- Email: [email protected]
Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm
ARTICLE IN PRESS
JID: COMM
[m5G;September 25, 2019;16:25]
Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Contents lists available at ScienceDirect
Computer Methods and Programs in Biomedicine journal homepage: www.elsevier.com/locate/cmpb
Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm Xiangdong Tian a,#,∗, Jian Wang a,#, Dongfeng Du b, Shuwen Li b, Changxiao Han b, Guangyu Zhu a, Yetong Tan a, Sheng Ma a, Handong Chen b, Ming Lei a a
Minimal Invasive Joint Department, The Third Affiliated Hospital of Beijing University of Chinese Medicine, No. 51 Anwai Xiaoguan Street, Chaoyang District, Beijing, 100029, China Graduate school of Beijing University of Chinese Medicine, Beijing, China
b
a r t i c l e
i n f o
Article history: Received 29 August 2019 Revised 10 September 2019 Accepted 14 September 2019 Available online xxx Keywords: Infrapatellar plica Magnetic resonance imaging Knee Arthroscopy Texture processing Laplacian algorithm
a b s t r a c t Objective: Nowadays, arthroscopy is widely applied to the treatment of joint diseases. The experimental trials were designed to determine whether the infrapatellar plica was symptomatic or not, and to appraise the medical effects of these patients who underwent arthroscopic treatment. An improved Laplacian image enhancement algorithm is added to the experiment. The medical image of the Subpatellar vertebral body under arthroscope is processed by the algorithm. The processed image is compared with the original image, and the advantages and disadvantages of the improved Laplacian image enhancement algorithm are analyzed. Methods: Retrospective Medical trial design was executed in our study. In addition, X-ray film and magnetic resonance imaging (MRI) were included in the study. Visual Analogue Scale (VAS) and Lysholm Score were carried out. Arthroscopy results, MRI findings, and Medical features were researched and analyzed carefully. Then we use the improved Laplacian image enhancement algorithm to process the image, which makes the image more convenient for analysis and improves the diagnostic accuracy. Results: Some of the experimental protomedical images are not clear enough, and the details and textures are difficult to judge, which hinders the diagnosis. After the improved Laplacian algorithm processing, the image effect has been significantly improved. From the image we get the result, although the wound healed after surgery, some patients have existence of transient swelling in recovery process but no effusion. The pain of all patients knee was sharply relieved and the function was improved. All patients’ conditions were most satisfactory. Conclusion: The findings in this study demonstrate a significant reduction in knee pain and improvement in function by releasing and removal of the symptomatic infrapatellar plica under arthroscopic surgery. The image processed by the improved Laplacian image enhancement algorithm can effectively retain the image details, which is conducive to diagnosis and improve the diagnostic accuracy. © 2019 Published by Elsevier B.V.
1. Introduction 1.1. Structure of infrapatellar fold The synovial membrane of the human knee is considered to be physiological structures that remained during the embryonic growing of the knee [1]. The infrapatellar plica of most people is left behind due to incomplete absorption. According to their different
∗
Corresponding author. E-mail address: [email protected] (X. Tian). # These authors contributed equally to the work and should be considered as cofirst authors.
origins, the synovial plicae can be divided into four smaller synovial plica: 1) patella medial synovial plica, 2) suprapatellar, 3) infrapatellar plica, and 4) patella lateral plicae [2]. Infrapatellar plica has a reputation of being a mucous membrane. It comes from its infrapatellar fat pad, which widens as it grows obliquely upward, over the anterior cruciate ligament (ACL), and adheres to the intercondylar notch of femur (Fig. 1). The plica is rarely large and thick [3]. There was obvious poor difference between knees, and it had no significant correlation with age [4]. 1.2. Medical symptoms of infrapatellar fold Many types of synovial plicae are often without symptoms and Medical signs. If synovial plicae failed their original elasticity or
https://doi.org/10.1016/j.cmpb.2019.105082 0169-2607/© 2019 Published by Elsevier B.V.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM 2
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Fig. 1. Anatomic study of the infrapatellar plica. The infrapatellar plica, which is also known as the ligamentum mucosum, is a normal structure in the knee that is present when the primitive embryologic septum between the medial and lateral compartments of the knee does not completely regress. The infrapatellar plica is located in the intercondylar area, anterior to the anterior cruciate ligament. It traverses from the intercondylar notch to the infrapatellar fat pad.
softness and turn to be fibrotic for various reasons, they could give rise to internal disorder of the knee joint [2,5,6]. The synovial plica may be symptomatic over time in the abnormal condition, and then it form what is called the plica syndrome of the knee [2]. The symptomatic infrapatellar plica becomes a kind of plica syndrome, and it is often obvious to the surgeons particularly in the process of arthrocentesis or arthroscopy. For these patients, large and thickened infrapatellar plica can form a hindrance to the pathway of arthroscope and some medical apparatuses [5,7]. Many outpatients complained of infrapatellar pain or anterior knee pain in a 30°–0° extension of the knee, with a transient “clicking” sound. Patients with light condition felt discomfort after a long walk, but felt no pain. It is often the early features of infrapatellar plica. It is possible that the infrapatellar plica stimulated and extruded the anterior cruciate ligament as well as the articular cartilage of the femoral condyle with flexion and extension of the knee joint. This leads to a marked infrapatellar plica with proliferation and cell activation due to continuous stimulation and force interaction. Such changes will result in chronic inflammation. In the process of arthrocentesis, lubricating fluid (such as sodium hyaluronate [21]) which should be injected to the joint, did not actually enter the articular cavity but stayed in the hyperplastic synovial membrane. Therefore, after repeated joint arthrocentesis, some patient’s condition did not improve significantly, but experienced knee joint swelling. Moreover, it can induce acute inflammation due to the aggravating condition. It is suggested that knee degeneracy results due to poor X-ray imaging of soft tissues. Unbelievably, the infrapatellar plica does not show up clearly on MRI. The MRI examination had many false appearances, which suggested injury of anterior or lateral meniscus, discoid meniscus and anterior cruciate ligament tear. All of these are contradictory to the intact medial and lateral meniscus and cruciate ligament under arthroscope. All of these illustrate the limitations of the MRI. Therefore, the patient’s symptoms did not improve significantly. Many medical experts do not recognize the condition as it is not obvious. It is often considered to be a meniscus or anterior cruciate ligament injury. Under arthroscopy, the infrapatellar plica has been observed intuitively [8]. Additionally, it had been widely reported in medical literature [9,10]. During our study, we observed the results of all the patients under our care. The purpose of this study is to determine whether the infrapatellar plicae was symptomatic or not, and to evaluate the Medical outcomes of all the patients.
1.3. Laplacian algorithm Image enhancement has the advantages of improving image quality, improving geometric registration accuracy and overcoming the image data incompleteness in object extraction and recognition. It has become an important information processing technology and has been widely used in remote sensing, medicine, aerospace and other fields. Image fusion refers to the use of various imaging sensors to obtain different images, synthesize complementary and redundant information of each image, and produce a new image to obtain more accurate, reliable and comprehensive image description. Image fusion includes three parts: pixel level fusion, feature level fusion and decision level fusion. In recent years, many image fusion methods have been proposed, among which the multi-resolution image fusion method in pixel level fusion is more common. Laplace pyramid decomposition method is one of the multi-resolution analysis methods. The classical Laplacian pyramid fusion algorithm can achieve the fusion effect without obvious stitching trace, but it has the defect of blurring the image due to the loss of details. In order to solve the above problems, this paper proposes an adaptive fusion area based on the Laplacian pyramid fusion algorithm of graph cutting, and proposes a weighted fusion method of multiple directions including horizontal direction. Using the complete detail information of source image, the fusion image of source image and Laplacian pyramid is fused according to this fusion method. Rules are fused to compensate the Laplace pyramid reconstruction error. This algorithm can effectively improve the details of the fused image, and the mosaic panorama is more realistic, which is in line with the characteristics of human vision. The improved Laplacian algorithm framework is shown in Fig. 2. 2. Materials and methods 2.1. Experimental subjects We performed a systematic research and analysed of all the patients who underwent diagnosis postoperatively with symptomatic infrapatellar plica at our department from October 2015 to March 2018. One hundred and six patients had knee internal lesions such as anterior cruciate ligament rupture, meniscal tears, chondromalacia patella, articular cartilage injury, infrapatellar plicae and medial plica syndrome at the beginning of our study. Those patients with symptomatic infrapatellar plica only were included in our study.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
3
ament injury in twelves cases (38%). In addition, the patients who suffered from tropatellar pain syndrome were found in nine cases (28%). All patients took conservative therapy including physical therapy, sticking plaster, non-steroid anti-inflammatory medications and progressive functional exercises for at least 8 weeks. Nonoperative treatments were unresolved to relieved symptoms. Arthroscopy was carried out for all the patients. Symptomatic infrapatellar plica was observed in 32 cases and divided into five types in the light of their morphological characteristics [8]: vertical septum pattern, separate types, split types, fenestra pattern and absent types. 2.3. Follow-up
Fig. 2. The improved Laplacian algorithm framework. This paper proposes an adaptive fusion area based on the Laplacian pyramid fusion algorithm of graph cutting, and proposes a weighted fusion method of multiple directions including horizontal direction. Using the complete detail information of source image, the fusion image of source image and Laplacian pyramid is fused according to this fusion method. Rules are fused to compensate the Laplace pyramid reconstruction error. This algorithm can effectively improve the details of the fused image, and the mosaic panorama is more realistic, which is in line with the characteristics of human vision.
