- Email: [email protected]
Three-dimensional versus two-dimensional sonography of the temporomandibular joint in comparison to MRI
European Journal of Radiology 61 (2007) 235–244
Three-dimensional versus two-dimensional sonography of the temporomandibular joint in comparison to MRI Constantin A. Landes a,∗ , Wojciech A. Goral a,1 , Robert Sader a,2 , Martin G. Mack b,3 a
Oral, Maxillofacial & Plastic Facial Surgery, The Frankfurt University Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt am Main, Germany b Department of Diagnostic and Interventional Radiology, The Frankfurt University Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt am Main, Germany Received 14 July 2006; received in revised form 21 September 2006; accepted 28 September 2006
Abstract Aim: To compare clinical feasibility of static two-dimensional (2D) to three-dimensional (3D) sonography of the temporomandibular joint (TMJ) in assessment of disk dislocation and joint degeneration compared to magnetic resonance imaging (MRI). Method: Thirty-three patients, 66 TMJ were prospectively sonographed 2D and 3D (8–12.5 MHz step motor scan), in occlusion and maximum opening with a probe position parallel inferior to the zygomatic arch. Axial 2D images were judged independent from the 3D scans; 3D volumes were cut axial, sagittal, frontal and rotated in real-time. Disk position and joint degeneration were assessed and compared to a subsequent MRI examination. Results: The specific appearance of the disk was hypoechogenic overlying a hyperechogenic condyle in axial (2D) or sagittal and frontal (3D) viewing. Specificity of 2D sonography for disk dislocation was 63%, sensitivity 58%, accuracy 64%, positive predictive value 46%, negative predictive value 73%; for joint degeneration synonymously 59/68/61/38/83%. 3D sonography for disk displacement reached synonymously 68/60/69/51/76%, for joint degeneration 75/65/73/48/86%. 2D sonographic diagnoses of disk dislocation in the closed mouth position and of joint degeneration showed significantly different results from the expected values (MRI) in χ2 testing; 3D diagnoses of disk dislocation in closed mouth position, of joint degeneration, 2D and 3D diagnoses in open mouth position were nonsignificant. Conclusions: Acceptable was the overall negative predictive value, as specificity and accuracy for joint degeneration in 3D. 3D appears superior diagnosing disk dislocation in closed mouth position as for overall joint degeneration. Sensitivity, accuracy and positive predictive value will have to ameliorate with future equipment of higher resolution in real-time 2D and 3D, if sonographic screening shall be clinically applied prior to MRI. © 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: Ultrasound; Computer applications 3D; Face; MR imaging
1. Introduction
∗
Corresponding author. Tel.: +49 69 6301 5643; fax: +49 69 6301 5644. E-mail addresses: [email protected] (C.A. Landes), [email protected] (W.A. Goral), [email protected] (R. Sader), [email protected] (M.G. Mack). 1 Present address: Oral, Maxillofacial & Plastic Facial Surgery, J.W. GoetheUniversity Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt/Main, Germany. Tel.: +49 17624520209. 2 Present address: Oral, Maxillofacial & Plastic Facial Surgery, J.W. GoetheUniversity Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt/Main, Germany. Tel.: +49 6963015643; fax: +49 6963015644. 3 Present address: Diagnostic and Interventional Radiology, J.W. GoetheUniversity Medical Centre, Frankfurt/Main, Germany. Tel.: +49 6963017277; fax: +49 6963017258. 0720-048X/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2006.09.015
Temporomandibular joint dysfunction (TMD) is a common disorder presenting with multiple symptoms of clinical joint dysfunction. A clinical classification [1] today is supported by radiological assessment using magnetic resonance imaging (MRI) [2]. The results repeatedly circumstantiated dislocated disk position, disk perforation, fibrosis [3–5] and joint degeneration [6–8] to be associated with clinical dysfunction. Exact diagnosis is necessary for the appropriate treatment, as joint degeneration can progress causing headaches, joint sounds, irregular, restricted jaw motion and tinnitus, that may require beyond conservative measures (splinting, physiotherapy [9]), surgical treatment as arthroscopy or arthroplasty [10]. MRI of the temporomandibular joint (TMJ) is the up-to-date diagnostic, widely used as it yields excellent anatomic detail in
236
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
static examination with high sensitivity, specificity and accuracy [2]. Fast imaging techniques permit reconstructed, alas no real-time motion cycles [11]. The use of MRI is limited by availability, costs, and the long examination time. On the other hand high-frequency sonographic transducers have been developed, and there are reports in the literature that two-dimensional (2D) sonography can assess condylar translation, disk position and osteoarthrosis fast and reliably [12–19]. TMJ sonography is
comfortable for the patient, highly available, transportable, fast and cost effective. However, multiplanar viewing is not yet possible but desirable, as due to traction forces of the upper belly of the lateral pterygoid muscles, anteromedial disk dislocation is the most frequent direction of dislocation. One of the main drawbacks of TMJ 2D sonography is the fact that the medial aspect of the joint cannot be visualized. Three-dimensional (3D) sonography is included into an increasing number of sonographic
Fig. 1. The standard (a) patient (a volunteer in this case) position for sonographic TMJ examination. The transducer was positioned parallel and inferior to the zygomatic arch (parallel to the Camper-line). (b) The 2D closed (left), and open mouth (right) condyle position and (c) the outlined anatomical structures of the identical sonogram which is an axial section reflecting the echoes from the lateral condyle head, overlying disk, capsule, anterior lying masseter muscle and lateral parotid gland. (d) A similar MRI cut, axial transecting the condylar head and the anterior lateral overlying disk. However, the insonation angle is slightly upward and the MRI cut is perpendicular and axial, therefore the zygomatic arch lies anterior in (d and e).
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
237
Fig. 1. (Continued ).
