- Email: [email protected]
A linear-array freehand 3-D endoscopic ultrasound
Ultrasound in Med. & Biol., Vol. 29, No. 7, pp. 1001–1006, 2003 Copyright © 2003 World Federation for Ultrasound in Medicine & Biology Printed in the USA. All rights reserved 0301-5629/03/$–see front matter
doi:10.1016/S0301-5629(03)00888-3
● Original Contribution A LINEAR-ARRAY FREEHAND 3-D ENDOSCOPIC ULTRASOUND KAZUKI SUMIYAMA, NAOKI SUZUKI and HISAO TAJIRI Department of Surgery, Department of Endoscopy, and Institute for High Dimensional Medical Imaging, Jikei University School of Medicine, Tokyo, Japan (Received 30 September 2002; revised 17 February 2003; in final form 27 February 2003)
Abstract—Recognition of the clinical importance of linear-array endoscopic ultrasound (EUS) has increased. In this study, we developed a linear-array 3-D EUS, a miniature position sensor attached to the tip of the echoendoscope used in freehand scanning. To evaluate the geometrical accuracy of the 3-D reconstruction of the system, the diameter of a sphere-shaped phantom (38 mm) was determined by five examiners and five measurers. Measured size of the sphere was 39.03 ⴞ 1.29 mm, with variance between examiners and measurers, and interaction of examiners with measurers was not significant. In animal and clinical studies, the system facilitated anatomical interpretation of the EUS images, especially in the pancreatobiliary area and vascular images. We concluded that this system is both accurate and reproducible, and may resolve difficulties in linear-array EUS. (E-mail: [email protected]) © 2003 World Federation for Ultrasound in Medicine & Biology. Key Words: Linear-array EUS, Three-dimension, Doppler, Freehand scanning, Reproducibility, Imaging accuracy, Vascular image, Clinical application, Geometrical analysis.
images without geometrical distortion in the stomach (Sumiyama et al. 2002). To resolve these problems and maximize the performance of the 3-D EUS using the linear array echoendoscope, we developed a new 3-D endoscoic US system using a linear-array echoendoscope with a miniature electromagnetic position sensor attached to the tip of the scope, which can be used in freehand scanning in any position. In this study, we evaluated the imaging accuracy of the system, and will discuss the clinical applicability of the system using images from animal experiments and clinical cases.
INTRODUCTION Several studies in the field of endoscopic ultrasound (EUS) technology have reported advantages and future visions for three-dimensional (3-D) EUS (Chung et al. 2000; Hashimoto et al. 1989, 1991; Hu¨nerbein et al. 1997, 1999; Kallimanis et al. 1995; Liu et al. 2000; Molin et al. 1999; Nesje et al. 1999; Yoshimoto 1998; Yoshino et al. 1999). EUS-guided diagnostic and therapeutic interventional procedures with the guidance of a linear-array EUS have also attracted attention from the clinical field (Chang et al. 2000; Fockens et al. 1997; Giovannini et al. 1995; Hoffmann et al. 1997; Lahoti et al. 2000; Seifert et al. 2000; Wiersema and Wiersema 1996; Wiersema et al. 1997; Williams et al. 1999). However, most 3-D EUS studies have been performed using a catheter-type miniature probe system. We previously reported the benefits of a prototype 3-D EUS using a linear-array echoendoscope for 3-D guidance of interventional procedures, but the applicable scanning method of this system was systemically limited and, because the US probe was not positioned at the tip of the endoscope, it was difficult to obtain clinically sufficient
MATERIALS AND METHODS Volume data acquisition and construction 3-D US scanning was performed by freehand rotation of a conventional convex scanning echoendoscope (FG36UX, PENTAX, Tokyo, Japan equipped with 5and 7-MHz transducers or UC30P, OLYMPUS, Tokyo, Japan) and a display unit (EUB525 HITACHI, Tokyo, Japan or Envision Plus, Dornier Medtech, Wessling, Germany). Animal studies were performed by FG36UX and EUB525. Phantom and clinical studies were performed by UC30P and EnvisonPlus. 3-D image reconstructions were done using graphic workstations (Precision 420 DEL, Austin, Texas, U.S.A.) connected to the video output. Volume data were constructed by spatial
Address correspondence to: Kazuki Sumiyama, M.D., Department of Surgery and Institute for High Dimensional Medical Imaging, Jikei University School of Medicine, 3-25-8 Nishi Shinbashi, Minatoku, Tokyo 105-8461, Japan. E-mail: [email protected] 1001
1002
Ultrasound in Medicine and Biology
Volume 29, Number 7, 2003
Fig. 1. Tip of the echoendoscpe with a receiver for electromagnetic position sensor attached.
placement of a series of 2-D images corresponding to the position of each video frame, which is computed from positional data and directions of the US probe obtained from an electromagnetic tracking sensor (mini-Bird, Ascension Technology Co., Burlington, Vermont U.S.A.) attached to the tip of the echoendoscope (Fig. 1). Visualization of 3-D images was executed simultaneously with US scanning and volume data reconstruction; the images were revised 15 to 20 times per s throughout the scanning process. The 3-D image was presented as a polyhedron representing the boundary of the reconstructed volume. Information on the inner volume was projected on each face of the polyhedron using minimum and maximum intensity projection methods or resliced cut plane in surface-rendered and volume-rendered images. In addition, rotation of the polyhedron using an interactive interface allowed the cut plane to be observed, providing inner information from any orientation. The software used in this study allowed the simultaneous display of a 2-D cross-sectional image in addition to the 3-D image. Phantom study Imaging accuracy of this system depends on the quality of the original 2-D image and the accuracy of the sensor position. Geometrical accuracy of the perpendicular plane to the scanning plane of the 2-D image is peculiar to this 3-D system and reflects the accuracy of the position sensor. To evaluate accuracy, consistency and reproducibility of this system, an agar sphere (diameter 38 mm, containing aluminum powder to enhance echogenicity) was scanned and its diameter measured. Five examiners each scanned the sphere 5 times; therefore 25 volume data sets were acquired. The examiners measured the diameter of the sphere two-dimensionally after the volume scan. The diameter of the sphere of each acquired volume data set was measured by five blinded
Fig. 2. (a) Measurement of diameter of the phantom 3-D image of the sphere visualized by maximum intensity projection method, (b) surface-rendered cut plane of the volume, and (c) schematic representation of the measurement of the sphere diameter.