Forty-one patients conform to our criteria from 106 knee arthroscopies conducted over the study period. Only 32 cases had followed up visits. There were twelve males and twenty females with a median age of 40 years (range: 27–70 years). There were 16 cases of left knee joint, and 18 cases of right knee joint. The postoperative follow-up time ranged from 6 months to 30 months (mean, 18 months). Diagnosis criteria were anterior knee pain, swelling, giving way, popping or snapping in process of flexion and extension. The feature of criteria was quite definitely; it consists of knee pain, which was frequent or having intermittent outbreak, and sometimes results in sudden painful clicking in the course of the activities such as jumping, stair activity and crouching. Definite trauma history existed in two cases. Eight cases have been duplicated injury due to daily activties such as line-dancing. Fourteen patients began to show symptoms with increased physical activity. Six cases had morbidity concealment. Finally, all the patients had anterior knee pain. Joint effusion was not displayed in these cases. Note that 5 cases had a 10° flexion contracture. In this study, X-ray, MRI, VAS and Lysholm scores were properly implemented for statistical analysis. 2.2. Research methods The frequently diagnosis was injury of the meniscus anterior horn shown in eight cases (25%). Patella medial synovial plica of knee was found in three cases (9%), and with anterior cruciate lig-
On the first day after surgery all the patients were asked to exercise their lower limbs in bed in case of vein thrombosis such as tightened legs and feet. All the patients were assigned to passive activity with CPM machine, crutch walking at 3 days postoperation, and recovered to normal activity or physical exercise 3 months later Preoperative MRI findings are to be consolidated. In addition, the postoperative situation were assessed as excellent, good or poor for visits [2]. Symptomless and a good and full healthy return with unrestricted activity were thought to an excellent rating. Then, the activity with mild symptoms was deemed to a ‘good’. Those patients with little symptomatic improvement were labelled as ‘poor’. visual analogue scale (VAS) is a 10 cm horizontal line with 0 at one end indicating no pain. Ten at the other end means severe pain; The middle represents different degrees of pain. Lysholm score consists of several items.Items include claudication, wring and swelling.The total score is 100 points, the higher the score, the better the curative effect. 2.4. Laplacian algorithm for image processing In order to make the image clearer and easier to diagnose the disease, we use a new image enhancement algorithm: Laplace algorithm based on graph cutting to process the image. The principle and flow chart of the algorithm are as follows. 2.4.1. Classical Laplacian pyramid fusion algorithms The Laplacian pyramid of image is evolved on the basis of the Gauss pyramid, so the first step is to decompose the image into Gauss pyramids. Assuming that the source image is, the bottom of the Gauss pyramid is I0 , and the first layer is Il , the image Il − 1 is filtered by Gauss low-pass filter and sampled downwards to get the image Il , that is, the image Il , whereby.
Il = Recude(Il−1 )
(1)
Repeat the above process to form the top of the Gauss Pyramid I0 , I1 ,…, In , In is the top of the Gauss pyramid. Secondly, image Il is decomposed by Laplace decomposition. Il is sampled up and Il∗ is obtained. The Laplace transform coefficients Pl of the corresponding layer 1 image are obtained by using the difference signal be∗ , that is, the Laplace transtween Il and the up-sampling signal Il+1 form coefficients Pl of the corresponding layer 1 image, whereby.
∗ Pl = Il − Il+1 , 0≤l≤N IN = Pl
(2)
Repeat the above process to achieve Laplace decomposition. In this way, Laplacian pyramid decomposition reflects the change of image in multiple resolutions. Different fusion rules are adopted for different frequency features and Laplacian coefficients are fused in corresponding layers. Finally, image I0 is reconstructed from top
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
ARTICLE IN PRESS
JID: COMM 4
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
to bottom according to Laplacian pyramid. The recursive formula is as follows:
∗ Il = Pl + Il+1 , 0≤l≤N IN = Pl
(3)
2.4.2. Improved Laplacian pyramid fusion method Aiming at the problem that classical Laplacian pyramid fusion algorithm can not be reconstructed accurately due to error noise, loss of details and blurring of image, this paper compensates the errors introduced in the process of image reconstruction, weights the fusion image of source image and Laplacian pyramid, and obtains the fusion image with improved details. After image registration, the optimal stitching line of overlapping area is obtained by graph cutting. Il and IR represent the two adjacent images and Iover is the overlapping area. x was extended to both sides through the suture of overlapping area. Because the scenes of different source image sequences are different, the effect of final image fusion is determined by x, so it is necessary to determine x adaptively and dynamically. Defined as:
√
ω × h × BS
x =
100
(4)
Fusion rules play a key role in the quality of the final panorama. Considering that there is no obvious stitching trace and details preservation, the following two main factors are taken into account when selecting fusion rules: 1. The closer to the optimal suture line, the greater the proportion of Laplace pyramid fusion, so as to achieve the effect of no obvious stitching trace. 2. The farther away from the optimal suture line, the larger the proportion of the source image is, and the details of the original image are retained as much as possible. Then the fusion of two images can be expressed as:
I(x, y ) = ωL × Il (x, y ) + ωs × Is (x, y )
(5)
In the formula: II(x, y) is a Laplacian pyramid fusion image; Is (x,y) is the source image. ωL + ωs = 1, ωs is the value:
ωs =
d (x,y ) x , d
(x, y ) < x 1, d (x, y ) = x
(6)
In formula d(x, y) is the distance in multiple directions. Weights ωs and ωL take into account the influence of regional pixels, the closer to the suture, the greater the weight value. ωs has the highest weight near the optimal suture, and gradually reduces to 0 on both sides of the suture; On the contrary, has the lowest weight at the optimal suture, and gradually increases to 1 on both sides of the suture. ωs and ωL take into account the influence of regional pixels, while taking into account the advantages of preserving the details of the source image and Laplacian pyramid fusion method without stitching trace. The reconstruction error is compensated by the complete details of the source image, and the source image and the Laplace pyramid image are weighted by multi-direction weighting method including horizontal direction. Finally, the fusion image with better detail preservation is obtained, and the fusion effect is improved. The schematic diagram depicting the distance d(x,y) is shown in Fig. 3.
Fig. 3. Distance d(x, y) schematic diagram. The reconstruction error is compensated by the complete details of the source image, and the source image and the Laplace pyramid image are weighted by multi-direction weighting method including horizontal direction. Finally, the fusion image with better detail preservation is obtained, and the fusion effect is improved.
out for pre-operation, post-operation, and the final follow-up respectively,and one-way anova was used for comparison before and after in the group. Note that p < 0.05 was considered to have statistical significant. 3. Results 3.1. Injury diagnostics The vertical septum pattern, separate types, split types, fenestra pattern and absent types were analyzed. These types are described as follows. Vertical septum types: the plica continued with the anterior surface of ACL (Fig. 4). Two parts were subdivided (Fig. 4a). Separate types: the plica was absolutely disassociated from ACL (Fig. 4b). Split types: the plica was different completely from the ACL and was split thoroughly (Fig. 4c). Fenestra types: the plica was complete synovium with a fenestra (Fig. 4d). Absent types: the remnant because of injury or impact (Fig. 4e). The routine procedure was complete resection of infrapatellar plica (Fig. 5e). 3.2. Follow-up and preoperative MRI findings Preoperative MRI findings are shown in Fig. 6, and the arthroscopy results and symptoms were recorded. The contents of follow up consist of symptoms and functions of knees, recurrence of the knee disease. These were assessed by VAS score and the Lysholm score.