equipments and permits multiplanar viewing, possibly more insight into the medial aspect of the joint and therefore superior diagnostic key data. This study evaluates the benefit of three-dimensional sonography comparing static 3D sonograms to static 2D B-scan and MRI. 2. Patients and methods From May to July 2004, 33 consecutive TMD patients referred to the TMJ clinic at our department (23 females, 10 males, mean age 34, 14–77 years) volunteered for prospective static 2D sonographic B-scan, 3D sonography and TMJ-MRI. All reported pain and dysfunction of the temporomandibular area. Patients agreed to the study procedure with informed consent and the guidelines of the declaration of Helsinki were maintained throughout. All 2D and 3D sonographic trials were performed by the first author on upright sitting patients using Voluson 730 equipment (General Electric-Kretz, Solingen, Germany). The transducer was an 8–12.5 MHz linear array, angulated by step motor in case of 3D scan. Probe positioning was parallel to the zygomatic arc; lateral to the joint and lateral condyle pole, the transducer was tilted for optimal insonation angle to the joint depending upon individual condyle-fossa conformation (Fig. 1). The orientation of the transducer was standardized according to the standard planes of head and neck sonography [20]. Static images were obtained in occlusion and maximal opening to adhere to repro-
ducible jaw positions in sonographic as in MRI examination. In 3D sonography after a 25◦ transducer sweep performed by step motor, the three-dimensional joint was reproduced in a tissue block, inspected and intersected deliberately sagittal, frontal and coronar. All images were stored on harddisk. At a later date the static images were scrutinized on-screen in randomized, independent, blinded fashion, i.e. consecutive sonograms of one patient were not interpreted successively. The investigator was blinded regarding clinical patient information. All images were mixed and interpreted without knowledge of the patient’s name or clinical examination results; scrutinized for disk position, effusion, disk degeneration, perforation. Furthermore scan-images were evaluated for condylar surface irregularity, flattening, exostoses, thickness of the lateral articular space and thickened or irregular joint capsule. The first (sonography) and the last author’s (MRI) diagnoses were reported independently to the second author for unbiased statistical evaluation. The disk position was considered normal, if the posterior ending of the hypoechogenic disk was located in a 12-o’clock or superior position relative to the condyle, which in case of a high fossa could also be slightly more anterior (e.g. 11o’clock). Disk displacement was diagnosed in patients in whom the posterior ending of the disk was situated anterior (10o’clock and below) or anterolateral relative to the condyle. If a disk–condyle relationship could not be appropriately differentiated, this was noted. Disk degeneration was diagnosed when the disk evinced increased echogenity, flattening or reduced mobility. TMJ osteoarthrosis was defined by the presence of condylar
238
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
flattening, surface irregularity, erosions, deformities (irregularities in the cortical surface) and exostoses (bone hypertrophy at the condylar surface). The patient was positioned in supine position for MRI imaging, performed by the last author with a 1.5-T MR scanner (Magnetom Vision; Siemens, Erlangen, Germany) and a dedicated polarized transmit–receive TMJ coil (7 cm diameter) which allows simultaneous examination of both temporomandibular joints. The imaging protocol included parasagittal and coronal T1 weighted spin echo (SE) sequences (TR/TE = 450/15 ms, imaging matrix 256 × 256, field of view 128 mm, pixel size 0.5 mm × 0.5 mm, slice thickness 3 mm) and parasagittal T2and proton density weighted turbo spin echo (TSE) sequence (TR/TE = 2840/103 − 15 ms, matrix 512 × 512, field of view 128 mm, pixel size 0.25 mm × 0.25 mm, slice thickness 3 mm). Images were corrected to the horizontal angulation. Sequential bilateral images were obtained in the closed mouth and maximum opening position. The MR images were interpreted without knowledge of the patient’s clinical or sonographic findings. MRI sequences were scrutinized for disk position, disk degeneration and osteoarthrosis. The disk position was considered normal if the posterior ligament of the disk was located in a 12-o’clock or superior position relative to the condyle. In high fossae or steep tubercular slopes this could also mean a slightly more anterior (11-o’clock) position. Disk displacement was diagnosed in patients in whom the posterior ending of the disk was situated anterior (10-o’clock and below), anterolateral, lateral, anteromedial or medial relative to the condyle. If a disk–condyle relationship could not be appropriately differentiated, this was noted. Disk degeneration was diagnosed when the disk had a thick posterior band, lengthening, was biconvex, folded, rounded, flattened or showed reduced mobility. TMJ osteoarthrosis was defined by the presence of condylar flattening, surface irregularity, erosion or presence of deformities, exostoses and subcortical sclerosis. The statistical parameters: sensitivity, specificity, accuracy, positive and negative predictive value of 2D and 3D static sonography were compared to MRI and evaluated for statistical differences using χ2 testing with α = 0.05 significance level
using Microsoft Excel 2003 (Microsoft Corp., Mountain View, CA, USA) and SPSS 14.0 (SPSS Inc., Chicago, IL, USA) software. 3. Results Sections parallel to the route of condylar translation were obtained by positioning the transducer in anterioposterior orientation parallel to the zygomatic arch. While sitting relaxed but erect, the patient moved the mandible from occlusion to maximal opening, to occlusion. The sitting position was considered more physiological or similar to patient posture when joint pain occurs during everyday situations. Static examination froze and saved images from occlusion and maximal mouth opening. In 2D sonography the disk was seen as a hypoechogenic band overlying the lateral condyle in-between the farther hyperechogenic line from the lateral condylar head and closer hyperechogenic line from the articular capsule (see Figs. 1 and 2). In joints with effusion, a broader hypoechogenic joint space surrounded the disk and lateral condyle pole without notable influence upon the diagnosis. 3D sonography permitted oversight in sequential images to assess the complete condylar head and disk including the joint capsule (Figs. 3 and 4). MRI documented disk dislocation in 30/66 TMJ in closed position and in 15/66 TMJ in open position. 2D sonography attained 63% (47% closed position, 78% open position) specificity. Sensitivity was 58% (63/43%), accuracy 64% (55/73%), positive predictive value 46% (50/42%), negative predictive value 73% (61/85%). 3D sonography attained 68% (53% closed, 82% open position) specificity. Sensitivity was 60% (67/53%), accuracy 69% (61/77%), positive predictive value 51% (54/47%), negative predictive value 76% (66/86%) (Tables 1–4). MRI documented disk degeneration in 18/66 TMJ and osteoarthrosis in 16/66 TMJ. 2D sonography attained 59% (44% disk degeneration, 74% osteoarthrosis) specificity. Sensitivity was 68% (67/69%), accuracy 61% (50/73%), positive predictive value 38% (31/46%), negative predictive value 83% (78/88%). 3D sonography attained 75% (73% disk degeneration, 78%
Fig. 2. A nonrepositioning disk displacement (a) with the outlined structures (b).
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
239
Fig. 3. (a) 3D-closed mouth sonogram. Above left the frontal, above right the top view and below left the lateral view. Below right the 3D view based upon the cut of the 3D block deliberately chosen by the investigator as pictured by the thin red line. (b) A normal disk position with closed mouth in axial section similar to the 2D image and (c) outlined structures, (d) open mouth position and (e) outlined structures (sagittal views). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
osteoarthrosis) specificity. Sensitivity was 65% (61/69%), accuracy 73% (70/76%), positive predictive value 48% (46/50%), negative predictive value 86% (83/89%) (Tables 5–8). It was helpful for the diagnosis in 3D sonography to cut the data block in several transections as multiplanar viewing. Initially, an axial section was preferred just transecting the upper condylar pole and disk, substituted later-on by a coro-
nar and sagittal section. The sagittal section was the most important for diagnostic evaluation of disk position, because it allowed easy determination of anterioposterior disk position. The coronar section proved to be helpful in judging the degree of median disk dislocation. The insight to the median joint region is better in 3D, but still limited by effacement of echoes behind highly echogenic osseous joint surfaces. Diag-
240
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
Fig. 4. (a) A dislocated disk position with closed mouth and marked condylar irregularity. (b) Outlined structures in sagittal aspect; (c) closed mouth transversal view and (d) outlined structures. The persistent extinction of echo behind the highly echogenic condyle towards the medial joint can be seen on the transversal cut. (e) Open mouth position with repositioned yet flattened and irregular disk and (f) outlined structures sagittal, (g) transversal view and (h) outlined structures.
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
241
Table 1 Crosstabulation of 2D vs. MRI diagnosing disk dislocation in closed mouth position with a significant difference from the expected to the actual results (p = 0.04 in χ2 testing with α = 0.05) Disk dislocation mouth closed + 2D sonography + 19 (TP) − 11 (FN) All with disk dislocation (30) Sensitivity = TP/(TP + FN) = 0.63
− 19 (FP) 17 (TN)
All with positive test (38) All with negative test (28)
Positive predictive value = TP/(TP + FP) = 0.5 Negative predictive value = TN/(FN + TN) = 0.61
All without disk dislocation (36) Specificity = TN/(FP + TN) = 0.47
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.45
True positives (TP), false positives (FP), true negatives (TN) and false negatives (FN) are given, as well as the statistical parameters derived therefrom. Table 2 Crosstabulation of 2D vs. MRI diagnosing disk dislocation with open mouth position and no significant difference from the expected to the prevalent results (p = 0.27 in χ2 testing with α = 0.05) Disk dislocation mouth open + 2D sonography + 8 (TP) − 7 (FN) All with disk dislocation (15) Sensitivity = TP/(TP + FN) = 0.43
− 11 (FP) 40 (TN)
All with positive test (19) All with negative test (47)
Positive predictive value = TP/(TP + FP) = 0.42 Negative predictive value = TN/(FN + TN) = 0.85
All without disk dislocation (51) Specificity = TN/(FP + TN) = 0.78
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.23
Table 3 Crosstabulation of 3D vs. MRI diagnosing disk dislocation with closed mouth position and no significant difference from the expected to the prevalent results (p = 0.08 in χ2 testing with α = 0.05) Disk dislocation mouth closed + 3D sonography + 20 (TP) − 10 (FN) All with disk dislocation (30) Sensitivity = TP/(TP + FN) = 0.67
− 17 (FP) 19 (TN) All without disk dislocation (36) Specificity = TN/(FP + TN) = 0.53
nosis of disk dislocation in the closed mouth position differed significantly from the expected (MRI) results in χ2 testing with α = 0.05 significance level in 2D (p = 0.04), not in 3D (p = 0.08); nonsignificant for the open mouth position (2D p = 0.3 and 3D p = 0.6). Disk degeneration assessment differed highly significant in 2D (p = 0.000) and not in 3D (p = 0.12) sonographic examination. Osteoarthrosis was significantly different in 2D
All with positive test (37) Positive predictive value = TP/(TP + FP) = 0.54 All with negative test (29) Negative predictive value = TN/(FN + TN) = 0.66 Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.45
and not in 3D sonography (p = 0.04 and p = 0.11). χ2 Testing for significant differences of 2D versus 3D sonography was significant (p = 0.03). Significances versus MRI were generally higher in 2D; 3D was nonsignificant in diagnosis of a disk dislocation in closed mouth position as joint degeneration. Therefore, a higher diagnostic value may be attributed to 3D sonography.