measurers who did not know the size or number of phantoms used. Sphere diameter measurements were determined by the surface-rendered cut plane, which was resliced perpendicularly to the scanning plane of the original 2-D images (Fig. 2). In total, 125 volume data sets and 5 2-D measurement data sets were expressed as the mean and SD. Intraexaminer variance, intrameasurer variance and interaction between observers and examiners were analyzed by two-way factorial analysis of variance (ANOVA). Animal studies and clinical cases Animal studies were performed using three swine of approximately 40 kg under general anesthesia. Images of 5 clinical cases were obtained; 3 with portal hypertension, 1 with a pancreatic intraductal papillary mucinous tumor and 1 with a pseudopancreatic cyst. All 3-D EUS were carried out after the instillation of a small volume (150 to 200 mL) of de-aerated water into the stomach, followed by the aspiration of as much air as possible from the stomach. The animal experiments complied with legal requirements and guidelines of our university. RESULTS In this system, the processing of the 3-D reconstructions from image data acquisition to 3-D image visualization was conveyed in real-time at approximately 10 to 15 times per s in color mode and 15 to 20 times per s in
Linear-array freehand 3-D EUS ● K. SUMIYAMA et al.
1003
Fig. 3. Images of the gallbladder of a swine (GB): GB was visualized just as cross-sectional image of the 2-D image. The cut plane obtained by this 3-D system presented a longitudinal image that visualizes both the GB and the cystic duct (CD) simultaneously.
Fig. 5. Left gastric vein (LGV) and abdominal aorta (Ao) (a) the left gastric vein bifurcating to several paraesophageal veins (PEV). (b) An image-extracted only Doppler signal allowed serial observation of enhanced vascular structure. (c) The abdominal aorta bifurcating to the celiac artery (CA) and the superior mesenteric vein (SMA).
the monochrome image. The system allowed arbitrary determination of the most appropriate volume for the target by the area scanned confirming the real-time 3-D image. This software also allowed choice of visualization methods; this study used adjusting opaque and contrast, volume rendering, surface rendering, maximum and minimum intensity projection methods.
sured sphere diameter was 38.54 ⫾ 2.13 mm. Variance of the sphere diameter measurement between five different examiners was not statistically significant (p ⫽ 0.4255 ⬎ 0.05) nor was variance between five different observers (p ⫽ 0.1496 ⬎ 0.05). In addition, variance in the interaction of examiners with observers was not significant (p ⫽ 0.8442 ⬎ 0.05). These results indicate that the 3-D images produced by this system were accurate and consistent.
Phantom study The mean and SD of the sphere diameter measured by 3-D image was 39.03 ⫾ 1.29 mm. The 2-D measured sphere diameter was 38.54 ⫾ 2.13 mm. The 2-D mea-
Fig. 4. 3-D Doppler images of the vascular structures in the animal study: (a) spleen; splenic artery (A) and vein (V) running parallel, (b) hepatic hilum, and (c) navigation mode of the system; a blue plane in the 3-D image indicating spatial position of (d) the 2-D image.
Images of animal studies and clinical cases In all animal studies and clinical cases, 3-D images of target organs (splenic vein and gallbladder in the
Fig. 6. Images of the pancreas: (left) pancreatic body and tail is clearly depicted as a low echoic region (P) with the splenic vein (SV) enhanced by Doppler. (right) 2-D image presenting the pancreas just as cross-section.
1004
Ultrasound in Medicine and Biology
Volume 29, Number 7, 2003
Fig. 7. Images of the pancreatic body and tail from the intraductal papillary mucinous tumor case. Dilated pancreatic duct (PD) of the (a) body and (b) tail; (c) papillary tumor visualized in the pancreatic duct of the body; (d) 2-D image of the dilated pancreatic duct.
Fig. 8. (a), (b) and (c) 3-D images and (d) 2-D image of the pseudopancreatic cyct: the 3-D image presenting good anatomical orientation of multiple cysts; small cysts (white arrows) surrounding a large cyst; (c) resliced image visualizing elevated lesion formed by the microcyst (black arrows).