2.5. Statistical methods
3.3. Retrospective MRI examinations
SPSS 24.0 was used for data analysis and technical graphics production. Measurement data was presented as mean ± standard deviation. The variance of the sample is homogeneous and accords with the normal distribution.VAS and Lysholm score were carried
All patients had returned to normal daily activities. Some cases had temporary swelling after operation. The important parts of follow-up: main complaints, motion of knee joint, VAS and Lysholm score.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
5
Fig. 4. Appearance of the infrapatellar plica that comes in five categories. Vertical septum type infrapatellar plica: The infrapatellar plica is in continuity with the anterior surface of the anterior cruciate ligament. Black arrowheads show the infrapatellar plica, and white arrow shows the anterior cruciate ligament (a). Separate type infrapatellar plica: the infrapatellar plica is completely separated from the anterior cruciate ligament. Black arrow shows the infrapatellar plica, and white arrow shows the anterior cruciate ligament (b). Split type infrapatellar plica: black arrows show the infrapatellar plica. The infrapatellar plica is completely separated from the anterior cruciate ligament (white arrow) and longitudinally split (c). Fenestra type infrapatellar plica: the plica is vertical septurn type with a fenestration. Black arrowheads show fenestration within the plica (d). Absent type infrapatellar plica: black arrow shows residual infrapatellar plica and white arrow shows the anterior cruciate ligament (e), and there is no synovium parallel to and above the anterior cruciate ligament.
Fig. 5. Symptomatic infrapatellar plica under arthroscope. The symptomatic infrapatellar plica (a); damaged the cartilage (b); and anterior cruciate ligament (c, and d) are demonstrated. Under knee arthroscopy, Resection of infrapatellar plica was performed (e).
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM 6
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Fig. 6. MRI manifestations of the infrapatellar plica. Sagittal fast spin-echo proton density-weighted MR image shows linear low intensity signal from infrapatellar plica (white arrows) in intercondylar notch anterior to anterior cruciate ligament. It can be found as a separate structure from the anterior cruciate ligament (ACL), fenestrated, or in continuity with the ACL.
The presenting symptom of the patient is the pain they experienced. Anterior knee pain was the most frequent as shown in 24 (75%) cases. The limit with flexion and tension existed in 20 (62.5%) cases. Representative swelling of knees were 8 (25%) and 4 (12.5%) cases respectively. MRI examinations conducted for all the cases were reviewed retrospectively on the basis of our research and extremely low signal intensity was shown by the sagittal images. Because some of the images in the experiment are relatively blurred, it is difficult to display some texture details, so we use the improved Laplacian image enhancement algorithm to process. We show the effect by comparing the original image with the processed image (Figs. 6 and 7). In line with the morphological characteristics assortments under arthroscopy, it was discovered many variable shapes, such as vertical septum pattern in eighteen cases, separate pattern in six cases, split type in four cases, fenestra pattern in two cases, and absent type in two cases. At earlier stages after the surgery, the symptoms were relieved in all patients, but swelling was reoccurring in six patients. Moreover, the flexion limitation of knee extension recognized initially and recovered by all patients. The median VAS of knee (5.94 ± 1.162) before operation and (0.22 ± 0.491) at the final follow-up visit (Table 1). We determined the VAS between 1 month post-operation and pre-operation (t = 14.296, p = 0.0 0 0). We also compared the VAS of 3 months with 1 month post-operation (t = 3.276, p = 0.003). Compared with the 3 months post-operative, VAS in 6 months after operation was evidently different (t = 2.339, p = 0.026). Additionally, we com-
Table 1 Measurement data of 32 patients that underwent arthoroscopy. Time
VAS score
Before arthroscopy 1 month after arthroscopy 3 months after arthroscopy 6 months after arthroscopy 12 months after arthroscopy
5.941 2.563 1.813 1.441 0.221
± ± ± ± ±
1.16 0.95 0.78 0.56 0.49
Lysholm score 50.59 79.34 85.31 92.91 94.25
± ± ± ± ±
9.48 3.01 3.15 2.71 2.15
VAS and Lysholm scores of 32 patients were measured and compared at pre-operation, 1 month, 3 months, 6 months and 12 months postoperation respectively. Measurement data was expressed using mean ± standard deviation. Based on the measured data, we can come to the conclusion that all of those who received the treatment experienced pain relief and function of knee joint improvement. Patient life has improved and returned to normal.
pared VAS 12 months with 6 months after operation (t = 10.459, p = 0.0 0 0). Note that p < 0.05, and have demonstrated sharp disparity. Meantime, we estimated that the Lysholm score before surgery was (37–67) with average of (50.59 ± 9.483) and the postoperative event visit score was (89–98) with the average of (94.25 ± 2.155) as shown by Table 1. Then, we compared the Lysholm scores of 1 month after operation with pre-operative (t = 15.943, p = 0.0 0 0). We compare the Lysholm scores of 3 months and 1 month postoperation (t = 8.450, p = 0.001). With 3 months after treatments, VAS of 6 months after treatments was very different (t = 10.210, p = 0.0 0 0). Ultimately, we compared Lysholm score of 12 months
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
7
Fig. 7. Imaging of human subjects with thickened infrapatellar plica causing anterior knee pain. X-ray showed degenerative arthritis (a, b). Sagittal proton density state images demonstrated a separate type of infrapatellar plica appearing as a fan-shaped thickened hypointense structure (arrow) coursing anterior to the anterior cruciate ligament (ACL). The fibers gradually become thinner anteriorly as they blend into Hoffa’s fat pad (c, d). Their respective enhanced images can depict these structures more clearly (e, f, g, h).
and 6 months after treatments (t = 2.109, p = 0.042). Note that p < 0.05 was thought to have distinct statistical meaning. Four cases were shown to be excellent, and twenty-eight cases were demonstrated to be good. No complications occurred except for two patients after operation. No second knee arthroscopy was performed. After arthroscopy, the function of knee joint improved significantly. 4. Discussion It is generally believed that the normal plica results due to the embryological vestigial remnants of synovial tissue of growing knees. It is an elastic tissue. Injury of the knee may result in inflammation, which lead to denaturation and inelasticity of the tissue. Finally, they result into fibrotic bands. But as is known to all, the suprapatellar plica is an important factor to cause symptoms [7]. The main features of infrapatellar plica concerned the problems caused by the membrane in process of arthroscopy and arthrography. It was considered that infrapatellar plica do not cause pain [2]. Most literatures on the pathology of plica claims that the infrapatellar plica has little Medical relevance and does not produce the symptoms [9]. Symptomatic infrapatellar plica was suggested when the correlative information was processed [10,11]. It was depicted as the fibrotic tissue, which is attached to the intercondylar notch, and tends to hinder further extension. Excision can be performed to increase the range of motion [11]. In our research, it was noticed that the 65 cases who have symptomatic infrapatellar plica with other injury of knee such as
anterior meniscus angle tears, injury of ACL, and chondral lesions fitted well with the diagnostic criteria as the cases that only have infrapatellar plica. Only in this way can the infrapatellar plica be treated correctly [12,13]. With regards to MRI, the symptomatic condition is better observed using the sagittal images [15,16], due to the curvilinear appearance in an anteroposterior orientation. Sagittal fast spin-echo T2-weighted MRI with fat suppression through intercondylar notch shows curvilinear high signal intensity along course of infrapatellar plica (Fig. 5). MRI is beneficial when it comes to excluding other lesions [17]. Because of different rates of identification of the plica imaging, anatomic studies are able to detect an extremely thin plica in comparison to using optical imaging [14]. Cothron et al. [18] believed that abnormal density of the infrapatellar plica shown on MRI may be due to trauma to plicae, which is probably linked to the Hoffa’s disease, and injury of the infrapatellar plica may imply an ACL injury. However, the MRI showed that the ACL is intact and also possesses a higher T2 signal around the infrapatellar plica in isolation. So they supported closely with the Medical symptoms and exclusion of other internal lesions and suggested injury of infrapatellar plica as an important cause knee pain [18]. In the process of research, we also encounter the problem of image blurring which affects the diagnosis. Image deblurring has attracted much attention in signal and image processing and other related fields. Aiming at this problem, researchers have proposed a variety of processing models and algorithms, but most of them have certain constraints. When applied to different blurred images, the stability of these models is not enough, which makes the
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM 8
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Fig. 8. Imaging of a human subject with thickened infrapatellar plica causing knee pain. X-ray showed no abnormal change (a, b). Sagittal proton density state images demonstrated a separate type of infrapatellar plica appearing as a fan-shaped thickened hypointense structure (arrow) coursing anterior to the anterior cruciate ligament (ACL). The fibers gradually become thinner anteriorly as they blend into Hoffa’s fat pad (c, d).Their respective enhanced images can depict these structures more clearly (e, f, g, h).