Table 4 Crosstabulation of 3D vs. MRI diagnosing disk dislocation with open mouth position and no significant difference from the expected to the prevalent results (p = 0.57 in χ2 testing with α = 0.05) Disk dislocation mouth open + 3D sonography + 8 (TP) − 7 (FN) All with disk dislocation (15) Sensitivity = TP/(TP + FN) = 0.53
− 9 (FP) 42 (TN)
All with positive test (17) All with negative test (49)
Positive predictive value = TP/(TP + FP) = 0.47 Negative predictive value = TN/(FN + TN) = 0.86
All without disk dislocation (51) Specificity = TN/(FP + TN) = 0.82
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.24
242
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
Table 5 Crosstabulation of 2D vs. MRI diagnosing disk degeneration and highly significant difference from the expected to the prevalent results (p = 0.0000 in χ2 testing with α = 0.05) Disk degeneration + 2D sonography + 12 (TP) − 6 (FN) All with disk degeneration (18) Sensitivity = TP/(TP + FN) = 0.67
− 27 (FP) 21 (TN)
All with positive test (39) All with negative test (27)
Positive predictive value = TP/(TP + FP) = 0.31 Negative predictive value = TN/(FN + TN) = 0.78
All without disk degeneration (48) Specificity = TN/(FP + TN) = 0.44
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.27
Table 6 Crosstabulation of 3D vs. MRI diagnosing disk degeneration and no significant differences from the expected to the prevalent results (p = 0.12 in χ2 testing with α = 0.05). Disk degeneration + 3D sonography + 11 (TP) − 7 (FN) All with disk dislocation (18) Sensitivity = TP/(TP + FN) = 0.61
− 13 (FP) 35 (TN)
All with positive test (24) All with negative test (42)
Positive predictive value = TP/(TP + FP) = 0.46 Negative predictive value = TN/(FN + TN) = 0.83
All without disk dislocation (48) Specificity = TN/(FP + TN) = 0.73
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.27
Table 7 Crosstabulation of 2D vs. MRI diagnosing osteoarthrosis and significant difference from the expected to the prevalent results (p = 0.04 in χ2 testing with α = 0.05). Osteoarthrosis + 2D sonography + 11 (TP) − 5 (FN) All with osteoarthrosis (16) Sensitivity = TP/(TP + FN) = 0.69
− 13 (FP) 37 (TN)
All with positive test (24) All with negative test (42)
Positive predictive value = TP/(TP + FP) = 0.46 Negative predictive value = TN/(FN + TN) = 0.88
All without osteoarthrosis (50) Specificity = TN/(FP + TN) = 0.74
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.24
Table 8 Crosstabulation of 3D vs. MRI diagnosing osteoarthrosis and no significant difference from the expected to the prevalent results (p = 0.11 in χ2 testing with α = 0.05) Osteoarthrosis + 3D sonography + 11 (TP) − 5 (FN) All with disk dislocation (16) Sensitivity = TP/(TP + FN) = 0.69
− 11 (FP) 39 (TN)
All with positive test (22) All with negative test (44)
All without disk dislocation (50) Specificity = TN/(FP + TN) = 0.78
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.24
4. Discussion Close association of TMD with disk displacements and osteoarthrosis has been repeatedly emphasized. Clinical assessment today is supported by MRI which is limited by patient compliance, availability, cost, and the quite long examination time. Sonography may yield a fast on-site screening diagnostic to exclude normal joints, before an MRI is performed. However, this would require a high negative predictive value. Splint therapy of disk dislocation and osteoarthrosis has been successful
Positive predictive value = TP/(TP + FP) = 0.50 Negative predictive value = TN/(FN + TN) = 0.89
[21,22]; but in cases of progressive arthritis, arthroscopic lavage, lysis and arthroplasty are performed today [10]. Therefore, correct pretherapeutic diagnosis is necessary to prevent potentially dangerous under- or overtreatment. An earlier report comparing static with dynamic 2D scan (disk position) can be compared to this report whose results are given in parentheses [15]. Static 2D sonography showed a sensitivity of 41% (this study 58% 2D, 60% 3D), specificity of 70% (this study 63% 2D, 68% 3D) and dynamic 2D sonography a sensitivity of 31% and a specificity of 95%. The data revealed that static 2D sonography was marginal
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
in detecting the presence of disk displacement, but dynamic 2D sonography was sensitive in detecting the absence of disk displacement. However, with a positive predictive value of 61% (this study 46% 2D, 51% 3D) and a negative predictive value of 51% (this study 73% 2D, 76% 3D) for static 2D sonography, and a positive predictive value of 88% and a negative predictive value of 55% for the dynamic 2D technique, the results indicated that both modalities are yet insufficient in establishing a highly reliable diagnosis for the presence or absence of disk displacement but dynamic technique showed higher diagnostic key data [15]. Recent comparison to cryosections showed 73% accuracy for disk displacement (this study 64% 2D, 69% 3D versus MRI), 93% for condylar erosion (this study 61% 2D, 73% 3D versus MRI) [16]. Differences may be due to equipment quality and investigator experience. Also the total blinding (one examination was not judged in a row but randomized from the thesaurus, picture by picture) in this evaluation may have downgraded the results. Scanning is mostly performed vertical or horizontal to the joint to obtain either frontal or axial sections. The patient either is sitting or in supine position [15,17]. The transducer should be tilted to optimal insonation angle. Bone landmarks of the mandibular condyle and the articular eminence are identified as hyperechogenic lines in vertical and in horizontal transducer position. The disk corresponds to a thin homogeneous hypoechoic band, lying adjacent to the lateral pole of the mandibular condyle. During the dynamic evaluation, the ultrasound beam must be kept in identical orientation to the disk surface. This avoids artefact related changes of disk echogenity which is a primary disadvantage of the vertical positioning [12]. Scanning 60◦ or more off the perpendicular in relation to the long axis of the disk leads to invisibility of the disk. Scanning parallel to the joint space is limited due to a lack of bony landmarks. However, it could be shown by this and previous studies [12–14] how joint-parallel transducer position combines information of disk position and joint degeneration with a metric assessment of translation distance and eventually dynamic visualization of the disk and associated ligament behaviour. If the beam does not intersect the disk at its lateral portion in the axial plane, higher echogenity results from vagabonding echoes between the osseous surfaces [24] and the disk appears scarred with higher echogenity, flat and dehydrated. The medial joint is not accessible with 2D sonography and thus a medial disk position cannot be seen, a potential reason for false negative results. The tissue block obtained by 3D sonography can be easily cut into several planes and allows multiplanar viewing of a joint portion situated slightly more towards the medial joint. This made the diagnosis of combined dislocations easier and 3D in-depth view of the hypoechogenic disk was possible. The better 3D sonographic results for disk dislocation in closed mouth position and joint degeneration versus MRI compared to 2D sonography may be due to the better viewing. A deep fossa with a physiologically further anteriorly situated disk may still become falsely diagnosed dislocated. Severe osteophytes may theoretically impede disk visualization, which has not been noted within this evaluation. However, as the difference in diagnostic key data is small but significant, future improvement of sonographic equipment
243