animal studies, and the portal vein, splenic vein and pancreas in the clinical cases) were obtained. Swine gallbladders were examined in the animal study (Fig. 3). 2-D imaging only visualized coronal cross-sections, but a resliced 3-D image visualized a longitudinal section of the gallbladder, which represents the gallbladder and the cystic duct simultaneously. Figure 4 represents 3-D Doppler images of (Fig. 4a) the spleen and (Fig. 4b and c) the hepatic hilum of the swine. Complicated bifurcating vascular structures can be identified serially. A blue plane projected in the 3-D image indicates the spatial position of the 2-D images (Fig. 4c and d). In the case studies examined, images of vascular structures in the portal hypertension cases were studied (Fig. 5). The left gastric vein was bifurcating to several paraesophageal veins (Fig. 5a and b). On the bottom, the volume-rendered image clearly depicts the abdominal aorta (Ao) bifurcating to the celiac artery (CA) and superior mesenteric artery (SMA) without the use of Doppler (Fig. 5c). Figure 6 is transversely resliced images of the pancreas from a patient with portal hypertension. The pancreas was depicted as a low echoic region with the splenic vein running parallel to the pancreas, enhanced by Doppler signal. Figure 7 shows images of the pancreatic body and tail from a patient with intraductal papillary mucinous tumor. The dilated tortuous main pancreatic duct was evident and the rough ductal wall with the papillary tumor and plaques were recognized spatially. Figure 8 is comprised of images of multiple pseudopancreatic cysts. The image facilitated objective evaluation of the aspect of the cystic wall (Fig. 8a and b). Resliced 3-D images demonstrate that the in-
creased lesions of the large cystic wall consisted of microcysts (Fig. 8c). DISCUSSION Linear-array EUS can provide diagnosis, staging, decision of surgical application and interventional therapy all in one procedure. However, most endoscopists routinely use radial scan EUS or miniature probe systems because of difficulties in scanning procedure and image interpretation of linear-array EUS. The major disadvantage of linear-array EUS is the smaller sector-shaped longitudinal acoustic window, compared with the 360° perpendicular window of radial scan EUS, which is similar in orientation to computerized tomography (CT) or magnetic resonance imaging (MRI). To obtain quality images of the target organ, specialized anatomical knowledge and skill for the linear-array EUS are required (Chang and Erickson 1996). Interpretation of images should be performed intraoperatively or just after the examination, by an examiner with knowledge of how the images were obtained, including the motion of the endoscope during scanning and anatomical characteristics of the patient. It is difficult for doctors other than the examiner to anatomically understand a 2-D sonographic picture after the examination. The results of this study indicate that this 3-D system may resolve these limitations of the linear-array EUS. 2-D linear-array EUS visualizes the gallbladder, pancreas and splenic vein in cross-section only; therefore, the observer has to interpret a series of 2-D images to evaluate the whole organ. In this system, the examiner can obtain any arbitrary cut plane of the volume, which
Linear-array freehand 3-D EUS ● K. SUMIYAMA et al.
provides any orientation for visualization of the whole organ. The system allows evaluation of a large amount of information from many 2-D images efficiently and intuitively within a short time. Several studies reported the detectability of lineararray EUS for lesions in the pancreatobiliary area was almost equivalent to radial scan EUS (Gress et al. 1997; Hunt and Faigel 2002; Kohut et al. 2002; Lachter et al. 2000; Muguruma et al., 2001). In addition to histologic confirmation by needle biopsy, linear-array EUS has an advantage over radial scan EUS with the availability of Doppler. Doppler EUS is recognized as a less invasive alternative imaging modality for abdominal vasculature, rather than angiography in distinguishing malignancy in the pancreas and in the observation of the portal system for hemodynamic analysis of esophageal and gastric varices (Becker et al. 2001; Hino et al. 2002; Lee et al. 2002; Nishida et al. 2001; Wiersema et al. 1995). However, the pancreatobiliary area, including portal system, is imaged through the stomach, with scanning and anatomical interpretation being difficult because of the large capacity and anatomical variation of the stomach. The relationship between the US probe and surrounding organs is greatly changed in each case. Visualization of these areas in this 3-D system was relatively easy when compared to that of 2-D scanning. In this system, the examiner was able to identify landmarks and organs with little effort. Scanning of these areas was performed by rotating the scope several times after positioning the probe toward the posterior wall of the stomach, gradually advancing the scope from the gastroesophago junction to the antrum of the stomach. In addition, this system facilitated postoperative evaluation of images, including geometrical analysis. In the phantom study, 3-D imaging accuracy and reproducibility of the length measurement was within acceptable limits. In both the animal study and clinical cases, the system postoperatively provided anatomical orientation of the 3-D images by reslicing and rotating the obtained volume without limitation of observable directions by the stomach anatomy after appropriate image processing for target organs. 3-D sonographic images of abdominal vascular structure, with or without Doppler, represented an angiograph-like view that permitted objective recording of complicated vascular structures, and facilitated anatomic interpretation and geometrical analysis. Bifurcation of the celiac artery and the abdominal aorta obtained in this study is an important landmark for EUSguided celiac plexus neurolysis (Gress et al. 2001; Gunaratnam et al. 2000). 3-D vascular images with high reproducibility may lead to precise planning of EUSguided interventional therapy. This system had not yet been directly compared with angiography, but 3-D EUS is less invasive and may have a benefit when a periodical
1005
