Reliability of MR Imaging Based Virtual Cystoscopy in the Diagnosis of Cancer of the Urinary Bladder

Transcription

1 Markus Lämmle 1,2 mbros eer 1 Marcus Settles 1 Christian Hannig 1 Hartwig Schwaibold 3 Carsten Drews 3 Received June 4, 2001; accepted after revision December 7, Department of Radiology, Klinikum rechts der Isar der Technischen Universität München, Ismaninger Str. 22, D Munich, Germany. 2 Present address: Division of Diagnostic Radiology, Mallinckrodt Institute of Radiology, Washington University School of Medicine, 510 S. Kingshighway lvd., St. Louis, MO ddress correspondence to M. Lämmle. 3 Department of Urology, Klinikum rechts der Isar der Technischen Universität München, D Munich, Germany. JR 2002;178: X/02/ merican Roentgen Ray Society Reliability of MR Imaging ased Virtual Cystoscopy in the Diagnosis of Cancer of the Urinary ladder OJECTIVE. Our purpose was to evaluate MR imaging based virtual endoscopy in patients with urinary bladder cancer compared with conventional cystoscopy as the gold standard. SUJECTS ND METHODS. Twenty-five patients with urinary bladder cancer diagnosed on conventional cystoscopy underwent MR imaging of the pelvis. Patients were examined without external bladder filling or administration of IV contrast medium. No medications were administered. The data obtained by MR imaging were reconstructed for virtual endoscopy on a workstation. The locations and sizes of tumors were individually determined and compared with results of conventional cystoscopy. RESULTS. Twenty-four patients were evaluated; one patient s examination was excluded from analysis because of metallic artifacts. Seventeen patients were diagnosed with a single bladder tumor. Five patients had two tumors each, and two patients had three tumors. Tumor diameter ranged from 0.4 to 6.4 cm. Thirty (90.9%) of 33 tumors detected on cystoscopy were visualized with virtual endoscopy. The detection rate for 23 tumors of 1 cm or greater was 100%. Difficult conditions for conventional cystoscopy, including hematuria, anterior wall involvement, and urethral strictures, had no deleterious impact on virtual cystoscopy. Difficulties in detection on virtual endoscopy were associated with flat bladder tumors with minimal surface elevation. CONCLUSION. The results of this study suggest a high reliability in the diagnosis of urinary bladder cancer by MR imaging based virtual cystoscopy a noninvasive method, independent of medication or contrast enhancement, that may be of value for screening, primary diagnosis, and surveillance. Virtual MR cystoscopy may be indicated when conventional cystoscopy cannot be performed or is ineffective. V irtual endoscopy is a recently developed imaging modality that requires further investigation with respect to its clinical indications and use. ssessment of this modality s clinical value and reliability is also needed [1]. Preliminary studies have already proven the feasibility of virtual endoscopy of the urinary tract [2 7]. Currently, most authors who discuss virtual cystoscopy of the bladder refer to CT-based virtual endoscopy, which requires air insufflation or administration of IV contrast medium [2 4, 6, 7]. Few reports are found in the literature regarding MR imaging based virtual endoscopy of the urinary bladder. Urinary bladder cancer is a frequently occuring malignancy with a high rate of recurrence and multifocal manifestations, and therefore it requires reliable diagnostic techniques [8 11]. The purpose of this study was to evaluate MR imaging based virtual endoscopy in patients with urinary bladder cancer compared with the gold standard that is, conventional cystoscopy and to determine the modality s detection rate and positive predictive value. In addition, we wanted to discover the advantages and disadvantages of the new method and to determine whether suspected urinary bladder cancer is a good indication for virtual endoscopy. Subjects and Methods Patients We studied twenty-five patients (20 men, five women) who were consecutively hospitalized under the care of the Department of Urology because of urinary bladder tumors diagnosed with conventional cystoscopy. The patients ranged in age from 45 to 89 years (mean age, 67.2 years; median age, 71 years). JR:178, June