should be expected before this is addressed with another study. In 3D sonography, the identical transducer position was used as anatomic orientation and maintenance of transducer-skin contact proved easier than a position perpendicular to the joint space. The observers have been involved with TMJ sonography and, respectively, MRI over 5 years and the learning curve has entered a steady slope. As individual observer variation can influence the result, retest examinations were performed and showed 93% concordance [12]. As ultrasound is always user-dependent, it has been the experience of the authors, that in order to achieve a certain standard and expertise in TMJ sonography, regular examination training upon a weekly basis for a year offers certainty as repeat reliability. Major TMD-clinics may prospectively benefit, offering such a diagnostic screening. MRI stays partly artificial in examination setting and an anxious patient may exhibit stronger dislocation as a consequence of bruxism in lying position. MRI is limited by speed (not real-time), cost and availability. Nevertheless cryosections have shown the very high correlation of MRI findings to post mortem morphological evaluation (accuracy 95% for disk position, 93% for osseous changes) and therefore it seems justified to correlate to MRI as gold standard in patient diagnostics [23]. Alternative arthrography and arthroscopy are invasive procedures and may cause pain, infection, allergic reaction and joint trauma. The 2D as the 3D static examination are fast and comfortable for the patient. Through fair specificity and high negative predictive value, TMJ sonography either 2D or prospectively favoured 3D appear suitable as a screening examination in TMD. Once resolution and contrast increase further and real-time 3D (4D) will be regularly available, it should have similar advantages as 2D real-time over static 2D. The principal advantage of 3D sonography was to obtain a more complete overview of the lateral condyle and disk, also more in the direction of the medial joint, not merely a transection. This made the interpretation more feasible, the transducer angulation has not got to be kept as exact as in 2D. This is because the step motor moves the probe for 25◦ . So if the ideal insonation angle is missed by the initial probe position, the movement of the probe by step motor adjusts for such inaccuracy as the obtained block of scanned tissue is cut at investigator will. Yet the upper margin of the selected three-dimensional “box” from the two-dimensional pilot image has to be adjusted as a box-cube and cannot be tailored to the disk–condyle arrangement. This would make direct interpretation of the resulting virtual tissue block easier and render virtual cuts superfluous. Diagnostic capacity may prospectively become reinforced by real-time viewing and higher equipment resolution (under www.constantinlandes.net a real-time normal joint in 2D- as well as 3D-dynamic viewing, performing opening, protrusion and mediotrusion can be seen; a flattened and arthritic joint with nonrepositioning anterior disk dislocation in 2D and 3D real-time viewing with a irregular condyle shape and nonrepositioning disk). Automatic structure recognition could prospectively implement visualization routines (colour-coding, transparency modification in favour of disk and condyle morphology, blinding out surrounding structures), that are even more end-user friendly.
244
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
5. Conclusions This investigation showed some value in static 2D and 3D sonography for exclusion of TMJ pathology (good negative predictive value). Static 3D sonography shows superior for assessment of disk position and joint degeneration over static 2D. Sensitivity, accuracy and positive predictive value will have to ameliorate dramatically with future systems of higher resolution in real-time 2D and real-time 3D. Further research on the diagnostic capacity of 3D sonography that permits multidimensional joint visualization is required to justify future routine screening of patients with joint dysfunction. References [1] Helkimo M. Epidemiological surveys of dysfunction of the masticatory system. Oral Sci Rev 1976;7:54–69. [2] Foucart JM, Carpentier P, Pajoni D, Marguelles-Bonnet R, Pharaboz C. MR of 732 TMJs: anterior, rotational, partial and sideways disc displacements. Eur J Radiol 1998;28:86–94. [3] Westesson PL, Bronstein SL, Liedberg J. Internal derangement of the temporomandibular joint: morphologic description with correlation to joint function. Oral Surg Oral Med Oral Pathol 1985;59:323–31. [4] Marguelles-Bonnet RE, Carpentier P, Yung LP, Defrennes D, Pharaboz C. Clinical diagnosis compared with findings of magnetic resonance in 242 patients with internal derangement of the TMJ. J Orofac Pain 1995;9:244–53. [5] Ribeiro RF, Tallents RH, Katzberg RW, et al. The prevalence of disc displacement in symptomatic and asymptomatic volunteers aged 6 to 25 years. J Orofac Pain 1997;11:37–47. [6] Rasmussen OC. Temporomandibular arthropathy—clinical, radiologic and therapeutic aspects with emphasis on diagnosis. Int J Oral Surg 1983;12:365–97. [7] Westesson PL. Structural hard-tissue changes in temporomandibular joints with internal derangement. Oral Surg Oral Med Oral Pathol 1985;59:220–4. [8] De Leeuw R, Boering G, Stegenga B, de Bont LG. Radiographic signs of temporomandibular joint osteoarthrosis and internal derangement 30 years after nonsurgical treatment. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1995;79:382–92. [9] Okeson JP. Long-term treatment of disc-interference disorders of the temporomandibular joint with anterior repositioning occlusal splints. J Prosthet Dent 1988;60:611–5.
[10] Nyberg J, Adell R, Svensson B. Temporomandibular joint discectomy for treatment of unilateral internal derangements—a 5 year follow-up evaluation. Int J Oral Maxillofac Surg 2004;33:8–12. [11] Behr M, Held P, Leibrock A, Fellner C, Handel G. Diagnostic potential of pseudo-dynamic MRI (CINE mode) for evaluation of internal derangement of the TMJ. Eur J Radiol 1996;23:212–5. [12] Landes C, Walendzik W, Klein C. Sonography of the temporomandibular joint from 60 examinations and comparison with MRI and axiography. J Craniomaxillofac Surg 2000;28:352–61. [13] Landes CA, Sterz M. Evaluation of condylar translation by sonography versus axiography in orthognathic surgery patients. J Oral Maxillofac Surg 2003;61:1410–7. [14] Landes CA. Proximal segment positioning in bilateral sagittal split osteotomy: intraoperative dynamic positioning and monitoring by sonography. J Oral Maxillofac Surg 2004;62:22–8. [15] Emshoff R, Bertram S, Rudisch A, Gaßner R. The diagnostic value of ultrasonography to determine the temporomandibular disc position. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1997;84: 688–96. [16] Rudisch A, Emshoff R, Maurer H, Kovacs P, Bodner G. Pathologic– sonographic correlation in temporomandibular joint pathology. Eur Radiol 2006;16:1750–6. Epub 2006 Mar 1. [17] Hayashi T, Ito J, Koyoma J, Yamada K. The accuracy of sonography for evaluation of internal derangement of the temporomandibular joint in asymptomatic elementary school children: comparison with MR and CT. AJNR 2001;22:728–34. [18] Emshoff R, Brandlmaier I, Bodner G, Rudisch A. Condylar erosion and disc displacement: detection with high-resolution ultrasonography. J Oral Maxillofac Surg 2003;61:877–81. [19] Sader R, Zeilhofer HF, Deppe H, Kling B, Wagner-Manslau C. Video and computer supported high-frequency TMJ arthrosonography. Dtsch Z Mund Kiefer Gesichts Chir 1995;19:12–4. [20] Siegert R. Ultrasound documentation in the head and neck region. Dtsch Z Mund Kiefer Gesichts Chir 1987;11:67–9. [21] Truelove EL, Sommers EE, LeResche L, Dworkin SF, Von Korff M. Clinical diagnostic criteria for TMD: new classification permits multiple diagnosis. J Am Dent Assoc 1992;123:47–54. [22] Okeson JP. Long-term treatment of disc-interference disorders of the temporomandibular joint with anterior repositioning occlusal splints. J Prosthet Dent 1988;60:611–5. [23] Tasaki MM, Westesson PL. Temporomandibular joint: diagnostic accuracy with sagittal and coronal MR imaging. Radiology 1993;186:723–9. [24] Malzer U. Schlierenoptische Untersuchungen am isolierten Kniegelenksmodell, Marburg. Munich, Germany: Videoproduktion Fa. Siemens; 1992.