follow-up is required. In addition, 3-D EUS can visualize surrounding organs simultaneously. In summary, a new 3-D EUS using linear-array echoendoscope was demonstrated as accurate and represented a consistent method in this fundamental study. In the animal study and clinical cases examined, the system facilitated anatomical interpretation of sonographic images and reduced procedural difficulty of scanning. These results indicate that this new 3-D EUS system may expand the application of linear-array EUS. REFERENCES Becker D, Strobel D, Bernatik T, Hahn EG. Echo-enhanced color- and power-Doppler EUS for the discrimination between focal pancreatitis and pancreatic carcinoma. Gastrointest Endosc 2001;53(79): 784–789. Chang KJ, Erickson RA. A primer on linear array endosonographic anatomy. Gastrointest Endosc 1996;43(2):S43–S47. Chang KJ, Nguyen PT, Thompson JA, et al. Phase I clinical trial of allogeneic mixed lymphocyte culture (cytoimplant) delivered by endoscopic ultrasound-guided fine-needle injection in patients with advanced pancreatic carcinoma. Cancer 2000;88:1325–1335. Chung CY, McCrary WH, Dhaliwal S, et al. Three-dimensional esophageal varix model quantification of variceal volume by high-resolution endoluminal US. Gastrointest Endosc 2000;52(1):87–91. Fockens P, Johnson TG, van Dullemen HM, Huibregtse K, Tytgat GN. Endosonographic imaging of pancreatic pseudocysts before endoscopic transmural drainage. Gastrointest Endosc 1997;46:412–416. Giovannini M, Seitz JF, Monges G, Perrier H, Rabbia I. Fine-needle aspiration cytology guided by endoscopic ultrasonography: Results in 141 patients. Endoscopy 1995;27:171–177. Gress F, Savides T, Cummings O, et al. Radial scanning and linear array endosonography for staging pancreatic cancer: A prospective randomized comparison. Gastrointest Endosc 1997;45(2):138–142. Gress F, Schmitt C, Scherman S, et al. Endoscopic ultrasound-guided celiac plexus block for managing abdominal pain associated with chronic pancreatitis: A prospective single center experience. Am J Gastroenterol 2001;96:409–416. Gunaratnam NT, Wong GT, Wiersema MJ. EUS-guided celiac plexus block for the management of pancreatic pain. Gastrointest Endosc 2000;52(6):28–34. Hashimoto H, Mitsunaga A, Suzuki S, Kurokawa K, Obata H. Evaluation of endoscopic ultrasonography for gastric tumors and presentation of three-dimensional display of endoscopic ultrasonography. Surg Endosc 1989;3:173–181. Hashimoto H, Yokoyama S, Nakano K, et al. The capability of three dimensional display during endoscopic ultrasonography. Dig Endosc 1991;3:194–198. Hino S, Kakutani H, Ikeda K, et al. Hemodynamic assessment of the left gastric vein in patients with esophageal varices with color Doppler EUS: Factors affecting development of esophageal varices. Gastrointest Endosc 2002;55(4):512–517. Hoffman BJ, Knapple WL, Bhutani MS, Verne GN, Hawes RH. Treatment of achalasia by injection of botulinum toxin under endoscopic ultrasound guidance. Gastrointest Endosc 1997;45:77–79. Hu¨nerbein M, Ghadimi BM, Gretschel S, Schlag PM. Three-dimensional endoluminal ultrasound: A new method for the evaluation of gastrointestinal tumors. Abdom Imaging 1999;24:445–448. Hu¨nerbein M, Gretschel S, Ghadimi BM, Schlag PM. Three-dimensional endoscopic ultrasound of the esophagus, preliminary experience. Surg Endosc 1997;11(10):991–994. Hunt GC, Faigel DO. Assessment of EUS for diagnosing, staging, and determining resectability of pancreatic cancer: A review. Gastrointest Endosc 2002;55(2):232–237. Kallimanis G, Garra SB, Tio T, et al. The feasibility of three-dimensional endoscopic ultrasonography: A preliminary report. Gastrointest Endosc 1995;41(3):235–239.
1006
Ultrasound in Medicine and Biology
Kohut M, Nowakowska-Dutawa E, Marek T, Kaczor R, Nowak A. Accuracy of linear endoscopic ultrasonography in the evaluation of patients with supected common bile stones. Endoscopy 2002;34: 299–303. Lachter J, Rubin A, Shiller M, et al. Linear EUS for bile duct stones. Gastrointest Endosc 2000;51(1):51–54. Lahoti S, Catalano MF, Alcocer E, Hogan WJ, Geenen JE. Obiteration of esophageal varices using EUS-guided sclerotherapy with color Doppler. Gastrointest Endosc 2000;51:331–333. Lee YT, Chan FK, Ching JY, et al. Diagnosis of gastroesophageal varices and portal collateral venous abnomalities by endosonography in cirrhotic patients. Endoscopy 2002;34:391–398. Liu J, Miller LS, Chung CY, et al. Validation of volume measurements in esophageal pseudotumors using 3-D endoluminal ultrasound. Ultrasound Med Biol 2000;26(5):735–741. Molin S-O, Nesje LB, Gilja OH, et al. 3D-endosonography in gastroenterology: Methodology and clinical applications. Eur J Ultrasound 1999;10:171–177. Muguruma N, Okamura S, Ichikawa S, et al. Endoscopic sonography in the diagnosis of gallbladder wall lesions in patients with gallstones. J Clin Ultrasound 2001;7:395–400. Nesje SO, Molin SO, Gilja OH, Hausken T. Three-dimensional intraluminal sonography in the evaluation of gastrointestinal diseases. Abdom Imaging 1999;24(5):449–451. Nishida H, Giostra E, Spahar L, et al. Validation of color Doppler EUS for azygos blood flow measurement in patients with cirrhosis:
Volume 29, Number 7, 2003 Application to the acute hemodynamic effects of somatostatin, octreotide, or placebo. Gastrointest Endosc 2001;54(1):24–30. Seifert H, Dietrich C, Schmitt T, Caspary W, Wehrmann T. Endoscopic ultrasound-guided one-step transmural drainage of cystic abdominal lesions with a large-channel echo endoscope. Endoscopy 2000; 32:255–259. Sumiyama K, Suzuki N, Kakutani H, et al. A novel 3-dimensional EUS technique for real-time visualization of the volume data reconstruction. Gastrointest Endosc 2002;55(6):723–728. Wiersema MJ, Chak A, Kopecky KK, Wiersema LM. Duplex Doppler endosonography in the diagnosis of splenic vein, portal vein, and portosystemic shunt thrombosis. Gastrointest Endosc 1995;42(1): 19–26. Wiersema MJ, Vilmann P, Giovannini M, Chang KJ, Wiersema LM. Endosonography – guided fine-needle aspiration biopsy: Diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087–1095. Wiersema MJ, Wiersema LM. Endosonography-guided celiac plexus neurolysis. Gastrointest Endosc 1996;44:656–662. Williams DB, Sahai AV, Aabakken L, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: A large single centre experience. Gut 1999;44:720–726. Yoshimoto K. Clinical application of ultrasound 3D imaging system in lesions of the gastrointestinal tract. Endoscopy 1998;30:145–148. Yoshino J, Nakazawa S, Inui K, et al. Surface-rendering imaging of gastrointestinal lesions by three-dimensional endoscopic ultrasonography. Endoscopy 1999;31(7):541–545.