2 Lämmle et al. ll patients underwent MR imaging of the pelvis between October 1999 and June 2000 after giving informed consent. The interval between conventional cystoscopy and virtual endoscopy was 6 29 days (mean time elapsed, 16.3 days; median time, 15 days). The patients clinical histories varied. Twentyone patients presented with painless hematuria or with dysuria, and a primary tumor was found in each patient. Four patients with a history of transurethral resection had recurrent tumor disease that was detected on follow-up cystoscopy. Of the 25 patients studied, one patient s examination was not acceptable for reconstruction because of strong artifacts from a hip prosthesis. That patient was excluded from analysis. The other 24 patients had examinations of satisfactory image quality and were evaluated. Cystoscopy Details regarding estimated size, location, and number of tumors found on cystoscopy were obtained from the cystoscopic reports provided by the Department of Urology. Techniques used for conventional cystoscopy included, in 21 patients, rigid cystoscopy with a field of view of and, in three patients, flexible fiberoptic cystoscopy with a 90 field of view. Number, location, and size of the tumors were individually determined and noted after cystoscopy for later comparison with the results of MR imaging based virtual endoscopy. MR Technique Fig. 1. Histogram depicts number of tumors found per patient. Left columns (gray) represent cystoscopy, and right columns (white) represent virtual endoscopy. Findings on conventional cystoscopy revealed that 17 patients had single bladder tumor, five had two tumors, and two had three tumors. Fig. 2. Histogram depicts number of tumors found on cystoscopy (gray columns) and virtual endoscopy (white columns) and their diameters. ll known tumors of 1 cm and larger were seen on virtual endoscopy. ll three tumors not identified on virtual endoscopy were smaller than 1 cm in diameter. Two tumors not discovered on cystoscopy but detected by virtual endoscopy are included in corresponding columns. MR imaging was performed without external bladder filling or administration of IV contrast medium. No medications were given. The patients were instructed to start drinking water 1 hr before the MR imaging study and to present with a full bladder. ll examinations were performed on a 1.5-T scanner (Gyroscan CS-NT; Philips Medical Systems, est, The Netherlands) using a five-element phased array coil. We performed a T2-weighted three-dimensional turbo spin-echo sequence for data acquisition with the following parameters: TR/TE, 2911/650; echo train, 128 (effective train length, 100); matrix, ; reconstruction matrix, ; flip angle, 90 ; field of view, 350 mm; excitations, 2; signal averages, 2; voxel size, 1.4 (x) 1.9 (y) 2.0 (z) mm 3 ; number of slices, 100; and interpolated slice thickness, 1.0 mm. The acquisition time of the sequence was 4 min 8 sec JR:178, June 2002

3 Virtual Cystoscopy of Cancer of the Urinary ladder Image nalysis coronal scout image was obtained to localize the urinary bladder and to visualize filling status and size of the bladder of patients undergoing MR imaging. Coronal imaging was performed as a standard scanning plane. Data were reconstructed on a postprocessing workstation (Easy Vision; Philips Medical Systems). The hardware consisted of a workstation (Sun Microsystems, Palo lto, C). The data were transferred via intranet using DICOM (Digital Imaging and COmmunications in Medicine) standards. Surface rendering was used to generate a standardized three-dimensional interactive intraluminal image of the urinary bladder. The region of interest was chosen from the data, and surrounding structures were segmented out. For reconstruction of a urinary bladder that presented with high signal intensity in comparison with the surrounding tissue, an upper threshold was selected on a scale with a range from 1 to 1000 units. The upper threshold was set at approximately 450 ( ) units for three-dimensional reconstruction. Using multiplanar reformation from source images, a central observation point was defined in the middle of the lumen of the bladder. The obtained reconstruction was visualized in two standard-perspective projections, rotated horizontally and vertically. In this manner, reconstruction was standardized to view internal bladder surfaces in all areas. First we created an overall view by spinning from the central observation point. field of view of 120 was applied to the vertical and horizontal planes, so that we could visualize the bladder neck; the vesical trigone with the ureteric orifices and the internal urethral orifice; the right, posterior, left, and anterior bladder wall; and the vesical dome. Then we used a field of view of 90 and the zoom function to scan the mucosa for small tumors. This search pattern was followed in all patients. Virtual endoscopy was performed and analyzed by one radiologist who was unaware of the other findings. The number, location, and size of the tumors were individually determined and noted after standardized virtual endoscopy for later comparison with the results of conventional cystoscopy. Statistical nalysis Findings from both methods conventional cystoscopy and virtual endoscopy were analyzed to determine the detection rate and positive predictive value of MR imaging based virtual endoscopy in the diagnosis of urinary bladder tumors. Results The patients examined for this study had a distribution of age and sex typical for patients with urinary bladder cancer and thus were representative of the population with this disease [10, 12]. MR imaging scanning was well