Three-dimensional versus two-dimensional sonography of the temporomandibular joint in comparison to MRI Constantin A. Landes a,∗ , Wojciech A. Goral a,1 , Robert Sader a,2 , Martin G. Mack b,3 a
Oral, Maxillofacial & Plastic Facial Surgery, The Frankfurt University Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt am Main, Germany b Department of Diagnostic and Interventional Radiology, The Frankfurt University Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt am Main, Germany Received 14 July 2006; received in revised form 21 September 2006; accepted 28 September 2006
Abstract Aim: To compare clinical feasibility of static two-dimensional (2D) to three-dimensional (3D) sonography of the temporomandibular joint (TMJ) in assessment of disk dislocation and joint degeneration compared to magnetic resonance imaging (MRI). Method: Thirty-three patients, 66 TMJ were prospectively sonographed 2D and 3D (8–12.5 MHz step motor scan), in occlusion and maximum opening with a probe position parallel inferior to the zygomatic arch. Axial 2D images were judged independent from the 3D scans; 3D volumes were cut axial, sagittal, frontal and rotated in real-time. Disk position and joint degeneration were assessed and compared to a subsequent MRI examination. Results: The specific appearance of the disk was hypoechogenic overlying a hyperechogenic condyle in axial (2D) or sagittal and frontal (3D) viewing. Specificity of 2D sonography for disk dislocation was 63%, sensitivity 58%, accuracy 64%, positive predictive value 46%, negative predictive value 73%; for joint degeneration synonymously 59/68/61/38/83%. 3D sonography for disk displacement reached synonymously 68/60/69/51/76%, for joint degeneration 75/65/73/48/86%. 2D sonographic diagnoses of disk dislocation in the closed mouth position and of joint degeneration showed significantly different results from the expected values (MRI) in χ2 testing; 3D diagnoses of disk dislocation in closed mouth position, of joint degeneration, 2D and 3D diagnoses in open mouth position were nonsignificant. Conclusions: Acceptable was the overall negative predictive value, as specificity and accuracy for joint degeneration in 3D. 3D appears superior diagnosing disk dislocation in closed mouth position as for overall joint degeneration. Sensitivity, accuracy and positive predictive value will have to ameliorate with future equipment of higher resolution in real-time 2D and 3D, if sonographic screening shall be clinically applied prior to MRI. © 2006 Elsevier Ireland Ltd. All rights reserved. Keywords: Ultrasound; Computer applications 3D; Face; MR imaging
1. Introduction
∗
Corresponding author. Tel.: +49 69 6301 5643; fax: +49 69 6301 5644. E-mail addresses: [email protected] (C.A. Landes), [email protected] (W.A. Goral), [email protected] (R. Sader), [email protected] (M.G. Mack). 1 Present address: Oral, Maxillofacial & Plastic Facial Surgery, J.W. GoetheUniversity Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt/Main, Germany. Tel.: +49 17624520209. 2 Present address: Oral, Maxillofacial & Plastic Facial Surgery, J.W. GoetheUniversity Medical Centre, Theodor-Stern-Kai 7, 60596 Frankfurt/Main, Germany. Tel.: +49 6963015643; fax: +49 6963015644. 3 Present address: Diagnostic and Interventional Radiology, J.W. GoetheUniversity Medical Centre, Frankfurt/Main, Germany. Tel.: +49 6963017277; fax: +49 6963017258. 0720-048X/$ – see front matter © 2006 Elsevier Ireland Ltd. All rights reserved. doi:10.1016/j.ejrad.2006.09.015
Temporomandibular joint dysfunction (TMD) is a common disorder presenting with multiple symptoms of clinical joint dysfunction. A clinical classification [1] today is supported by radiological assessment using magnetic resonance imaging (MRI) [2]. The results repeatedly circumstantiated dislocated disk position, disk perforation, fibrosis [3–5] and joint degeneration [6–8] to be associated with clinical dysfunction. Exact diagnosis is necessary for the appropriate treatment, as joint degeneration can progress causing headaches, joint sounds, irregular, restricted jaw motion and tinnitus, that may require beyond conservative measures (splinting, physiotherapy [9]), surgical treatment as arthroscopy or arthroplasty [10]. MRI of the temporomandibular joint (TMJ) is the up-to-date diagnostic, widely used as it yields excellent anatomic detail in
236
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
static examination with high sensitivity, specificity and accuracy [2]. Fast imaging techniques permit reconstructed, alas no real-time motion cycles [11]. The use of MRI is limited by availability, costs, and the long examination time. On the other hand high-frequency sonographic transducers have been developed, and there are reports in the literature that two-dimensional (2D) sonography can assess condylar translation, disk position and osteoarthrosis fast and reliably [12–19]. TMJ sonography is
comfortable for the patient, highly available, transportable, fast and cost effective. However, multiplanar viewing is not yet possible but desirable, as due to traction forces of the upper belly of the lateral pterygoid muscles, anteromedial disk dislocation is the most frequent direction of dislocation. One of the main drawbacks of TMJ 2D sonography is the fact that the medial aspect of the joint cannot be visualized. Three-dimensional (3D) sonography is included into an increasing number of sonographic
Fig. 1. The standard (a) patient (a volunteer in this case) position for sonographic TMJ examination. The transducer was positioned parallel and inferior to the zygomatic arch (parallel to the Camper-line). (b) The 2D closed (left), and open mouth (right) condyle position and (c) the outlined anatomical structures of the identical sonogram which is an axial section reflecting the echoes from the lateral condyle head, overlying disk, capsule, anterior lying masseter muscle and lateral parotid gland. (d) A similar MRI cut, axial transecting the condylar head and the anterior lateral overlying disk. However, the insonation angle is slightly upward and the MRI cut is perpendicular and axial, therefore the zygomatic arch lies anterior in (d and e).
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
237
Fig. 1. (Continued ).
equipments and permits multiplanar viewing, possibly more insight into the medial aspect of the joint and therefore superior diagnostic key data. This study evaluates the benefit of three-dimensional sonography comparing static 3D sonograms to static 2D B-scan and MRI. 2. Patients and methods From May to July 2004, 33 consecutive TMD patients referred to the TMJ clinic at our department (23 females, 10 males, mean age 34, 14–77 years) volunteered for prospective static 2D sonographic B-scan, 3D sonography and TMJ-MRI. All reported pain and dysfunction of the temporomandibular area. Patients agreed to the study procedure with informed consent and the guidelines of the declaration of Helsinki were maintained throughout. All 2D and 3D sonographic trials were performed by the first author on upright sitting patients using Voluson 730 equipment (General Electric-Kretz, Solingen, Germany). The transducer was an 8–12.5 MHz linear array, angulated by step motor in case of 3D scan. Probe positioning was parallel to the zygomatic arc; lateral to the joint and lateral condyle pole, the transducer was tilted for optimal insonation angle to the joint depending upon individual condyle-fossa conformation (Fig. 1). The orientation of the transducer was standardized according to the standard planes of head and neck sonography [20]. Static images were obtained in occlusion and maximal opening to adhere to repro-
ducible jaw positions in sonographic as in MRI examination. In 3D sonography after a 25◦ transducer sweep performed by step motor, the three-dimensional joint was reproduced in a tissue block, inspected and intersected deliberately sagittal, frontal and coronar. All images were stored on harddisk. At a later date the static images were scrutinized on-screen in randomized, independent, blinded fashion, i.e. consecutive sonograms of one patient were not interpreted successively. The investigator was blinded regarding clinical patient information. All images were mixed and interpreted without knowledge of the patient’s name or clinical examination results; scrutinized for disk position, effusion, disk degeneration, perforation. Furthermore scan-images were evaluated for condylar surface irregularity, flattening, exostoses, thickness of the lateral articular space and thickened or irregular joint capsule. The first (sonography) and the last author’s (MRI) diagnoses were reported independently to the second author for unbiased statistical evaluation. The disk position was considered normal, if the posterior ending of the hypoechogenic disk was located in a 12-o’clock or superior position relative to the condyle, which in case of a high fossa could also be slightly more anterior (e.g. 11o’clock). Disk displacement was diagnosed in patients in whom the posterior ending of the disk was situated anterior (10o’clock and below) or anterolateral relative to the condyle. If a disk–condyle relationship could not be appropriately differentiated, this was noted. Disk degeneration was diagnosed when the disk evinced increased echogenity, flattening or reduced mobility. TMJ osteoarthrosis was defined by the presence of condylar
238