doi:10.1016/S0301-5629(03)00888-3
● Original Contribution A LINEAR-ARRAY FREEHAND 3-D ENDOSCOPIC ULTRASOUND KAZUKI SUMIYAMA, NAOKI SUZUKI and HISAO TAJIRI Department of Surgery, Department of Endoscopy, and Institute for High Dimensional Medical Imaging, Jikei University School of Medicine, Tokyo, Japan (Received 30 September 2002; revised 17 February 2003; in final form 27 February 2003)
Abstract—Recognition of the clinical importance of linear-array endoscopic ultrasound (EUS) has increased. In this study, we developed a linear-array 3-D EUS, a miniature position sensor attached to the tip of the echoendoscope used in freehand scanning. To evaluate the geometrical accuracy of the 3-D reconstruction of the system, the diameter of a sphere-shaped phantom (38 mm) was determined by five examiners and five measurers. Measured size of the sphere was 39.03 ⴞ 1.29 mm, with variance between examiners and measurers, and interaction of examiners with measurers was not significant. In animal and clinical studies, the system facilitated anatomical interpretation of the EUS images, especially in the pancreatobiliary area and vascular images. We concluded that this system is both accurate and reproducible, and may resolve difficulties in linear-array EUS. (E-mail: [email protected]) © 2003 World Federation for Ultrasound in Medicine & Biology. Key Words: Linear-array EUS, Three-dimension, Doppler, Freehand scanning, Reproducibility, Imaging accuracy, Vascular image, Clinical application, Geometrical analysis.
images without geometrical distortion in the stomach (Sumiyama et al. 2002). To resolve these problems and maximize the performance of the 3-D EUS using the linear array echoendoscope, we developed a new 3-D endoscoic US system using a linear-array echoendoscope with a miniature electromagnetic position sensor attached to the tip of the scope, which can be used in freehand scanning in any position. In this study, we evaluated the imaging accuracy of the system, and will discuss the clinical applicability of the system using images from animal experiments and clinical cases.
INTRODUCTION Several studies in the field of endoscopic ultrasound (EUS) technology have reported advantages and future visions for three-dimensional (3-D) EUS (Chung et al. 2000; Hashimoto et al. 1989, 1991; Hu¨nerbein et al. 1997, 1999; Kallimanis et al. 1995; Liu et al. 2000; Molin et al. 1999; Nesje et al. 1999; Yoshimoto 1998; Yoshino et al. 1999). EUS-guided diagnostic and therapeutic interventional procedures with the guidance of a linear-array EUS have also attracted attention from the clinical field (Chang et al. 2000; Fockens et al. 1997; Giovannini et al. 1995; Hoffmann et al. 1997; Lahoti et al. 2000; Seifert et al. 2000; Wiersema and Wiersema 1996; Wiersema et al. 1997; Williams et al. 1999). However, most 3-D EUS studies have been performed using a catheter-type miniature probe system. We previously reported the benefits of a prototype 3-D EUS using a linear-array echoendoscope for 3-D guidance of interventional procedures, but the applicable scanning method of this system was systemically limited and, because the US probe was not positioned at the tip of the endoscope, it was difficult to obtain clinically sufficient
MATERIALS AND METHODS Volume data acquisition and construction 3-D US scanning was performed by freehand rotation of a conventional convex scanning echoendoscope (FG36UX, PENTAX, Tokyo, Japan equipped with 5and 7-MHz transducers or UC30P, OLYMPUS, Tokyo, Japan) and a display unit (EUB525 HITACHI, Tokyo, Japan or Envision Plus, Dornier Medtech, Wessling, Germany). Animal studies were performed by FG36UX and EUB525. Phantom and clinical studies were performed by UC30P and EnvisonPlus. 3-D image reconstructions were done using graphic workstations (Precision 420 DEL, Austin, Texas, U.S.A.) connected to the video output. Volume data were constructed by spatial
Address correspondence to: Kazuki Sumiyama, M.D., Department of Surgery and Institute for High Dimensional Medical Imaging, Jikei University School of Medicine, 3-25-8 Nishi Shinbashi, Minatoku, Tokyo 105-8461, Japan. E-mail: [email protected] 1001
1002
Ultrasound in Medicine and Biology
Volume 29, Number 7, 2003
Fig. 1. Tip of the echoendoscpe with a receiver for electromagnetic position sensor attached.
placement of a series of 2-D images corresponding to the position of each video frame, which is computed from positional data and directions of the US probe obtained from an electromagnetic tracking sensor (mini-Bird, Ascension Technology Co., Burlington, Vermont U.S.A.) attached to the tip of the echoendoscope (Fig. 1). Visualization of 3-D images was executed simultaneously with US scanning and volume data reconstruction; the images were revised 15 to 20 times per s throughout the scanning process. The 3-D image was presented as a polyhedron representing the boundary of the reconstructed volume. Information on the inner volume was projected on each face of the polyhedron using minimum and maximum intensity projection methods or resliced cut plane in surface-rendered and volume-rendered images. In addition, rotation of the polyhedron using an interactive interface allowed the cut plane to be observed, providing inner information from any orientation. The software used in this study allowed the simultaneous display of a 2-D cross-sectional image in addition to the 3-D image. Phantom study Imaging accuracy of this system depends on the quality of the original 2-D image and the accuracy of the sensor position. Geometrical accuracy of the perpendicular plane to the scanning plane of the 2-D image is peculiar to this 3-D system and reflects the accuracy of the position sensor. To evaluate accuracy, consistency and reproducibility of this system, an agar sphere (diameter 38 mm, containing aluminum powder to enhance echogenicity) was scanned and its diameter measured. Five examiners each scanned the sphere 5 times; therefore 25 volume data sets were acquired. The examiners measured the diameter of the sphere two-dimensionally after the volume scan. The diameter of the sphere of each acquired volume data set was measured by five blinded
Fig. 2. (a) Measurement of diameter of the phantom 3-D image of the sphere visualized by maximum intensity projection method, (b) surface-rendered cut plane of the volume, and (c) schematic representation of the measurement of the sphere diameter.