adaptability of the algorithm poor. Therefore, people are still diligently exploring a high performance. An algorithm that can and has universality for general real images. The core idea of traditional Laplacian pyramid image fusion is to construct pyramids with decreasing resolution, extract information from pyramids at different levels according to certain rules, and reconstruct the fusion image layer by layer using upper information. It mainly includes the following three steps: the establishment of image Gauss pyramid, the establishment of image Laplacian pyramid and the inverse transformation of Laplacian pyramid. The source image is filtered by Gauss lowpass filter separately, and then sampled by interlacing rows and columns. Through this process, a sub-image similar to the source image can be formed. Then the sub-image is inserted into an appropriate value to make it expand. Then the difference between the sub-image and the previous image is calculated and the band-pass component is obtained. This is the first layer of the Laplacian Pyramid. Repeat this approximate image to do the above decomposition process. By doing so, we can get other layers, which constitute the Laplacian Pyramid. Therefore, an image fusion algorithm based on Laplacian pyramid is proposed. The experimental results show that this method achieves good results for different focusing images and ordinary photographic images. The quality of the fused image is significantly improved compared with the general method, which has certain practical value.
In our study, we observed that the infrapatellar plica was located in the intercondylar area, anterior to the ACL and it traversed from the intercondylar notch to the infrapatellar fat pad (Fig. 7). The location and appearance have been well defined with arthroscopy (Fig. 7a). The thickened and inelastic infrapatellar plica had done serious damage to the femoral condylar cartilage (Fig. 7b). At the same time anterior cruciate ligament suffered from injury of the symptomatic infrapatellar plica in flexion and extension of the knee joint (Fig. 7c). Under arthroscopy, symptomatic infrapatellar plica is impacted on the adjacent cartilage and anterior cruciate ligament (ACL) in the flexion and extension of the knee joint (Fig. 7d). In some literatures, infrapatellar plica was described as its wing ligament. But we took the excised infrapatellar plica for pathological examinations under the arthroscope. Pathological examination showed that it consists of most synovial tissues, adipose tissue and some inflammatory cells (Fig. 9a). Degeneration of many synovial tissues was observed in histology. They demonstrate fibrosis, hyalinization, and calcification [19] as shown (Fig. 9b). The infrapatellar plica is often observed during the arthroscopic surgical operation [20]. The difference between the orthopedic and radiology is due to a lack of knowledge pertaining to the anatomical architecture. In addition, the treatments of symptomatic infrapatellar plica are limited, and the arthroscopic imaging and their Medical features are vivid.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
ARTICLE IN PRESS
JID: COMM
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
9
Fig. 9. Pathological study based on symptomatic infrapatellar plica. According to the pathological investigation, the symptomatic infrapatellar plica consists of synovial tissues, adipose tissues and inflammatory cells (a); and the synovial tissues demonstrate fibrosis, hyalinization, and calcification (b).
5. Conclusion Medical imaging coupled with enhanced image processing is a necessary and effective advanced method to make a diagnosis and give treatment for infrapatellar plica. The findings in this study demonstrate a significant reduction in knee pain and improvement in function by releasing and removal of the symptomatic infrapatellar plica under arthroscopic surgery. The image processed by the improved Laplacian image enhancement algorithm can effectively retain the image details, which is conducive to diagnosis and improve the diagnostic accuracy. Compliance with ethical standards The study was funded by personal (XDT) research and development fund (303-02-01-06-04). And authors whose names appear on the manuscript have contributed sufficiently to the scientific work and therefore share collective responsibility and accountability for the results. The authors declare that they have no conflict of interest. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. Declaration of Competing Interest The authors declared that there was no conflict of interest in this study. CRediT authorship contribution statement Xiangdong Tian: Conceptualization, Formal analysis, Visualization, Writing - original draft. Jian Wang: Formal analysis, Visualization, Writing - original draft, Resources, Software. Dongfeng Du: Data curation, Project administration, Methodology. Shuwen Li: Data curation. Changxiao Han: Methodology. Guangyu Zhu: Formal analysis, Project administration. Yetong Tan: Formal analysis. Ming Lei: Data curation.
thopedic Center of The Third Affiliated Hospital of Beijing University of Chinese Medicine. This work was supported in part by Graduate school of Beijing University of Chinese Medicine, Beijing. References [1] N.L. Tindel, B. Nisonson, The plica syndrome, Orthop. Clin. North Am. 23 (1992) 613–618. [2] W.T. Hardaker, T.L. Whipple, F.H. Bassett, Diagnosis and treatment of the plica syndrome of the knee, J. Bone Joint Surg. Am. 62 (2) (1980) 221–225. [3] B. Reider, J.L. Marshall, R.F. Warren, Persistent vertical septum in the human knee joint, Brief Note 63 (7) (1981) 1185–1187. [4] P.Y.F. Lee, et al., Synovial plica syndrome of the knee: a commonly overlooked cause of anterior knee pain, Surg. J. 1 (3) (2017) 9–16. [5] D. Patel, Arthroscopy of the plicae-synovial folds and their significance, Am. J. Sports Med. 6 (1978) 217–225. [6] D. Patel, Plica as a cause of anterior knee pain, Orthop. Clin. North Am. 17 (2) (1986) 273–277. [7] G.L. Muse, W.A. Grana, S. Hollingsworth, Arthroscopic treatment of medial shelf syndrome, Arthroscopy 1 (1) (1985) 63–67. [8] S.J. Kim, B.H. Min, H.K. Kim, Arthroscopic anatomy of the infrapatellar plica, Arthroscopy 12 (5) (1996) 561–564. [9] A.M. Jackson, Anterior knee pain, J. Bone Joint Surg. Br. 83 (2001) 937–948. [10] S.J. Kim, W. Choe, Pathological infrapatellar plica: a report of two cases and literature review, Arthroscopy 12 (2) (1996) 236–239. [11] S.J. Kim, J.Y. Kim, J.W. Lee, Pathologic infrapatellar plica, Arthroscopy 18 (5) (2002) 25–28. [12] D. Kohn, S. Deiler, M. Rudert, Arterial blood supply of the infrapatellar fat pad anatomy and medical consequences, Arch. Orthop. Trauma Surg. 114 (1995) 72–75. [13] M. Magi, A. Branca, C. Bucca, et al., Hoffa disease, Ital. J. Orthop. Trauma 17 (1991) 675–691. [14] F.J. Koserek, C.A. Helms, The MR appearance of the infrapatellar plica, Am. J. Roentgenol. 172 (2) (1999) 481–484. [15] E. Stathopulu, E. Baildam, Anterior knee pain: a long term follow-up, Rheumatology 42 (2003) 380–382. [16] R. Garcia-Valtuille, F. Abascal, L. Cerezal, et al., Anatomy and MR imaging appearances of synovial plicae of the knee, Radiographics 22 (2002) 775–784. [17] N.M. Major, C.A. Helms, MR imaging of the knee: findings in asymptomatic collegiate basketball players, Am. J. Roentgenol. 179 (2002) 641–644. [18] R.L. Cothran, P.M. McGuire, C.A. Helms, et al., MR imaging of infrapatellar plica injury, Am. J. Roen. 180 (5) (2003) 1443–1447. [19] Christine R. Boyd, Colin Eakin, Gordon O. Matheson, Infrapatellar plica as a cause of anterior knee pain, Clin. J. Sport Med. 15 (2005) 98–103. [20] D.P. Johnson, D.M. Eastwood, P.J. Witherow, Symptomatic synovial plicae of the knee, J. Bone Joint Surg. Am. 75 (1993) 1485–1496. [21] C.-W. Ha, Y.-B. Park, C.-H. Choi, et al., Efficacy and safety of single injection of cross-linked sodium hyaluronate vs. three injections of high molecular weight sodium hyaluronate for osteoarthritis of the knee: a double-blind, randomized, multi-center, non-inferiority study, BMC Musculoskelet Disord. 18 (1) (2017) 223.