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
flattening, surface irregularity, erosions, deformities (irregularities in the cortical surface) and exostoses (bone hypertrophy at the condylar surface). The patient was positioned in supine position for MRI imaging, performed by the last author with a 1.5-T MR scanner (Magnetom Vision; Siemens, Erlangen, Germany) and a dedicated polarized transmit–receive TMJ coil (7 cm diameter) which allows simultaneous examination of both temporomandibular joints. The imaging protocol included parasagittal and coronal T1 weighted spin echo (SE) sequences (TR/TE = 450/15 ms, imaging matrix 256 × 256, field of view 128 mm, pixel size 0.5 mm × 0.5 mm, slice thickness 3 mm) and parasagittal T2and proton density weighted turbo spin echo (TSE) sequence (TR/TE = 2840/103 − 15 ms, matrix 512 × 512, field of view 128 mm, pixel size 0.25 mm × 0.25 mm, slice thickness 3 mm). Images were corrected to the horizontal angulation. Sequential bilateral images were obtained in the closed mouth and maximum opening position. The MR images were interpreted without knowledge of the patient’s clinical or sonographic findings. MRI sequences were scrutinized for disk position, disk degeneration and osteoarthrosis. The disk position was considered normal if the posterior ligament of the disk was located in a 12-o’clock or superior position relative to the condyle. In high fossae or steep tubercular slopes this could also mean a slightly more anterior (11-o’clock) position. Disk displacement was diagnosed in patients in whom the posterior ending of the disk was situated anterior (10-o’clock and below), anterolateral, lateral, anteromedial or medial relative to the condyle. If a disk–condyle relationship could not be appropriately differentiated, this was noted. Disk degeneration was diagnosed when the disk had a thick posterior band, lengthening, was biconvex, folded, rounded, flattened or showed reduced mobility. TMJ osteoarthrosis was defined by the presence of condylar flattening, surface irregularity, erosion or presence of deformities, exostoses and subcortical sclerosis. The statistical parameters: sensitivity, specificity, accuracy, positive and negative predictive value of 2D and 3D static sonography were compared to MRI and evaluated for statistical differences using χ2 testing with α = 0.05 significance level
using Microsoft Excel 2003 (Microsoft Corp., Mountain View, CA, USA) and SPSS 14.0 (SPSS Inc., Chicago, IL, USA) software. 3. Results Sections parallel to the route of condylar translation were obtained by positioning the transducer in anterioposterior orientation parallel to the zygomatic arch. While sitting relaxed but erect, the patient moved the mandible from occlusion to maximal opening, to occlusion. The sitting position was considered more physiological or similar to patient posture when joint pain occurs during everyday situations. Static examination froze and saved images from occlusion and maximal mouth opening. In 2D sonography the disk was seen as a hypoechogenic band overlying the lateral condyle in-between the farther hyperechogenic line from the lateral condylar head and closer hyperechogenic line from the articular capsule (see Figs. 1 and 2). In joints with effusion, a broader hypoechogenic joint space surrounded the disk and lateral condyle pole without notable influence upon the diagnosis. 3D sonography permitted oversight in sequential images to assess the complete condylar head and disk including the joint capsule (Figs. 3 and 4). MRI documented disk dislocation in 30/66 TMJ in closed position and in 15/66 TMJ in open position. 2D sonography attained 63% (47% closed position, 78% open position) specificity. Sensitivity was 58% (63/43%), accuracy 64% (55/73%), positive predictive value 46% (50/42%), negative predictive value 73% (61/85%). 3D sonography attained 68% (53% closed, 82% open position) specificity. Sensitivity was 60% (67/53%), accuracy 69% (61/77%), positive predictive value 51% (54/47%), negative predictive value 76% (66/86%) (Tables 1–4). MRI documented disk degeneration in 18/66 TMJ and osteoarthrosis in 16/66 TMJ. 2D sonography attained 59% (44% disk degeneration, 74% osteoarthrosis) specificity. Sensitivity was 68% (67/69%), accuracy 61% (50/73%), positive predictive value 38% (31/46%), negative predictive value 83% (78/88%). 3D sonography attained 75% (73% disk degeneration, 78%
Fig. 2. A nonrepositioning disk displacement (a) with the outlined structures (b).
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
239
Fig. 3. (a) 3D-closed mouth sonogram. Above left the frontal, above right the top view and below left the lateral view. Below right the 3D view based upon the cut of the 3D block deliberately chosen by the investigator as pictured by the thin red line. (b) A normal disk position with closed mouth in axial section similar to the 2D image and (c) outlined structures, (d) open mouth position and (e) outlined structures (sagittal views). (For interpretation of the references to colour in this figure legend, the reader is referred to the web version of the article.)
osteoarthrosis) specificity. Sensitivity was 65% (61/69%), accuracy 73% (70/76%), positive predictive value 48% (46/50%), negative predictive value 86% (83/89%) (Tables 5–8). It was helpful for the diagnosis in 3D sonography to cut the data block in several transections as multiplanar viewing. Initially, an axial section was preferred just transecting the upper condylar pole and disk, substituted later-on by a coro-
nar and sagittal section. The sagittal section was the most important for diagnostic evaluation of disk position, because it allowed easy determination of anterioposterior disk position. The coronar section proved to be helpful in judging the degree of median disk dislocation. The insight to the median joint region is better in 3D, but still limited by effacement of echoes behind highly echogenic osseous joint surfaces. Diag-
240
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
Fig. 4. (a) A dislocated disk position with closed mouth and marked condylar irregularity. (b) Outlined structures in sagittal aspect; (c) closed mouth transversal view and (d) outlined structures. The persistent extinction of echo behind the highly echogenic condyle towards the medial joint can be seen on the transversal cut. (e) Open mouth position with repositioned yet flattened and irregular disk and (f) outlined structures sagittal, (g) transversal view and (h) outlined structures.
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
241
Table 1 Crosstabulation of 2D vs. MRI diagnosing disk dislocation in closed mouth position with a significant difference from the expected to the actual results (p = 0.04 in χ2 testing with α = 0.05) Disk dislocation mouth closed + 2D sonography + 19 (TP) − 11 (FN) All with disk dislocation (30) Sensitivity = TP/(TP + FN) = 0.63
− 19 (FP) 17 (TN)
All with positive test (38) All with negative test (28)
Positive predictive value = TP/(TP + FP) = 0.5 Negative predictive value = TN/(FN + TN) = 0.61
All without disk dislocation (36) Specificity = TN/(FP + TN) = 0.47
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.45
True positives (TP), false positives (FP), true negatives (TN) and false negatives (FN) are given, as well as the statistical parameters derived therefrom. Table 2 Crosstabulation of 2D vs. MRI diagnosing disk dislocation with open mouth position and no significant difference from the expected to the prevalent results (p = 0.27 in χ2 testing with α = 0.05) Disk dislocation mouth open + 2D sonography + 8 (TP) − 7 (FN) All with disk dislocation (15) Sensitivity = TP/(TP + FN) = 0.43
− 11 (FP) 40 (TN)
All with positive test (19) All with negative test (47)
Positive predictive value = TP/(TP + FP) = 0.42 Negative predictive value = TN/(FN + TN) = 0.85
All without disk dislocation (51) Specificity = TN/(FP + TN) = 0.78
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.23
Table 3 Crosstabulation of 3D vs. MRI diagnosing disk dislocation with closed mouth position and no significant difference from the expected to the prevalent results (p = 0.08 in χ2 testing with α = 0.05) Disk dislocation mouth closed + 3D sonography + 20 (TP) − 10 (FN) All with disk dislocation (30) Sensitivity = TP/(TP + FN) = 0.67
− 17 (FP) 19 (TN) All without disk dislocation (36) Specificity = TN/(FP + TN) = 0.53
nosis of disk dislocation in the closed mouth position differed significantly from the expected (MRI) results in χ2 testing with α = 0.05 significance level in 2D (p = 0.04), not in 3D (p = 0.08); nonsignificant for the open mouth position (2D p = 0.3 and 3D p = 0.6). Disk degeneration assessment differed highly significant in 2D (p = 0.000) and not in 3D (p = 0.12) sonographic examination. Osteoarthrosis was significantly different in 2D
All with positive test (37) Positive predictive value = TP/(TP + FP) = 0.54 All with negative test (29) Negative predictive value = TN/(FN + TN) = 0.66 Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.45
and not in 3D sonography (p = 0.04 and p = 0.11). χ2 Testing for significant differences of 2D versus 3D sonography was significant (p = 0.03). Significances versus MRI were generally higher in 2D; 3D was nonsignificant in diagnosis of a disk dislocation in closed mouth position as joint degeneration. Therefore, a higher diagnostic value may be attributed to 3D sonography.