measurers who did not know the size or number of phantoms used. Sphere diameter measurements were determined by the surface-rendered cut plane, which was resliced perpendicularly to the scanning plane of the original 2-D images (Fig. 2). In total, 125 volume data sets and 5 2-D measurement data sets were expressed as the mean and SD. Intraexaminer variance, intrameasurer variance and interaction between observers and examiners were analyzed by two-way factorial analysis of variance (ANOVA). Animal studies and clinical cases Animal studies were performed using three swine of approximately 40 kg under general anesthesia. Images of 5 clinical cases were obtained; 3 with portal hypertension, 1 with a pancreatic intraductal papillary mucinous tumor and 1 with a pseudopancreatic cyst. All 3-D EUS were carried out after the instillation of a small volume (150 to 200 mL) of de-aerated water into the stomach, followed by the aspiration of as much air as possible from the stomach. The animal experiments complied with legal requirements and guidelines of our university. RESULTS In this system, the processing of the 3-D reconstructions from image data acquisition to 3-D image visualization was conveyed in real-time at approximately 10 to 15 times per s in color mode and 15 to 20 times per s in
Linear-array freehand 3-D EUS ● K. SUMIYAMA et al.
1003
Fig. 3. Images of the gallbladder of a swine (GB): GB was visualized just as cross-sectional image of the 2-D image. The cut plane obtained by this 3-D system presented a longitudinal image that visualizes both the GB and the cystic duct (CD) simultaneously.
Fig. 5. Left gastric vein (LGV) and abdominal aorta (Ao) (a) the left gastric vein bifurcating to several paraesophageal veins (PEV). (b) An image-extracted only Doppler signal allowed serial observation of enhanced vascular structure. (c) The abdominal aorta bifurcating to the celiac artery (CA) and the superior mesenteric vein (SMA).
the monochrome image. The system allowed arbitrary determination of the most appropriate volume for the target by the area scanned confirming the real-time 3-D image. This software also allowed choice of visualization methods; this study used adjusting opaque and contrast, volume rendering, surface rendering, maximum and minimum intensity projection methods.
sured sphere diameter was 38.54 ⫾ 2.13 mm. Variance of the sphere diameter measurement between five different examiners was not statistically significant (p ⫽ 0.4255 ⬎ 0.05) nor was variance between five different observers (p ⫽ 0.1496 ⬎ 0.05). In addition, variance in the interaction of examiners with observers was not significant (p ⫽ 0.8442 ⬎ 0.05). These results indicate that the 3-D images produced by this system were accurate and consistent.
Phantom study The mean and SD of the sphere diameter measured by 3-D image was 39.03 ⫾ 1.29 mm. The 2-D measured sphere diameter was 38.54 ⫾ 2.13 mm. The 2-D mea-
Fig. 4. 3-D Doppler images of the vascular structures in the animal study: (a) spleen; splenic artery (A) and vein (V) running parallel, (b) hepatic hilum, and (c) navigation mode of the system; a blue plane in the 3-D image indicating spatial position of (d) the 2-D image.
Images of animal studies and clinical cases In all animal studies and clinical cases, 3-D images of target organs (splenic vein and gallbladder in the
Fig. 6. Images of the pancreas: (left) pancreatic body and tail is clearly depicted as a low echoic region (P) with the splenic vein (SV) enhanced by Doppler. (right) 2-D image presenting the pancreas just as cross-section.
1004
Ultrasound in Medicine and Biology
Volume 29, Number 7, 2003
Fig. 7. Images of the pancreatic body and tail from the intraductal papillary mucinous tumor case. Dilated pancreatic duct (PD) of the (a) body and (b) tail; (c) papillary tumor visualized in the pancreatic duct of the body; (d) 2-D image of the dilated pancreatic duct.
Fig. 8. (a), (b) and (c) 3-D images and (d) 2-D image of the pseudopancreatic cyct: the 3-D image presenting good anatomical orientation of multiple cysts; small cysts (white arrows) surrounding a large cyst; (c) resliced image visualizing elevated lesion formed by the microcyst (black arrows).