Acknowledgement The authors express their appreciation for the Medical support of orthopedics and valuable discussions with surgeons at the Or-
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
[m5G;September 25, 2019;16:25]
Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Contents lists available at ScienceDirect
Computer Methods and Programs in Biomedicine journal homepage: www.elsevier.com/locate/cmpb
Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm Xiangdong Tian a,#,∗, Jian Wang a,#, Dongfeng Du b, Shuwen Li b, Changxiao Han b, Guangyu Zhu a, Yetong Tan a, Sheng Ma a, Handong Chen b, Ming Lei a a
Minimal Invasive Joint Department, The Third Affiliated Hospital of Beijing University of Chinese Medicine, No. 51 Anwai Xiaoguan Street, Chaoyang District, Beijing, 100029, China Graduate school of Beijing University of Chinese Medicine, Beijing, China
b
a r t i c l e
i n f o
Article history: Received 29 August 2019 Revised 10 September 2019 Accepted 14 September 2019 Available online xxx Keywords: Infrapatellar plica Magnetic resonance imaging Knee Arthroscopy Texture processing Laplacian algorithm
a b s t r a c t Objective: Nowadays, arthroscopy is widely applied to the treatment of joint diseases. The experimental trials were designed to determine whether the infrapatellar plica was symptomatic or not, and to appraise the medical effects of these patients who underwent arthroscopic treatment. An improved Laplacian image enhancement algorithm is added to the experiment. The medical image of the Subpatellar vertebral body under arthroscope is processed by the algorithm. The processed image is compared with the original image, and the advantages and disadvantages of the improved Laplacian image enhancement algorithm are analyzed. Methods: Retrospective Medical trial design was executed in our study. In addition, X-ray film and magnetic resonance imaging (MRI) were included in the study. Visual Analogue Scale (VAS) and Lysholm Score were carried out. Arthroscopy results, MRI findings, and Medical features were researched and analyzed carefully. Then we use the improved Laplacian image enhancement algorithm to process the image, which makes the image more convenient for analysis and improves the diagnostic accuracy. Results: Some of the experimental protomedical images are not clear enough, and the details and textures are difficult to judge, which hinders the diagnosis. After the improved Laplacian algorithm processing, the image effect has been significantly improved. From the image we get the result, although the wound healed after surgery, some patients have existence of transient swelling in recovery process but no effusion. The pain of all patients knee was sharply relieved and the function was improved. All patients’ conditions were most satisfactory. Conclusion: The findings in this study demonstrate a significant reduction in knee pain and improvement in function by releasing and removal of the symptomatic infrapatellar plica under arthroscopic surgery. The image processed by the improved Laplacian image enhancement algorithm can effectively retain the image details, which is conducive to diagnosis and improve the diagnostic accuracy. © 2019 Published by Elsevier B.V.
1. Introduction 1.1. Structure of infrapatellar fold The synovial membrane of the human knee is considered to be physiological structures that remained during the embryonic growing of the knee [1]. The infrapatellar plica of most people is left behind due to incomplete absorption. According to their different
∗
Corresponding author. E-mail address: [email protected] (X. Tian). # These authors contributed equally to the work and should be considered as cofirst authors.
origins, the synovial plicae can be divided into four smaller synovial plica: 1) patella medial synovial plica, 2) suprapatellar, 3) infrapatellar plica, and 4) patella lateral plicae [2]. Infrapatellar plica has a reputation of being a mucous membrane. It comes from its infrapatellar fat pad, which widens as it grows obliquely upward, over the anterior cruciate ligament (ACL), and adheres to the intercondylar notch of femur (Fig. 1). The plica is rarely large and thick [3]. There was obvious poor difference between knees, and it had no significant correlation with age [4]. 1.2. Medical symptoms of infrapatellar fold Many types of synovial plicae are often without symptoms and Medical signs. If synovial plicae failed their original elasticity or
https://doi.org/10.1016/j.cmpb.2019.105082 0169-2607/© 2019 Published by Elsevier B.V.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM 2
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Fig. 1. Anatomic study of the infrapatellar plica. The infrapatellar plica, which is also known as the ligamentum mucosum, is a normal structure in the knee that is present when the primitive embryologic septum between the medial and lateral compartments of the knee does not completely regress. The infrapatellar plica is located in the intercondylar area, anterior to the anterior cruciate ligament. It traverses from the intercondylar notch to the infrapatellar fat pad.
softness and turn to be fibrotic for various reasons, they could give rise to internal disorder of the knee joint [2,5,6]. The synovial plica may be symptomatic over time in the abnormal condition, and then it form what is called the plica syndrome of the knee [2]. The symptomatic infrapatellar plica becomes a kind of plica syndrome, and it is often obvious to the surgeons particularly in the process of arthrocentesis or arthroscopy. For these patients, large and thickened infrapatellar plica can form a hindrance to the pathway of arthroscope and some medical apparatuses [5,7]. Many outpatients complained of infrapatellar pain or anterior knee pain in a 30°–0° extension of the knee, with a transient “clicking” sound. Patients with light condition felt discomfort after a long walk, but felt no pain. It is often the early features of infrapatellar plica. It is possible that the infrapatellar plica stimulated and extruded the anterior cruciate ligament as well as the articular cartilage of the femoral condyle with flexion and extension of the knee joint. This leads to a marked infrapatellar plica with proliferation and cell activation due to continuous stimulation and force interaction. Such changes will result in chronic inflammation. In the process of arthrocentesis, lubricating fluid (such as sodium hyaluronate [21]) which should be injected to the joint, did not actually enter the articular cavity but stayed in the hyperplastic synovial membrane. Therefore, after repeated joint arthrocentesis, some patient’s condition did not improve significantly, but experienced knee joint swelling. Moreover, it can induce acute inflammation due to the aggravating condition. It is suggested that knee degeneracy results due to poor X-ray imaging of soft tissues. Unbelievably, the infrapatellar plica does not show up clearly on MRI. The MRI examination had many false appearances, which suggested injury of anterior or lateral meniscus, discoid meniscus and anterior cruciate ligament tear. All of these are contradictory to the intact medial and lateral meniscus and cruciate ligament under arthroscope. All of these illustrate the limitations of the MRI. Therefore, the patient’s symptoms did not improve significantly. Many medical experts do not recognize the condition as it is not obvious. It is often considered to be a meniscus or anterior cruciate ligament injury. Under arthroscopy, the infrapatellar plica has been observed intuitively [8]. Additionally, it had been widely reported in medical literature [9,10]. During our study, we observed the results of all the patients under our care. The purpose of this study is to determine whether the infrapatellar plicae was symptomatic or not, and to evaluate the Medical outcomes of all the patients.
1.3. Laplacian algorithm Image enhancement has the advantages of improving image quality, improving geometric registration accuracy and overcoming the image data incompleteness in object extraction and recognition. It has become an important information processing technology and has been widely used in remote sensing, medicine, aerospace and other fields. Image fusion refers to the use of various imaging sensors to obtain different images, synthesize complementary and redundant information of each image, and produce a new image to obtain more accurate, reliable and comprehensive image description. Image fusion includes three parts: pixel level fusion, feature level fusion and decision level fusion. In recent years, many image fusion methods have been proposed, among which the multi-resolution image fusion method in pixel level fusion is more common. Laplace pyramid decomposition method is one of the multi-resolution analysis methods. The classical Laplacian pyramid fusion algorithm can achieve the fusion effect without obvious stitching trace, but it has the defect of blurring the image due to the loss of details. In order to solve the above problems, this paper proposes an adaptive fusion area based on the Laplacian pyramid fusion algorithm of graph cutting, and proposes a weighted fusion method of multiple directions including horizontal direction. Using the complete detail information of source image, the fusion image of source image and Laplacian pyramid is fused according to this fusion method. Rules are fused to compensate the Laplace pyramid reconstruction error. This algorithm can effectively improve the details of the fused image, and the mosaic panorama is more realistic, which is in line with the characteristics of human vision. The improved Laplacian algorithm framework is shown in Fig. 2. 2. Materials and methods 2.1. Experimental subjects We performed a systematic research and analysed of all the patients who underwent diagnosis postoperatively with symptomatic infrapatellar plica at our department from October 2015 to March 2018. One hundred and six patients had knee internal lesions such as anterior cruciate ligament rupture, meniscal tears, chondromalacia patella, articular cartilage injury, infrapatellar plicae and medial plica syndrome at the beginning of our study. Those patients with symptomatic infrapatellar plica only were included in our study.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
3
ament injury in twelves cases (38%). In addition, the patients who suffered from tropatellar pain syndrome were found in nine cases (28%). All patients took conservative therapy including physical therapy, sticking plaster, non-steroid anti-inflammatory medications and progressive functional exercises for at least 8 weeks. Nonoperative treatments were unresolved to relieved symptoms. Arthroscopy was carried out for all the patients. Symptomatic infrapatellar plica was observed in 32 cases and divided into five types in the light of their morphological characteristics [8]: vertical septum pattern, separate types, split types, fenestra pattern and absent types. 2.3. Follow-up
Fig. 2. The improved Laplacian algorithm framework. This paper proposes an adaptive fusion area based on the Laplacian pyramid fusion algorithm of graph cutting, and proposes a weighted fusion method of multiple directions including horizontal direction. Using the complete detail information of source image, the fusion image of source image and Laplacian pyramid is fused according to this fusion method. Rules are fused to compensate the Laplace pyramid reconstruction error. This algorithm can effectively improve the details of the fused image, and the mosaic panorama is more realistic, which is in line with the characteristics of human vision.