Table 4 Crosstabulation of 3D vs. MRI diagnosing disk dislocation with open mouth position and no significant difference from the expected to the prevalent results (p = 0.57 in χ2 testing with α = 0.05) Disk dislocation mouth open + 3D sonography + 8 (TP) − 7 (FN) All with disk dislocation (15) Sensitivity = TP/(TP + FN) = 0.53
− 9 (FP) 42 (TN)
All with positive test (17) All with negative test (49)
Positive predictive value = TP/(TP + FP) = 0.47 Negative predictive value = TN/(FN + TN) = 0.86
All without disk dislocation (51) Specificity = TN/(FP + TN) = 0.82
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.24
242
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
Table 5 Crosstabulation of 2D vs. MRI diagnosing disk degeneration and highly significant difference from the expected to the prevalent results (p = 0.0000 in χ2 testing with α = 0.05) Disk degeneration + 2D sonography + 12 (TP) − 6 (FN) All with disk degeneration (18) Sensitivity = TP/(TP + FN) = 0.67
− 27 (FP) 21 (TN)
All with positive test (39) All with negative test (27)
Positive predictive value = TP/(TP + FP) = 0.31 Negative predictive value = TN/(FN + TN) = 0.78
All without disk degeneration (48) Specificity = TN/(FP + TN) = 0.44
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.27
Table 6 Crosstabulation of 3D vs. MRI diagnosing disk degeneration and no significant differences from the expected to the prevalent results (p = 0.12 in χ2 testing with α = 0.05). Disk degeneration + 3D sonography + 11 (TP) − 7 (FN) All with disk dislocation (18) Sensitivity = TP/(TP + FN) = 0.61
− 13 (FP) 35 (TN)
All with positive test (24) All with negative test (42)
Positive predictive value = TP/(TP + FP) = 0.46 Negative predictive value = TN/(FN + TN) = 0.83
All without disk dislocation (48) Specificity = TN/(FP + TN) = 0.73
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.27
Table 7 Crosstabulation of 2D vs. MRI diagnosing osteoarthrosis and significant difference from the expected to the prevalent results (p = 0.04 in χ2 testing with α = 0.05). Osteoarthrosis + 2D sonography + 11 (TP) − 5 (FN) All with osteoarthrosis (16) Sensitivity = TP/(TP + FN) = 0.69
− 13 (FP) 37 (TN)
All with positive test (24) All with negative test (42)
Positive predictive value = TP/(TP + FP) = 0.46 Negative predictive value = TN/(FN + TN) = 0.88
All without osteoarthrosis (50) Specificity = TN/(FP + TN) = 0.74
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.24
Table 8 Crosstabulation of 3D vs. MRI diagnosing osteoarthrosis and no significant difference from the expected to the prevalent results (p = 0.11 in χ2 testing with α = 0.05) Osteoarthrosis + 3D sonography + 11 (TP) − 5 (FN) All with disk dislocation (16) Sensitivity = TP/(TP + FN) = 0.69
− 11 (FP) 39 (TN)
All with positive test (22) All with negative test (44)
All without disk dislocation (50) Specificity = TN/(FP + TN) = 0.78
Everyone (TP + FP + FN + TN) Pre-test probability = (TP + FN)/(TP + FP + FN + TN) = 0.24
4. Discussion Close association of TMD with disk displacements and osteoarthrosis has been repeatedly emphasized. Clinical assessment today is supported by MRI which is limited by patient compliance, availability, cost, and the quite long examination time. Sonography may yield a fast on-site screening diagnostic to exclude normal joints, before an MRI is performed. However, this would require a high negative predictive value. Splint therapy of disk dislocation and osteoarthrosis has been successful
Positive predictive value = TP/(TP + FP) = 0.50 Negative predictive value = TN/(FN + TN) = 0.89
[21,22]; but in cases of progressive arthritis, arthroscopic lavage, lysis and arthroplasty are performed today [10]. Therefore, correct pretherapeutic diagnosis is necessary to prevent potentially dangerous under- or overtreatment. An earlier report comparing static with dynamic 2D scan (disk position) can be compared to this report whose results are given in parentheses [15]. Static 2D sonography showed a sensitivity of 41% (this study 58% 2D, 60% 3D), specificity of 70% (this study 63% 2D, 68% 3D) and dynamic 2D sonography a sensitivity of 31% and a specificity of 95%. The data revealed that static 2D sonography was marginal
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
in detecting the presence of disk displacement, but dynamic 2D sonography was sensitive in detecting the absence of disk displacement. However, with a positive predictive value of 61% (this study 46% 2D, 51% 3D) and a negative predictive value of 51% (this study 73% 2D, 76% 3D) for static 2D sonography, and a positive predictive value of 88% and a negative predictive value of 55% for the dynamic 2D technique, the results indicated that both modalities are yet insufficient in establishing a highly reliable diagnosis for the presence or absence of disk displacement but dynamic technique showed higher diagnostic key data [15]. Recent comparison to cryosections showed 73% accuracy for disk displacement (this study 64% 2D, 69% 3D versus MRI), 93% for condylar erosion (this study 61% 2D, 73% 3D versus MRI) [16]. Differences may be due to equipment quality and investigator experience. Also the total blinding (one examination was not judged in a row but randomized from the thesaurus, picture by picture) in this evaluation may have downgraded the results. Scanning is mostly performed vertical or horizontal to the joint to obtain either frontal or axial sections. The patient either is sitting or in supine position [15,17]. The transducer should be tilted to optimal insonation angle. Bone landmarks of the mandibular condyle and the articular eminence are identified as hyperechogenic lines in vertical and in horizontal transducer position. The disk corresponds to a thin homogeneous hypoechoic band, lying adjacent to the lateral pole of the mandibular condyle. During the dynamic evaluation, the ultrasound beam must be kept in identical orientation to the disk surface. This avoids artefact related changes of disk echogenity which is a primary disadvantage of the vertical positioning [12]. Scanning 60◦ or more off the perpendicular in relation to the long axis of the disk leads to invisibility of the disk. Scanning parallel to the joint space is limited due to a lack of bony landmarks. However, it could be shown by this and previous studies [12–14] how joint-parallel transducer position combines information of disk position and joint degeneration with a metric assessment of translation distance and eventually dynamic visualization of the disk and associated ligament behaviour. If the beam does not intersect the disk at its lateral portion in the axial plane, higher echogenity results from vagabonding echoes between the osseous surfaces [24] and the disk appears scarred with higher echogenity, flat and dehydrated. The medial joint is not accessible with 2D sonography and thus a medial disk position cannot be seen, a potential reason for false negative results. The tissue block obtained by 3D sonography can be easily cut into several planes and allows multiplanar viewing of a joint portion situated slightly more towards the medial joint. This made the diagnosis of combined dislocations easier and 3D in-depth view of the hypoechogenic disk was possible. The better 3D sonographic results for disk dislocation in closed mouth position and joint degeneration versus MRI compared to 2D sonography may be due to the better viewing. A deep fossa with a physiologically further anteriorly situated disk may still become falsely diagnosed dislocated. Severe osteophytes may theoretically impede disk visualization, which has not been noted within this evaluation. However, as the difference in diagnostic key data is small but significant, future improvement of sonographic equipment
243