tolerated by all patients, and no complications occurred. Twenty-four data sets of MR imaging examinations were used to perform virtual endoscopic reconstruction. Evaluation of the results of tumor detection by cystoscopy and MR imaging based virtual endoscopy is shown in Figures 1 and 2. On conventional cystoscopy, 33 tumors were found in 24 patients. mong those, 17 patients (70.8%) were diagnosed with a single bladder tumor. Five patients (20.8%) had two tumor locations, and two patients (8.3%) had three tumors. One or more tumors were found on virtual endoscopy in all patients, and all known single tumors were identified. In two patients with two known tumors each, only one tumor was detected; in one patient with three tumor locations, only two were identified by virtual endoscopy. Thus, in total, 30 of 33 known tumors were identified by virtual MR cystoscopy (Fig. 1). Two tumors that were not identified on initial cystoscopy were detected by virtual endoscopy. In one patient, the tumor was very small (0.5 cm in maximum diameter) and was located on the anterior bladder wall. dditionally, cystoscopy was complicated by moderate hematuria in this patient. In the second patient, a 2.1-cm tumor located at the dome of the bladder was not seen on initial cystoscopy because vision was restricted by Fig year-old man with primary urinary bladder cancer., nteroposterior virtual endoscopic image depicts posterior bladder wall and vesical trigone. The trigone is demarcated by two ureteric orifices (arrowheads) and internal urethral orifice (arrow). Marked prostatic protrusion visible in foreground is consistent with prostatic hypertrophy., Virtual endoscopic intravesical image of anterior bladder wall (posteroanterior view: field of view, 90 ) depicts two papillary tumors (arrowheads) 0.8 and 1.4 cm in diameter at left anterior bladder wall and right ventral dome. Fig year-old woman with recurrent urinary bladder cancer. and, Virtual endoscopic images of four bladder wall diverticula before zooming () and after zooming () show tumor at neck of one diverticulum (arrow). Tumor diameter is 0.4 cm one of smallest found in our study. JR:178, June

4 Lämmle et al. the presence of marked hematuria. These findings were confirmed by repeated conventional cystoscopy after coagulation of bleeding tumors and bladder washings at the time of transurethral resection. Tumor size ranged from 0.4 to 6.4 cm in diameter (mean size, 2.1 cm). Virtual endoscopy depicted all 23 known tumors with a diameter of 1 cm or greater and seven of 10 known tumors smaller than 1 cm (Fig. 2). Thus the detection rate for those small tumors was 70%. Detection rate for tumors larger than 1 cm was 100%. Thirty of 33 tumors detected on conventional cystoscopy were identified with virtual endoscopy, corresponding to an overall detection rate of 90.9%. Virtual endoscopy revealed two tumors that were not seen on cystoscopy, C Fig year-old man with solitary urinary bladder tumor. and, Conventional cystoscopic image () and virtual endoscopic image (: field of view, 30 ) depict tumor 2.3 cm in diameter. C and D, Coronal T2-weighted MR image (C) and virtual endoscopic image with overall view (D, 120 ) show tumor s location next to vesical trigone ventrolateral to left ureteric orifice. Internal urethral orifice can be identified in lower mid portion of D. D which represented our gold standard. Therefore, per definition, the positive predictive value of virtual endoscopy is 93.8%. However, the positive predictive value is 100% if we consider the fact that the two tumors revealed on virtual endoscopy were confirmed later intraoperatively, and, therefore, no false-positive findings resulted from virtual endoscopy. Difficult conditions for conventional cystoscopy, including hematuria, anterior wall involvement, and urethral strictures, had no impact on virtual endoscopy. Standardized reconstruction of data sets guaranteed a complete examination of all areas of the mucosa of the bladder. Even very small tumors were easily identified in virtual endoscopic reconstruction using the zoom function in the software tools. Difficulties in detection were associated with small, flat bladder tumors that caused only a slight elevation of surface structure. Three masses smaller than 1 cm in diameter and without significant contour elevation from the bladder wall into the lumen were not found on virtual endoscopy. The three undetected tumors were histologically superficial tumors. Two of the tumors were staged pta, G1 and one was staged pt1(is), G3 on histology after transurethral resection. Discussion Most tumors of the urinary bladder are malignant. Carcinoma of the urinary bladder represents 4.4% of all malignancies and is associated with a relatively high mortality (2.2% of all cancer deaths) [10, 12]. Thus, cystoscopy is required on a general basis for early detection of bladder tumors in at-risk patients [8, 9]. The urinary bladder tumor has a high tendency