Forty-one patients conform to our criteria from 106 knee arthroscopies conducted over the study period. Only 32 cases had followed up visits. There were twelve males and twenty females with a median age of 40 years (range: 27–70 years). There were 16 cases of left knee joint, and 18 cases of right knee joint. The postoperative follow-up time ranged from 6 months to 30 months (mean, 18 months). Diagnosis criteria were anterior knee pain, swelling, giving way, popping or snapping in process of flexion and extension. The feature of criteria was quite definitely; it consists of knee pain, which was frequent or having intermittent outbreak, and sometimes results in sudden painful clicking in the course of the activities such as jumping, stair activity and crouching. Definite trauma history existed in two cases. Eight cases have been duplicated injury due to daily activties such as line-dancing. Fourteen patients began to show symptoms with increased physical activity. Six cases had morbidity concealment. Finally, all the patients had anterior knee pain. Joint effusion was not displayed in these cases. Note that 5 cases had a 10° flexion contracture. In this study, X-ray, MRI, VAS and Lysholm scores were properly implemented for statistical analysis. 2.2. Research methods The frequently diagnosis was injury of the meniscus anterior horn shown in eight cases (25%). Patella medial synovial plica of knee was found in three cases (9%), and with anterior cruciate lig-
On the first day after surgery all the patients were asked to exercise their lower limbs in bed in case of vein thrombosis such as tightened legs and feet. All the patients were assigned to passive activity with CPM machine, crutch walking at 3 days postoperation, and recovered to normal activity or physical exercise 3 months later Preoperative MRI findings are to be consolidated. In addition, the postoperative situation were assessed as excellent, good or poor for visits [2]. Symptomless and a good and full healthy return with unrestricted activity were thought to an excellent rating. Then, the activity with mild symptoms was deemed to a ‘good’. Those patients with little symptomatic improvement were labelled as ‘poor’. visual analogue scale (VAS) is a 10 cm horizontal line with 0 at one end indicating no pain. Ten at the other end means severe pain; The middle represents different degrees of pain. Lysholm score consists of several items.Items include claudication, wring and swelling.The total score is 100 points, the higher the score, the better the curative effect. 2.4. Laplacian algorithm for image processing In order to make the image clearer and easier to diagnose the disease, we use a new image enhancement algorithm: Laplace algorithm based on graph cutting to process the image. The principle and flow chart of the algorithm are as follows. 2.4.1. Classical Laplacian pyramid fusion algorithms The Laplacian pyramid of image is evolved on the basis of the Gauss pyramid, so the first step is to decompose the image into Gauss pyramids. Assuming that the source image is, the bottom of the Gauss pyramid is I0 , and the first layer is Il , the image Il − 1 is filtered by Gauss low-pass filter and sampled downwards to get the image Il , that is, the image Il , whereby.
Il = Recude(Il−1 )
(1)
Repeat the above process to form the top of the Gauss Pyramid I0 , I1 ,…, In , In is the top of the Gauss pyramid. Secondly, image Il is decomposed by Laplace decomposition. Il is sampled up and Il∗ is obtained. The Laplace transform coefficients Pl of the corresponding layer 1 image are obtained by using the difference signal be∗ , that is, the Laplace transtween Il and the up-sampling signal Il+1 form coefficients Pl of the corresponding layer 1 image, whereby.
∗ Pl = Il − Il+1 , 0≤l≤N IN = Pl
(2)
Repeat the above process to achieve Laplace decomposition. In this way, Laplacian pyramid decomposition reflects the change of image in multiple resolutions. Different fusion rules are adopted for different frequency features and Laplacian coefficients are fused in corresponding layers. Finally, image I0 is reconstructed from top
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
ARTICLE IN PRESS
JID: COMM 4
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
to bottom according to Laplacian pyramid. The recursive formula is as follows:
∗ Il = Pl + Il+1 , 0≤l≤N IN = Pl
(3)
2.4.2. Improved Laplacian pyramid fusion method Aiming at the problem that classical Laplacian pyramid fusion algorithm can not be reconstructed accurately due to error noise, loss of details and blurring of image, this paper compensates the errors introduced in the process of image reconstruction, weights the fusion image of source image and Laplacian pyramid, and obtains the fusion image with improved details. After image registration, the optimal stitching line of overlapping area is obtained by graph cutting. Il and IR represent the two adjacent images and Iover is the overlapping area. x was extended to both sides through the suture of overlapping area. Because the scenes of different source image sequences are different, the effect of final image fusion is determined by x, so it is necessary to determine x adaptively and dynamically. Defined as:
√
ω × h × BS
x =
100
(4)
Fusion rules play a key role in the quality of the final panorama. Considering that there is no obvious stitching trace and details preservation, the following two main factors are taken into account when selecting fusion rules: 1. The closer to the optimal suture line, the greater the proportion of Laplace pyramid fusion, so as to achieve the effect of no obvious stitching trace. 2. The farther away from the optimal suture line, the larger the proportion of the source image is, and the details of the original image are retained as much as possible. Then the fusion of two images can be expressed as:
I(x, y ) = ωL × Il (x, y ) + ωs × Is (x, y )
(5)
In the formula: II(x, y) is a Laplacian pyramid fusion image; Is (x,y) is the source image. ωL + ωs = 1, ωs is the value:
ωs =
d (x,y ) x , d
(x, y ) < x 1, d (x, y ) = x
(6)
In formula d(x, y) is the distance in multiple directions. Weights ωs and ωL take into account the influence of regional pixels, the closer to the suture, the greater the weight value. ωs has the highest weight near the optimal suture, and gradually reduces to 0 on both sides of the suture; On the contrary, has the lowest weight at the optimal suture, and gradually increases to 1 on both sides of the suture. ωs and ωL take into account the influence of regional pixels, while taking into account the advantages of preserving the details of the source image and Laplacian pyramid fusion method without stitching trace. The reconstruction error is compensated by the complete details of the source image, and the source image and the Laplace pyramid image are weighted by multi-direction weighting method including horizontal direction. Finally, the fusion image with better detail preservation is obtained, and the fusion effect is improved. The schematic diagram depicting the distance d(x,y) is shown in Fig. 3.
Fig. 3. Distance d(x, y) schematic diagram. The reconstruction error is compensated by the complete details of the source image, and the source image and the Laplace pyramid image are weighted by multi-direction weighting method including horizontal direction. Finally, the fusion image with better detail preservation is obtained, and the fusion effect is improved.
out for pre-operation, post-operation, and the final follow-up respectively,and one-way anova was used for comparison before and after in the group. Note that p < 0.05 was considered to have statistical significant. 3. Results 3.1. Injury diagnostics The vertical septum pattern, separate types, split types, fenestra pattern and absent types were analyzed. These types are described as follows. Vertical septum types: the plica continued with the anterior surface of ACL (Fig. 4). Two parts were subdivided (Fig. 4a). Separate types: the plica was absolutely disassociated from ACL (Fig. 4b). Split types: the plica was different completely from the ACL and was split thoroughly (Fig. 4c). Fenestra types: the plica was complete synovium with a fenestra (Fig. 4d). Absent types: the remnant because of injury or impact (Fig. 4e). The routine procedure was complete resection of infrapatellar plica (Fig. 5e). 3.2. Follow-up and preoperative MRI findings Preoperative MRI findings are shown in Fig. 6, and the arthroscopy results and symptoms were recorded. The contents of follow up consist of symptoms and functions of knees, recurrence of the knee disease. These were assessed by VAS score and the Lysholm score.