should be expected before this is addressed with another study. In 3D sonography, the identical transducer position was used as anatomic orientation and maintenance of transducer-skin contact proved easier than a position perpendicular to the joint space. The observers have been involved with TMJ sonography and, respectively, MRI over 5 years and the learning curve has entered a steady slope. As individual observer variation can influence the result, retest examinations were performed and showed 93% concordance [12]. As ultrasound is always user-dependent, it has been the experience of the authors, that in order to achieve a certain standard and expertise in TMJ sonography, regular examination training upon a weekly basis for a year offers certainty as repeat reliability. Major TMD-clinics may prospectively benefit, offering such a diagnostic screening. MRI stays partly artificial in examination setting and an anxious patient may exhibit stronger dislocation as a consequence of bruxism in lying position. MRI is limited by speed (not real-time), cost and availability. Nevertheless cryosections have shown the very high correlation of MRI findings to post mortem morphological evaluation (accuracy 95% for disk position, 93% for osseous changes) and therefore it seems justified to correlate to MRI as gold standard in patient diagnostics [23]. Alternative arthrography and arthroscopy are invasive procedures and may cause pain, infection, allergic reaction and joint trauma. The 2D as the 3D static examination are fast and comfortable for the patient. Through fair specificity and high negative predictive value, TMJ sonography either 2D or prospectively favoured 3D appear suitable as a screening examination in TMD. Once resolution and contrast increase further and real-time 3D (4D) will be regularly available, it should have similar advantages as 2D real-time over static 2D. The principal advantage of 3D sonography was to obtain a more complete overview of the lateral condyle and disk, also more in the direction of the medial joint, not merely a transection. This made the interpretation more feasible, the transducer angulation has not got to be kept as exact as in 2D. This is because the step motor moves the probe for 25◦ . So if the ideal insonation angle is missed by the initial probe position, the movement of the probe by step motor adjusts for such inaccuracy as the obtained block of scanned tissue is cut at investigator will. Yet the upper margin of the selected three-dimensional “box” from the two-dimensional pilot image has to be adjusted as a box-cube and cannot be tailored to the disk–condyle arrangement. This would make direct interpretation of the resulting virtual tissue block easier and render virtual cuts superfluous. Diagnostic capacity may prospectively become reinforced by real-time viewing and higher equipment resolution (under www.constantinlandes.net a real-time normal joint in 2D- as well as 3D-dynamic viewing, performing opening, protrusion and mediotrusion can be seen; a flattened and arthritic joint with nonrepositioning anterior disk dislocation in 2D and 3D real-time viewing with a irregular condyle shape and nonrepositioning disk). Automatic structure recognition could prospectively implement visualization routines (colour-coding, transparency modification in favour of disk and condyle morphology, blinding out surrounding structures), that are even more end-user friendly.
244
C.A. Landes et al. / European Journal of Radiology 61 (2007) 235–244
5. Conclusions This investigation showed some value in static 2D and 3D sonography for exclusion of TMJ pathology (good negative predictive value). Static 3D sonography shows superior for assessment of disk position and joint degeneration over static 2D. Sensitivity, accuracy and positive predictive value will have to ameliorate dramatically with future systems of higher resolution in real-time 2D and real-time 3D. Further research on the diagnostic capacity of 3D sonography that permits multidimensional joint visualization is required to justify future routine screening of patients with joint dysfunction. References [1] Helkimo M. Epidemiological surveys of dysfunction of the masticatory system. Oral Sci Rev 1976;7:54–69. [2] Foucart JM, Carpentier P, Pajoni D, Marguelles-Bonnet R, Pharaboz C. MR of 732 TMJs: anterior, rotational, partial and sideways disc displacements. Eur J Radiol 1998;28:86–94. [3] Westesson PL, Bronstein SL, Liedberg J. Internal derangement of the temporomandibular joint: morphologic description with correlation to joint function. Oral Surg Oral Med Oral Pathol 1985;59:323–31. [4] Marguelles-Bonnet RE, Carpentier P, Yung LP, Defrennes D, Pharaboz C. Clinical diagnosis compared with findings of magnetic resonance in 242 patients with internal derangement of the TMJ. J Orofac Pain 1995;9:244–53. [5] Ribeiro RF, Tallents RH, Katzberg RW, et al. The prevalence of disc displacement in symptomatic and asymptomatic volunteers aged 6 to 25 years. J Orofac Pain 1997;11:37–47. [6] Rasmussen OC. Temporomandibular arthropathy—clinical, radiologic and therapeutic aspects with emphasis on diagnosis. Int J Oral Surg 1983;12:365–97. [7] Westesson PL. Structural hard-tissue changes in temporomandibular joints with internal derangement. Oral Surg Oral Med Oral Pathol 1985;59:220–4. [8] De Leeuw R, Boering G, Stegenga B, de Bont LG. Radiographic signs of temporomandibular joint osteoarthrosis and internal derangement 30 years after nonsurgical treatment. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1995;79:382–92. [9] Okeson JP. Long-term treatment of disc-interference disorders of the temporomandibular joint with anterior repositioning occlusal splints. J Prosthet Dent 1988;60:611–5.
[10] Nyberg J, Adell R, Svensson B. Temporomandibular joint discectomy for treatment of unilateral internal derangements—a 5 year follow-up evaluation. Int J Oral Maxillofac Surg 2004;33:8–12. [11] Behr M, Held P, Leibrock A, Fellner C, Handel G. Diagnostic potential of pseudo-dynamic MRI (CINE mode) for evaluation of internal derangement of the TMJ. Eur J Radiol 1996;23:212–5. [12] Landes C, Walendzik W, Klein C. Sonography of the temporomandibular joint from 60 examinations and comparison with MRI and axiography. J Craniomaxillofac Surg 2000;28:352–61. [13] Landes CA, Sterz M. Evaluation of condylar translation by sonography versus axiography in orthognathic surgery patients. J Oral Maxillofac Surg 2003;61:1410–7. [14] Landes CA. Proximal segment positioning in bilateral sagittal split osteotomy: intraoperative dynamic positioning and monitoring by sonography. J Oral Maxillofac Surg 2004;62:22–8. [15] Emshoff R, Bertram S, Rudisch A, Gaßner R. The diagnostic value of ultrasonography to determine the temporomandibular disc position. Oral Surg Oral Med Oral Pathol Oral Radiol Endod 1997;84: 688–96. [16] Rudisch A, Emshoff R, Maurer H, Kovacs P, Bodner G. Pathologic– sonographic correlation in temporomandibular joint pathology. Eur Radiol 2006;16:1750–6. Epub 2006 Mar 1. [17] Hayashi T, Ito J, Koyoma J, Yamada K. The accuracy of sonography for evaluation of internal derangement of the temporomandibular joint in asymptomatic elementary school children: comparison with MR and CT. AJNR 2001;22:728–34. [18] Emshoff R, Brandlmaier I, Bodner G, Rudisch A. Condylar erosion and disc displacement: detection with high-resolution ultrasonography. J Oral Maxillofac Surg 2003;61:877–81. [19] Sader R, Zeilhofer HF, Deppe H, Kling B, Wagner-Manslau C. Video and computer supported high-frequency TMJ arthrosonography. Dtsch Z Mund Kiefer Gesichts Chir 1995;19:12–4. [20] Siegert R. Ultrasound documentation in the head and neck region. Dtsch Z Mund Kiefer Gesichts Chir 1987;11:67–9. [21] Truelove EL, Sommers EE, LeResche L, Dworkin SF, Von Korff M. Clinical diagnostic criteria for TMD: new classification permits multiple diagnosis. J Am Dent Assoc 1992;123:47–54. [22] Okeson JP. Long-term treatment of disc-interference disorders of the temporomandibular joint with anterior repositioning occlusal splints. J Prosthet Dent 1988;60:611–5. [23] Tasaki MM, Westesson PL. Temporomandibular joint: diagnostic accuracy with sagittal and coronal MR imaging. Radiology 1993;186:723–9. [24] Malzer U. Schlierenoptische Untersuchungen am isolierten Kniegelenksmodell, Marburg. Munich, Germany: Videoproduktion Fa. Siemens; 1992.