toward multifocality ( 50% of all cases) [10] and recurs after treatment either at the initial tumor site or elsewhere throughout the transitional epithelium in 30 80% of patients [8, 9, 11]. Considering the specific features of the disease, it is important to apply diagnostic methods with a maximum sensitivity and the capability to reliably visualize the entire organ. Virtual MR endoscopy is a recently developed noninvasive method to detect tumors protruding from the walls of hollow organs. promising advantage of this imaging modality is that views not possible in conventional endoscopic examination can be created [1]. t present, it is important to gather information about the reliability of the new method compared with conventional cystoscopy, which is the gold standard for detection of bladder tumors. This pilot study contributes to the growing experience in this new field. lthough further clinical trials must be performed to reach statistical significance, our study found a high overall detection rate (90.9%) for bladder tumors. The detection rate for tumors of 1 cm in diameter or greater was 100%. Virtual endoscopy revealed two tumors that were not seen on cystoscopy, our gold standard. Therefore, per definition, the positive predictive value for virtual endoscopy is 93.8%. Considering the two tumors revealed by virtual endoscopy as true-positive because they were confirmed later intraoperatively, the positive predictive value is 100% JR:178, June 2002

5 Virtual Cystoscopy of Cancer of the Urinary ladder Given the detection rate and positive predictive value found in this study, a high sensitivity and specificity can be expected for the presented method. The primary advantage of virtual endoscopy versus cystoscopy is that virtual MR cystoscopy is a noninvasive technique. Cystoscopy is an invasive and uncomfortable procedure for patients, and complications such as infections, perforation, scarring, and stricture of the urethra have been observed [13 17]. No complications resulted from MR imaging based virtual endoscopy in our study. Cystoscopy is often difficult to perform adequately when exploring the anterior bladder wall or the lumen of a diverticulum [18, 19]. The anterior bladder wall is a common location of bladder tumors [20]. Primary intradiverticular carcinomas are rare, but diagnosis is often difficult and prognosis is relatively poor. Therefore, imaging plays an important role in these cases [18, 19, 21 23]. ccess to the anterior bladder wall (Fig. 3) or the lumen of a diverticulum is not restricted in virtual endoscopy because various software reconstruction tools can be used. virtual observation point can be placed individually, and individual path tracking permits easy access to all areas of the hollow organ and visualization of the mucosa. Free choice of angle and the ability to magnify a focal point (zoom in) allow detection of small lesions (Fig. 4). In our study, at least two instances of neoplasm were not initially detected on conventional cystoscopy because of poor visibility related to hematuria. In distinction, visibility and detection of tumors on virtual endoscopy is not affected by hematuria. The disadvantages of virtual MR cystoscopy versus cystoscopy include difficulties in detecting flat tumors, and the fact that it is not possible to visualize the lumen of the urethra as is routinely done with conventional cystoscopy. Differentiation between small tumors and inflammatory swelling of the mucosa may be difficult, especially in patients with unsatisfactory bladder filling in whom folding of the vesical mucosa occurs. Inflammatory swelling of the mucosa thus could be misdiagnosed as a tumor, or small tumors could be missed on virtual endoscopy. nother disadvantage is that patients with metal prostheses or implants cannot be adequately examined. Finally, in MR imaging based virtual endoscopy, the scanning and reconstructing process is lengthy and costly. This factor, at present, limits the usefulness of this technique as a screening tool. Raw data from MR imaging provide a sufficient basis for reconstruction. Patients do not need to undergo CT and the external bladder filling used in CT-based virtual endoscopy, which has been described by authors who report a tumor detection rate in their studies that is similar to the results of our MR imaging based study [2 4]. Performing standardized reconstruction of our MR imaging raw data sets provided accurate and clinically useful visualization (Fig. 5). Our experience with virtual MR cystoscopy showed that once the operator becomes familiar with the intraluminal perspective, time needed for postprocessing could be reduced by establishing a routine in performing and analyzing endoscopic views. Postprocessing and evaluation of virtual cystoscopic views were performed in approximately min, depending on reviewers experience, hard disk usage of the workstation, zoom status, artifacts, mucosal morphology, and the