animal studies, and the portal vein, splenic vein and pancreas in the clinical cases) were obtained. Swine gallbladders were examined in the animal study (Fig. 3). 2-D imaging only visualized coronal cross-sections, but a resliced 3-D image visualized a longitudinal section of the gallbladder, which represents the gallbladder and the cystic duct simultaneously. Figure 4 represents 3-D Doppler images of (Fig. 4a) the spleen and (Fig. 4b and c) the hepatic hilum of the swine. Complicated bifurcating vascular structures can be identified serially. A blue plane projected in the 3-D image indicates the spatial position of the 2-D images (Fig. 4c and d). In the case studies examined, images of vascular structures in the portal hypertension cases were studied (Fig. 5). The left gastric vein was bifurcating to several paraesophageal veins (Fig. 5a and b). On the bottom, the volume-rendered image clearly depicts the abdominal aorta (Ao) bifurcating to the celiac artery (CA) and superior mesenteric artery (SMA) without the use of Doppler (Fig. 5c). Figure 6 is transversely resliced images of the pancreas from a patient with portal hypertension. The pancreas was depicted as a low echoic region with the splenic vein running parallel to the pancreas, enhanced by Doppler signal. Figure 7 shows images of the pancreatic body and tail from a patient with intraductal papillary mucinous tumor. The dilated tortuous main pancreatic duct was evident and the rough ductal wall with the papillary tumor and plaques were recognized spatially. Figure 8 is comprised of images of multiple pseudopancreatic cysts. The image facilitated objective evaluation of the aspect of the cystic wall (Fig. 8a and b). Resliced 3-D images demonstrate that the in-
creased lesions of the large cystic wall consisted of microcysts (Fig. 8c). DISCUSSION Linear-array EUS can provide diagnosis, staging, decision of surgical application and interventional therapy all in one procedure. However, most endoscopists routinely use radial scan EUS or miniature probe systems because of difficulties in scanning procedure and image interpretation of linear-array EUS. The major disadvantage of linear-array EUS is the smaller sector-shaped longitudinal acoustic window, compared with the 360° perpendicular window of radial scan EUS, which is similar in orientation to computerized tomography (CT) or magnetic resonance imaging (MRI). To obtain quality images of the target organ, specialized anatomical knowledge and skill for the linear-array EUS are required (Chang and Erickson 1996). Interpretation of images should be performed intraoperatively or just after the examination, by an examiner with knowledge of how the images were obtained, including the motion of the endoscope during scanning and anatomical characteristics of the patient. It is difficult for doctors other than the examiner to anatomically understand a 2-D sonographic picture after the examination. The results of this study indicate that this 3-D system may resolve these limitations of the linear-array EUS. 2-D linear-array EUS visualizes the gallbladder, pancreas and splenic vein in cross-section only; therefore, the observer has to interpret a series of 2-D images to evaluate the whole organ. In this system, the examiner can obtain any arbitrary cut plane of the volume, which
Linear-array freehand 3-D EUS ● K. SUMIYAMA et al.
provides any orientation for visualization of the whole organ. The system allows evaluation of a large amount of information from many 2-D images efficiently and intuitively within a short time. Several studies reported the detectability of lineararray EUS for lesions in the pancreatobiliary area was almost equivalent to radial scan EUS (Gress et al. 1997; Hunt and Faigel 2002; Kohut et al. 2002; Lachter et al. 2000; Muguruma et al., 2001). In addition to histologic confirmation by needle biopsy, linear-array EUS has an advantage over radial scan EUS with the availability of Doppler. Doppler EUS is recognized as a less invasive alternative imaging modality for abdominal vasculature, rather than angiography in distinguishing malignancy in the pancreas and in the observation of the portal system for hemodynamic analysis of esophageal and gastric varices (Becker et al. 2001; Hino et al. 2002; Lee et al. 2002; Nishida et al. 2001; Wiersema et al. 1995). However, the pancreatobiliary area, including portal system, is imaged through the stomach, with scanning and anatomical interpretation being difficult because of the large capacity and anatomical variation of the stomach. The relationship between the US probe and surrounding organs is greatly changed in each case. Visualization of these areas in this 3-D system was relatively easy when compared to that of 2-D scanning. In this system, the examiner was able to identify landmarks and organs with little effort. Scanning of these areas was performed by rotating the scope several times after positioning the probe toward the posterior wall of the stomach, gradually advancing the scope from the gastroesophago junction to the antrum of the stomach. In addition, this system facilitated postoperative evaluation of images, including geometrical analysis. In the phantom study, 3-D imaging accuracy and reproducibility of the length measurement was within acceptable limits. In both the animal study and clinical cases, the system postoperatively provided anatomical orientation of the 3-D images by reslicing and rotating the obtained volume without limitation of observable directions by the stomach anatomy after appropriate image processing for target organs. 3-D sonographic images of abdominal vascular structure, with or without Doppler, represented an angiograph-like view that permitted objective recording of complicated vascular structures, and facilitated anatomic interpretation and geometrical analysis. Bifurcation of the celiac artery and the abdominal aorta obtained in this study is an important landmark for EUSguided celiac plexus neurolysis (Gress et al. 2001; Gunaratnam et al. 2000). 3-D vascular images with high reproducibility may lead to precise planning of EUSguided interventional therapy. This system had not yet been directly compared with angiography, but 3-D EUS is less invasive and may have a benefit when a periodical
1005