2.5. Statistical methods
3.3. Retrospective MRI examinations
SPSS 24.0 was used for data analysis and technical graphics production. Measurement data was presented as mean ± standard deviation. The variance of the sample is homogeneous and accords with the normal distribution.VAS and Lysholm score were carried
All patients had returned to normal daily activities. Some cases had temporary swelling after operation. The important parts of follow-up: main complaints, motion of knee joint, VAS and Lysholm score.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
5
Fig. 4. Appearance of the infrapatellar plica that comes in five categories. Vertical septum type infrapatellar plica: The infrapatellar plica is in continuity with the anterior surface of the anterior cruciate ligament. Black arrowheads show the infrapatellar plica, and white arrow shows the anterior cruciate ligament (a). Separate type infrapatellar plica: the infrapatellar plica is completely separated from the anterior cruciate ligament. Black arrow shows the infrapatellar plica, and white arrow shows the anterior cruciate ligament (b). Split type infrapatellar plica: black arrows show the infrapatellar plica. The infrapatellar plica is completely separated from the anterior cruciate ligament (white arrow) and longitudinally split (c). Fenestra type infrapatellar plica: the plica is vertical septurn type with a fenestration. Black arrowheads show fenestration within the plica (d). Absent type infrapatellar plica: black arrow shows residual infrapatellar plica and white arrow shows the anterior cruciate ligament (e), and there is no synovium parallel to and above the anterior cruciate ligament.
Fig. 5. Symptomatic infrapatellar plica under arthroscope. The symptomatic infrapatellar plica (a); damaged the cartilage (b); and anterior cruciate ligament (c, and d) are demonstrated. Under knee arthroscopy, Resection of infrapatellar plica was performed (e).
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM 6
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Fig. 6. MRI manifestations of the infrapatellar plica. Sagittal fast spin-echo proton density-weighted MR image shows linear low intensity signal from infrapatellar plica (white arrows) in intercondylar notch anterior to anterior cruciate ligament. It can be found as a separate structure from the anterior cruciate ligament (ACL), fenestrated, or in continuity with the ACL.
The presenting symptom of the patient is the pain they experienced. Anterior knee pain was the most frequent as shown in 24 (75%) cases. The limit with flexion and tension existed in 20 (62.5%) cases. Representative swelling of knees were 8 (25%) and 4 (12.5%) cases respectively. MRI examinations conducted for all the cases were reviewed retrospectively on the basis of our research and extremely low signal intensity was shown by the sagittal images. Because some of the images in the experiment are relatively blurred, it is difficult to display some texture details, so we use the improved Laplacian image enhancement algorithm to process. We show the effect by comparing the original image with the processed image (Figs. 6 and 7). In line with the morphological characteristics assortments under arthroscopy, it was discovered many variable shapes, such as vertical septum pattern in eighteen cases, separate pattern in six cases, split type in four cases, fenestra pattern in two cases, and absent type in two cases. At earlier stages after the surgery, the symptoms were relieved in all patients, but swelling was reoccurring in six patients. Moreover, the flexion limitation of knee extension recognized initially and recovered by all patients. The median VAS of knee (5.94 ± 1.162) before operation and (0.22 ± 0.491) at the final follow-up visit (Table 1). We determined the VAS between 1 month post-operation and pre-operation (t = 14.296, p = 0.0 0 0). We also compared the VAS of 3 months with 1 month post-operation (t = 3.276, p = 0.003). Compared with the 3 months post-operative, VAS in 6 months after operation was evidently different (t = 2.339, p = 0.026). Additionally, we com-
Table 1 Measurement data of 32 patients that underwent arthoroscopy. Time
VAS score
Before arthroscopy 1 month after arthroscopy 3 months after arthroscopy 6 months after arthroscopy 12 months after arthroscopy
5.941 2.563 1.813 1.441 0.221
± ± ± ± ±
1.16 0.95 0.78 0.56 0.49
Lysholm score 50.59 79.34 85.31 92.91 94.25
± ± ± ± ±
9.48 3.01 3.15 2.71 2.15
VAS and Lysholm scores of 32 patients were measured and compared at pre-operation, 1 month, 3 months, 6 months and 12 months postoperation respectively. Measurement data was expressed using mean ± standard deviation. Based on the measured data, we can come to the conclusion that all of those who received the treatment experienced pain relief and function of knee joint improvement. Patient life has improved and returned to normal.
pared VAS 12 months with 6 months after operation (t = 10.459, p = 0.0 0 0). Note that p < 0.05, and have demonstrated sharp disparity. Meantime, we estimated that the Lysholm score before surgery was (37–67) with average of (50.59 ± 9.483) and the postoperative event visit score was (89–98) with the average of (94.25 ± 2.155) as shown by Table 1. Then, we compared the Lysholm scores of 1 month after operation with pre-operative (t = 15.943, p = 0.0 0 0). We compare the Lysholm scores of 3 months and 1 month postoperation (t = 8.450, p = 0.001). With 3 months after treatments, VAS of 6 months after treatments was very different (t = 10.210, p = 0.0 0 0). Ultimately, we compared Lysholm score of 12 months
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
7
Fig. 7. Imaging of human subjects with thickened infrapatellar plica causing anterior knee pain. X-ray showed degenerative arthritis (a, b). Sagittal proton density state images demonstrated a separate type of infrapatellar plica appearing as a fan-shaped thickened hypointense structure (arrow) coursing anterior to the anterior cruciate ligament (ACL). The fibers gradually become thinner anteriorly as they blend into Hoffa’s fat pad (c, d). Their respective enhanced images can depict these structures more clearly (e, f, g, h).
and 6 months after treatments (t = 2.109, p = 0.042). Note that p < 0.05 was thought to have distinct statistical meaning. Four cases were shown to be excellent, and twenty-eight cases were demonstrated to be good. No complications occurred except for two patients after operation. No second knee arthroscopy was performed. After arthroscopy, the function of knee joint improved significantly. 4. Discussion It is generally believed that the normal plica results due to the embryological vestigial remnants of synovial tissue of growing knees. It is an elastic tissue. Injury of the knee may result in inflammation, which lead to denaturation and inelasticity of the tissue. Finally, they result into fibrotic bands. But as is known to all, the suprapatellar plica is an important factor to cause symptoms [7]. The main features of infrapatellar plica concerned the problems caused by the membrane in process of arthroscopy and arthrography. It was considered that infrapatellar plica do not cause pain [2]. Most literatures on the pathology of plica claims that the infrapatellar plica has little Medical relevance and does not produce the symptoms [9]. Symptomatic infrapatellar plica was suggested when the correlative information was processed [10,11]. It was depicted as the fibrotic tissue, which is attached to the intercondylar notch, and tends to hinder further extension. Excision can be performed to increase the range of motion [11]. In our research, it was noticed that the 65 cases who have symptomatic infrapatellar plica with other injury of knee such as
anterior meniscus angle tears, injury of ACL, and chondral lesions fitted well with the diagnostic criteria as the cases that only have infrapatellar plica. Only in this way can the infrapatellar plica be treated correctly [12,13]. With regards to MRI, the symptomatic condition is better observed using the sagittal images [15,16], due to the curvilinear appearance in an anteroposterior orientation. Sagittal fast spin-echo T2-weighted MRI with fat suppression through intercondylar notch shows curvilinear high signal intensity along course of infrapatellar plica (Fig. 5). MRI is beneficial when it comes to excluding other lesions [17]. Because of different rates of identification of the plica imaging, anatomic studies are able to detect an extremely thin plica in comparison to using optical imaging [14]. Cothron et al. [18] believed that abnormal density of the infrapatellar plica shown on MRI may be due to trauma to plicae, which is probably linked to the Hoffa’s disease, and injury of the infrapatellar plica may imply an ACL injury. However, the MRI showed that the ACL is intact and also possesses a higher T2 signal around the infrapatellar plica in isolation. So they supported closely with the Medical symptoms and exclusion of other internal lesions and suggested injury of infrapatellar plica as an important cause knee pain [18]. In the process of research, we also encounter the problem of image blurring which affects the diagnosis. Image deblurring has attracted much attention in signal and image processing and other related fields. Aiming at this problem, researchers have proposed a variety of processing models and algorithms, but most of them have certain constraints. When applied to different blurred images, the stability of these models is not enough, which makes the
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
JID: COMM 8
ARTICLE IN PRESS
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
Fig. 8. Imaging of a human subject with thickened infrapatellar plica causing knee pain. X-ray showed no abnormal change (a, b). Sagittal proton density state images demonstrated a separate type of infrapatellar plica appearing as a fan-shaped thickened hypointense structure (arrow) coursing anterior to the anterior cruciate ligament (ACL). The fibers gradually become thinner anteriorly as they blend into Hoffa’s fat pad (c, d).Their respective enhanced images can depict these structures more clearly (e, f, g, h).
adaptability of the algorithm poor. Therefore, people are still diligently exploring a high performance. An algorithm that can and has universality for general real images. The core idea of traditional Laplacian pyramid image fusion is to construct pyramids with decreasing resolution, extract information from pyramids at different levels according to certain rules, and reconstruct the fusion image layer by layer using upper information. It mainly includes the following three steps: the establishment of image Gauss pyramid, the establishment of image Laplacian pyramid and the inverse transformation of Laplacian pyramid. The source image is filtered by Gauss lowpass filter separately, and then sampled by interlacing rows and columns. Through this process, a sub-image similar to the source image can be formed. Then the sub-image is inserted into an appropriate value to make it expand. Then the difference between the sub-image and the previous image is calculated and the band-pass component is obtained. This is the first layer of the Laplacian Pyramid. Repeat this approximate image to do the above decomposition process. By doing so, we can get other layers, which constitute the Laplacian Pyramid. Therefore, an image fusion algorithm based on Laplacian pyramid is proposed. The experimental results show that this method achieves good results for different focusing images and ordinary photographic images. The quality of the fused image is significantly improved compared with the general method, which has certain practical value.