number of tumors. In assessing virtual endoscopic images, we observed striping and wrap-around artifacts. The causes are probably related to motion artifacts resulting from long TE. Possible causes of motion artifacts are scanner vibration and detrusor contractions. The artifacts possibly interfere with detection of Fig year-old man with primary urinary bladder cancer., Coronal T2-weighted MR image shows two lesions (arrows) in right and left lower part of bladder in area of vesical trigone., Virtual endoscopic image focused on vesical trigone reveals that only one tumor is present (1.4 cm in diameter, arrow). Tumor is located near left ureteric orifice. Structure seen on right side in coronal T2-weighted image () is identified on virtual endoscopic image as part of normal vesical trigone and not tumor. Internal urethral orifice can be identified in lower mid portion. small lesions and thus may affect the detection rate and positive predictive value of the method. reath-triggered examinations were not successful in preventing artifacts; therefore, it seems that breathing motion may not affect image quality. Postprocessed images must be established from cross-sectional source data that can be used as a tool for the assessment of virtual endoscopic images. For interpretation of the reconstructed virtual endoscopic image, using the coronal T2-weighted source images is recommended for spatial orientation and quick assessment of the dimensions of a tumor that is already visible in cross-sectional imaging. Source images were used to provide a first overview. However, cross-sectional imaging for tumor detection is often confusing and is, therefore, not recommended (Fig. 6). In patients with diverticula, it is often impossible to differentiate between wall segments of diverticula and small tumors. It is useful, in assessing the intravesical image, to focus first on the vesical trigone with the ureteric orifices and the internal urethral orifice in anteroposterior orientation similar to the generally practiced standard in conventional cystoscopy (Figs. 3, 5D, and 6). From this position, it is possible to proceed to successively visualize all areas of the mucosa of the bladder and perform a standardized reconstruction. Standardizing reconstruction allows easy orien- JR:178, June

6 Lämmle et al. tation inside the lumen of the bladder and identification of the exact tumor position on the internal bladder wall. Therefore, mapping of the internal surface of the bladder wall is possible. This technique allows examination of the complete organ and provides accurate and reproducible results. These characteristics are important to fulfill the requirements of diagnosing urinary bladder cancer, which has a high rate of multifocality and recurrence. When MR imaging is used as the basis for virtual endoscopic reconstruction, there is an obvious interest in simultaneous contribution of the MR images to clinical staging of the bladder tumors by revealing information about depth of invasion, metastasis of lymph nodes, and infiltration of pelvic organs. arentsz et al. [24] compared accuracy of different staging techniques for bladder carcinoma and showed the potential use of MR imaging in this field. eyersdorff et al. [25] identified even small tumors with a T2-weighted turbo spin-echo sequence after intravesical administration of superparamagnetic iron oxide particles, but they found it was not possible to reliably differentiate the layers of the bladder wall. However, MR imaging can provide only indicative information for staging of bladder cancer. Staging of bladder cancer and the deduced prognosis and therapeutic strategies depend primarily on histologic analysis; staging; and grading of a specimen from transurethral resection, biopsies, or cytology [8, 9, 11, 26]. MR imaging cannot yet provide differentiation of histologic layers of the urinary bladder wall or provide a definitive diagnosis without histopathologic confirmation [24, 25, 27]. On the basis of the results of this study and our experience with conventional cystoscopy and virtual endoscopy, we propose that virtual cystoscopy is indicated when conventional cystoscopy is difficult to perform, restricted in feasibility, unsatisfactory in interpretation, or contraindicated. Contraindications for cystoscopy are bacteriuria, acute cystitis, urethritis, prostatitis, obstructive prostatic hypertrophy, and stricture or rupture of the urethra. Factors that limit the technical success of cystoscopy, thereby decreasing its reliability, include marked hematuria, bladder diverticula, and poor visibility of the anterior urinary bladder wall. Occasionally, after cystoscopy, there are no pathologic findings in spite of a strong clinical suspicion of bladder cancer. In these patients, virtual cystoscopy can be used complementary to conventional cystoscopy to enhance diagnostic performance. MR imaging based virtual endoscopy is also indicated for patients who