follow-up is required. In addition, 3-D EUS can visualize surrounding organs simultaneously. In summary, a new 3-D EUS using linear-array echoendoscope was demonstrated as accurate and represented a consistent method in this fundamental study. In the animal study and clinical cases examined, the system facilitated anatomical interpretation of sonographic images and reduced procedural difficulty of scanning. These results indicate that this new 3-D EUS system may expand the application of linear-array EUS. REFERENCES Becker D, Strobel D, Bernatik T, Hahn EG. Echo-enhanced color- and power-Doppler EUS for the discrimination between focal pancreatitis and pancreatic carcinoma. Gastrointest Endosc 2001;53(79): 784–789. Chang KJ, Erickson RA. A primer on linear array endosonographic anatomy. Gastrointest Endosc 1996;43(2):S43–S47. Chang KJ, Nguyen PT, Thompson JA, et al. Phase I clinical trial of allogeneic mixed lymphocyte culture (cytoimplant) delivered by endoscopic ultrasound-guided fine-needle injection in patients with advanced pancreatic carcinoma. Cancer 2000;88:1325–1335. Chung CY, McCrary WH, Dhaliwal S, et al. Three-dimensional esophageal varix model quantification of variceal volume by high-resolution endoluminal US. Gastrointest Endosc 2000;52(1):87–91. Fockens P, Johnson TG, van Dullemen HM, Huibregtse K, Tytgat GN. Endosonographic imaging of pancreatic pseudocysts before endoscopic transmural drainage. Gastrointest Endosc 1997;46:412–416. Giovannini M, Seitz JF, Monges G, Perrier H, Rabbia I. Fine-needle aspiration cytology guided by endoscopic ultrasonography: Results in 141 patients. Endoscopy 1995;27:171–177. Gress F, Savides T, Cummings O, et al. Radial scanning and linear array endosonography for staging pancreatic cancer: A prospective randomized comparison. Gastrointest Endosc 1997;45(2):138–142. Gress F, Schmitt C, Scherman S, et al. Endoscopic ultrasound-guided celiac plexus block for managing abdominal pain associated with chronic pancreatitis: A prospective single center experience. Am J Gastroenterol 2001;96:409–416. Gunaratnam NT, Wong GT, Wiersema MJ. EUS-guided celiac plexus block for the management of pancreatic pain. Gastrointest Endosc 2000;52(6):28–34. Hashimoto H, Mitsunaga A, Suzuki S, Kurokawa K, Obata H. Evaluation of endoscopic ultrasonography for gastric tumors and presentation of three-dimensional display of endoscopic ultrasonography. Surg Endosc 1989;3:173–181. Hashimoto H, Yokoyama S, Nakano K, et al. The capability of three dimensional display during endoscopic ultrasonography. Dig Endosc 1991;3:194–198. Hino S, Kakutani H, Ikeda K, et al. Hemodynamic assessment of the left gastric vein in patients with esophageal varices with color Doppler EUS: Factors affecting development of esophageal varices. Gastrointest Endosc 2002;55(4):512–517. Hoffman BJ, Knapple WL, Bhutani MS, Verne GN, Hawes RH. Treatment of achalasia by injection of botulinum toxin under endoscopic ultrasound guidance. Gastrointest Endosc 1997;45:77–79. Hu¨nerbein M, Ghadimi BM, Gretschel S, Schlag PM. Three-dimensional endoluminal ultrasound: A new method for the evaluation of gastrointestinal tumors. Abdom Imaging 1999;24:445–448. Hu¨nerbein M, Gretschel S, Ghadimi BM, Schlag PM. Three-dimensional endoscopic ultrasound of the esophagus, preliminary experience. Surg Endosc 1997;11(10):991–994. Hunt GC, Faigel DO. Assessment of EUS for diagnosing, staging, and determining resectability of pancreatic cancer: A review. Gastrointest Endosc 2002;55(2):232–237. Kallimanis G, Garra SB, Tio T, et al. The feasibility of three-dimensional endoscopic ultrasonography: A preliminary report. Gastrointest Endosc 1995;41(3):235–239.
1006
Ultrasound in Medicine and Biology
Kohut M, Nowakowska-Dutawa E, Marek T, Kaczor R, Nowak A. Accuracy of linear endoscopic ultrasonography in the evaluation of patients with supected common bile stones. Endoscopy 2002;34: 299–303. Lachter J, Rubin A, Shiller M, et al. Linear EUS for bile duct stones. Gastrointest Endosc 2000;51(1):51–54. Lahoti S, Catalano MF, Alcocer E, Hogan WJ, Geenen JE. Obiteration of esophageal varices using EUS-guided sclerotherapy with color Doppler. Gastrointest Endosc 2000;51:331–333. Lee YT, Chan FK, Ching JY, et al. Diagnosis of gastroesophageal varices and portal collateral venous abnomalities by endosonography in cirrhotic patients. Endoscopy 2002;34:391–398. Liu J, Miller LS, Chung CY, et al. Validation of volume measurements in esophageal pseudotumors using 3-D endoluminal ultrasound. Ultrasound Med Biol 2000;26(5):735–741. Molin S-O, Nesje LB, Gilja OH, et al. 3D-endosonography in gastroenterology: Methodology and clinical applications. Eur J Ultrasound 1999;10:171–177. Muguruma N, Okamura S, Ichikawa S, et al. Endoscopic sonography in the diagnosis of gallbladder wall lesions in patients with gallstones. J Clin Ultrasound 2001;7:395–400. Nesje SO, Molin SO, Gilja OH, Hausken T. Three-dimensional intraluminal sonography in the evaluation of gastrointestinal diseases. Abdom Imaging 1999;24(5):449–451. Nishida H, Giostra E, Spahar L, et al. Validation of color Doppler EUS for azygos blood flow measurement in patients with cirrhosis:
Volume 29, Number 7, 2003 Application to the acute hemodynamic effects of somatostatin, octreotide, or placebo. Gastrointest Endosc 2001;54(1):24–30. Seifert H, Dietrich C, Schmitt T, Caspary W, Wehrmann T. Endoscopic ultrasound-guided one-step transmural drainage of cystic abdominal lesions with a large-channel echo endoscope. Endoscopy 2000; 32:255–259. Sumiyama K, Suzuki N, Kakutani H, et al. A novel 3-dimensional EUS technique for real-time visualization of the volume data reconstruction. Gastrointest Endosc 2002;55(6):723–728. Wiersema MJ, Chak A, Kopecky KK, Wiersema LM. Duplex Doppler endosonography in the diagnosis of splenic vein, portal vein, and portosystemic shunt thrombosis. Gastrointest Endosc 1995;42(1): 19–26. Wiersema MJ, Vilmann P, Giovannini M, Chang KJ, Wiersema LM. Endosonography – guided fine-needle aspiration biopsy: Diagnostic accuracy and complication assessment. Gastroenterology 1997;112:1087–1095. Wiersema MJ, Wiersema LM. Endosonography-guided celiac plexus neurolysis. Gastrointest Endosc 1996;44:656–662. Williams DB, Sahai AV, Aabakken L, et al. Endoscopic ultrasound guided fine needle aspiration biopsy: A large single centre experience. Gut 1999;44:720–726. Yoshimoto K. Clinical application of ultrasound 3D imaging system in lesions of the gastrointestinal tract. Endoscopy 1998;30:145–148. Yoshino J, Nakazawa S, Inui K, et al. Surface-rendering imaging of gastrointestinal lesions by three-dimensional endoscopic ultrasonography. Endoscopy 1999;31(7):541–545.