In our study, we observed that the infrapatellar plica was located in the intercondylar area, anterior to the ACL and it traversed from the intercondylar notch to the infrapatellar fat pad (Fig. 7). The location and appearance have been well defined with arthroscopy (Fig. 7a). The thickened and inelastic infrapatellar plica had done serious damage to the femoral condylar cartilage (Fig. 7b). At the same time anterior cruciate ligament suffered from injury of the symptomatic infrapatellar plica in flexion and extension of the knee joint (Fig. 7c). Under arthroscopy, symptomatic infrapatellar plica is impacted on the adjacent cartilage and anterior cruciate ligament (ACL) in the flexion and extension of the knee joint (Fig. 7d). In some literatures, infrapatellar plica was described as its wing ligament. But we took the excised infrapatellar plica for pathological examinations under the arthroscope. Pathological examination showed that it consists of most synovial tissues, adipose tissue and some inflammatory cells (Fig. 9a). Degeneration of many synovial tissues was observed in histology. They demonstrate fibrosis, hyalinization, and calcification [19] as shown (Fig. 9b). The infrapatellar plica is often observed during the arthroscopic surgical operation [20]. The difference between the orthopedic and radiology is due to a lack of knowledge pertaining to the anatomical architecture. In addition, the treatments of symptomatic infrapatellar plica are limited, and the arthroscopic imaging and their Medical features are vivid.
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082
ARTICLE IN PRESS
JID: COMM
[m5G;September 25, 2019;16:25]
X. Tian, J. Wang and D. Du et al. / Computer Methods and Programs in Biomedicine xxx (xxxx) xxx
9
Fig. 9. Pathological study based on symptomatic infrapatellar plica. According to the pathological investigation, the symptomatic infrapatellar plica consists of synovial tissues, adipose tissues and inflammatory cells (a); and the synovial tissues demonstrate fibrosis, hyalinization, and calcification (b).
5. Conclusion Medical imaging coupled with enhanced image processing is a necessary and effective advanced method to make a diagnosis and give treatment for infrapatellar plica. The findings in this study demonstrate a significant reduction in knee pain and improvement in function by releasing and removal of the symptomatic infrapatellar plica under arthroscopic surgery. The image processed by the improved Laplacian image enhancement algorithm can effectively retain the image details, which is conducive to diagnosis and improve the diagnostic accuracy. Compliance with ethical standards The study was funded by personal (XDT) research and development fund (303-02-01-06-04). And authors whose names appear on the manuscript have contributed sufficiently to the scientific work and therefore share collective responsibility and accountability for the results. The authors declare that they have no conflict of interest. All procedures performed in studies involving human participants were in accordance with the ethical standards of the institutional and/or national research committee and with the 1964 Helsinki declaration and its later amendments or comparable ethical standards. Informed consent was obtained from all individual participants included in the study. Declaration of Competing Interest The authors declared that there was no conflict of interest in this study. CRediT authorship contribution statement Xiangdong Tian: Conceptualization, Formal analysis, Visualization, Writing - original draft. Jian Wang: Formal analysis, Visualization, Writing - original draft, Resources, Software. Dongfeng Du: Data curation, Project administration, Methodology. Shuwen Li: Data curation. Changxiao Han: Methodology. Guangyu Zhu: Formal analysis, Project administration. Yetong Tan: Formal analysis. Ming Lei: Data curation.
thopedic Center of The Third Affiliated Hospital of Beijing University of Chinese Medicine. This work was supported in part by Graduate school of Beijing University of Chinese Medicine, Beijing. References [1] N.L. Tindel, B. Nisonson, The plica syndrome, Orthop. Clin. North Am. 23 (1992) 613–618. [2] W.T. Hardaker, T.L. Whipple, F.H. Bassett, Diagnosis and treatment of the plica syndrome of the knee, J. Bone Joint Surg. Am. 62 (2) (1980) 221–225. [3] B. Reider, J.L. Marshall, R.F. Warren, Persistent vertical septum in the human knee joint, Brief Note 63 (7) (1981) 1185–1187. [4] P.Y.F. Lee, et al., Synovial plica syndrome of the knee: a commonly overlooked cause of anterior knee pain, Surg. J. 1 (3) (2017) 9–16. [5] D. Patel, Arthroscopy of the plicae-synovial folds and their significance, Am. J. Sports Med. 6 (1978) 217–225. [6] D. Patel, Plica as a cause of anterior knee pain, Orthop. Clin. North Am. 17 (2) (1986) 273–277. [7] G.L. Muse, W.A. Grana, S. Hollingsworth, Arthroscopic treatment of medial shelf syndrome, Arthroscopy 1 (1) (1985) 63–67. [8] S.J. Kim, B.H. Min, H.K. Kim, Arthroscopic anatomy of the infrapatellar plica, Arthroscopy 12 (5) (1996) 561–564. [9] A.M. Jackson, Anterior knee pain, J. Bone Joint Surg. Br. 83 (2001) 937–948. [10] S.J. Kim, W. Choe, Pathological infrapatellar plica: a report of two cases and literature review, Arthroscopy 12 (2) (1996) 236–239. [11] S.J. Kim, J.Y. Kim, J.W. Lee, Pathologic infrapatellar plica, Arthroscopy 18 (5) (2002) 25–28. [12] D. Kohn, S. Deiler, M. Rudert, Arterial blood supply of the infrapatellar fat pad anatomy and medical consequences, Arch. Orthop. Trauma Surg. 114 (1995) 72–75. [13] M. Magi, A. Branca, C. Bucca, et al., Hoffa disease, Ital. J. Orthop. Trauma 17 (1991) 675–691. [14] F.J. Koserek, C.A. Helms, The MR appearance of the infrapatellar plica, Am. J. Roentgenol. 172 (2) (1999) 481–484. [15] E. Stathopulu, E. Baildam, Anterior knee pain: a long term follow-up, Rheumatology 42 (2003) 380–382. [16] R. Garcia-Valtuille, F. Abascal, L. Cerezal, et al., Anatomy and MR imaging appearances of synovial plicae of the knee, Radiographics 22 (2002) 775–784. [17] N.M. Major, C.A. Helms, MR imaging of the knee: findings in asymptomatic collegiate basketball players, Am. J. Roentgenol. 179 (2002) 641–644. [18] R.L. Cothran, P.M. McGuire, C.A. Helms, et al., MR imaging of infrapatellar plica injury, Am. J. Roen. 180 (5) (2003) 1443–1447. [19] Christine R. Boyd, Colin Eakin, Gordon O. Matheson, Infrapatellar plica as a cause of anterior knee pain, Clin. J. Sport Med. 15 (2005) 98–103. [20] D.P. Johnson, D.M. Eastwood, P.J. Witherow, Symptomatic synovial plicae of the knee, J. Bone Joint Surg. Am. 75 (1993) 1485–1496. [21] C.-W. Ha, Y.-B. Park, C.-H. Choi, et al., Efficacy and safety of single injection of cross-linked sodium hyaluronate vs. three injections of high molecular weight sodium hyaluronate for osteoarthritis of the knee: a double-blind, randomized, multi-center, non-inferiority study, BMC Musculoskelet Disord. 18 (1) (2017) 223.
Acknowledgement The authors express their appreciation for the Medical support of orthopedics and valuable discussions with surgeons at the Or-
Please cite this article as: X. Tian, J. Wang and D. Du et al., Medical imaging and diagnosis of subpatellar vertebrae based on improved Laplacian image enhancement algorithm, Computer Methods and Programs in Biomedicine, https://doi.org/10.1016/j.cmpb.2019.105082