are at risk of complications such as hemorrhage, perforation, infection, or pain, and for the examination of young patients. The modality may also be used at the patient s request. In the future, we expect improved detection rate and sensitivity for even very small tumors through continued development and advancement of hardware and software, with the possible use of virtual MR cystoscopy for screening and follow-up of bladder cancer, and in the clinical setting in which cystoscopy cannot be performed or is ineffective. In conclusion, the results of our study suggest a high reliability in the diagnosis of urinary bladder cancer by MR imaging based virtual cystoscopy. This noninvasive method, independent of any medication or contrast enhancement, can be of value for screening, primary diagnosis, and surveillance of bladder cancer, which is a frequent malignancy with a high rate of recurrence and multifocal manifestations. t this time, virtual MR cystoscopy may be indicated as a clinical routine when conventional cystoscopy is contraindicated or restricted in feasibility or interpretation; when there is risk of hemorrhage, perforation, or pain; for young patients; or if a patient wishes a noninvasive assessment of bladder mucosa. References 1. lezek DJ, Robb R. Evaluating virtual endoscopy for clinical use. J Digit Imaging 1997;10: Merkle EM, Wunderlich, schoff J, et al. Virtual cystoscopy based on helical CT scan data sets: perspectives and limitations. r J Radiol 1998;71: Fenlon HM, ell TV, hari HK, Hussain S. Virtual cystoscopy: early clinical experience. Radiology 1997;205: Gualdi GF, Casciani E, Rojas M, Polettini E. Virtual cystoscopy of bladder neoplasms: preliminary experience [in Italian]. Radiol Med (Torino) 1999;97: Nolte-Ernsting CC, Krombach G, Staatz G, Kilbinger M, dam G, Grunther RW. Virtual endoscopy of the upper urinary tract based on contrast-enhanced MR urography data sets [in German]. Rofo Fortschr Geb Rontgenstr Neuen ildgeb Verfahr 1999;170: Schreyer G, Fielding JR, Warfield SK, et al. Virtual CT cystoscopy: color mapping of bladder wall thickness. Invest Radiol 2000;35: Frank R, Stenzl, Fede T. Three-dimensional computed tomography of the reconstructed lower urinary tract: technique and findings. Eur Radiol 1998;8: Soloway MS, Perito PE. Superficial bladder cancer: diagnosis, surveillance and treatment. J Cell iochem 1992;16I[suppl]:120S 127S 9. Soloway MS. Selecting initial therapy for bladder cancer. Cancer 1987;60: Cheng L, Cheville JC, Neumann RM, et al. Survival of patients with carcinoma in situ of the urinary bladder. Cancer 1999;85: Soloway MS. Diagnosis and management of superficial bladder cancer. Semin Surg Oncol 1989; 5: merican Cancer Society. Cancer facts and figures. vailable at: STT/F&F2001.pdf. ccessed March 18, Westenfelder M, Rosset K, Pelz K. Development of nosocomial and iatrogenic urinary tract infections (UTI) following urological interventions: a prospective clinical study. Scand J Urol Nephrol 1987;104[suppl]:59S 63S 14. Clark KR, Higgs MJ. Urinary infection following out-patient flexible cystoscopy. r J Urol 1990; 66: avetta S, Olsha O, Fenely J. Spreading sepsis by cystoscopy. Postgrad Med J 1990;66: Mosbah, Kane, Zhani R, Hattab C. Iatrogenic urethral strictures of the male urethra [in French]. cta Urol elg 1990;58: Golomb J, Waizbard E, Iellin, Merimsky E. Recurrent bladder perforation in chronic irradiation cystitis. J Urol (Paris) 1986;92: ubert J, Dombriz M, Dore. Intradiverticular bladder tumor: apropos of 3 cases [in French]. J Urol (Paris) 1982;88: Durfee SM, Schwartz LH, Panicek DM, Russo P. MR imaging of carcinoma within urinary bladder diverticulum. Clin Imaging 1997;21: Stephenson WT, Holmes FF, Noble MJ, Gerald K. nalysis of bladder carcinoma by subsite: cystoscopic location may have prognostic value. Cancer 1990;66: aniel J, Vishna T. Primary transitional cell carcinoma in vesical diverticula. Urology 1997;50: Melekos MD, sbach HW, arbalias G. Vesical diverticula: etiology, diagnosis, tumorigenesis, and treatment analysis of 74 cases. Urology 1987;30: Dondalski M, White EM, Ghahremani GG, Patel SK. Carcinoma arising in urinary bladder diverticula: imaging findings in six patients. JR 1993;161: arentsz JO, Engelbrecht MR, Witjes JM, de la Rosette JJ, Van der Graaf M. MR imaging of the male pelvis. Eur Radiol 1999;9: eyersdorff D, Taupitz M, Giessing M, et al. The staging of bladder tumors in MRT: the value of the intravesical application of an iron oxide-containing contrast medium in combination with high-resolution T2-weighted imaging [in German]. Rofo Fortschr Geb Rontgenstr Neuen ildgeb Verfahr 2000;172: Sobin L, Wittekind C. TNM classification of malignant tumours, 5th ed. Union Internationale Contre le Cancer (UICC). New York: Wiley, 1997: ltwein JE, Kurth KH, Hohenfellner R. ladder carcinoma: therapeutic concept of the Urology Department of the University Hospital, Mainz [in German]. Urologe 1977;16: JR:178, June 2002