| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
|
|
|
|||||||
|
|||||||||
Gastrointestinal Imaging |
1 From the Departments of Surgery (G.N.B., A.B.W., D.T., C.R.G.C.) and Intestinal Imaging (S.H., C.I.B.), St. Mark’s Hospital, Level 4V, Watford Road, Northwick Park, Harrow, London HA1 3UJ, England. Supported by a Kodak Bursary from the Royal College of Radiologists. Received October 24, 2003; revision requested January 13, 2004; revision received March 3; accepted April 8. Address correspondence to S.H. (e-mail: s.halligan@imperial.ac.uk).
 |
ABSTRACT |
|---|
 TOP ABSTRACT INTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
MATERIALS AND METHODS: Ethical committee approval and informed consent were obtained. A total of 104 patients who were suspected of having fistula in ano underwent preoperative digital examination, 10-MHz anal endosonography, and body-coil MR imaging. Fistula classification was determined with each modality, with reviewers blinded to findings of other assessments. For fistula classification, an outcome-derived reference standard was based on a combination of subsequent surgical and MR imaging findings and clinical outcome after surgery. The proportion of patients correctly classified and agreement between the preoperative assessment and reference standard were determined with trend tests and 
statistics, respectively.
RESULTS: There was a significant linear trend (P < .001) in the proportion of fistula tracks (n = 108) correctly classified with each modality, as follows: clinical examination, 66 (61%) patients; endosonography, 87 (81%) patients; MR imaging, 97 (90%) patients. Similar trends were found for the correct anatomic classification of abscesses (P < .001), horseshoe extensions (P = .003), and internal openings (n = 99, P < .001); endosonography was used to correctly identify the internal opening in 90 (91%) patients versus 96 (97%) patients with MR imaging. Agreement between the outcome-derived reference standard and digital examination, endosonography, and MR imaging for classification of the primary track was fair ( = 0.38), good ( = 0.68), and very good ( = 0.84), respectively, and fair ( = 0.29), good ( = 0.64), and very good ( = 0.88), respectively, for classification of abscesses and horseshoe extensions combined.
CONCLUSION: Endosonography with a high-frequency transducer is superior to digital examination for the preoperative classification of fistula in ano. While MR imaging remains superior in all respects, endosonography is a viable alternative for identification of the internal opening.
© RSNA, 2004
Index terms: Anus, abnormalities, 757.245 • Anus, MR, 757.1214 • Endoscopy, 757.1298 • Fistula, gastrointestinal tract, 757.245
|
INTRODUCTION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Although now generally available, there are many health care systems worldwide where access to MR imaging remains restricted. MR imaging is also relatively expensive. Fistulography, although less expensive and more readily available, has been shown to be inaccurate (9), and assessment with computed tomography (CT) is hindered by inadequate tissue contrast (10). Anal endosonography is an alternative technique that is used for imaging the anal sphincter complex (11), notably in incontinent patients (12). While preoperative investigation of fistula in ano was one of the earliest applications of anal endosonography (13), its benefit for this indication remains uncertain. Some authors have found it to be very useful (14,15), while others have found it to be no better than digital examination (16). Direct comparisons have demonstrated anal endosonography to be superior (17), equivalent (18,19), and inferior (20,21) to MR imaging. Sonographic probe technology has improved dramatically in recent years (22), and the role of anal endosonography now needs further definition in light of these developments and in relation to the simultaneous advances made with MR imaging. Thus, the purpose of our study was to prospectively evaluate the relative accuracy of digital examination, anal endosonography, and MR imaging in the preoperative assessment of fistula in ano by comparison with an outcome-derived reference standard.
|
MATERIALS AND METHODS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Clinical Assessment
Clinical evaluation in the outpatient department was performed either by one of eight consultant colorectal surgeons with 9–30 years of colorectal experience or by one of 15 senior colorectal trainees who were in their last year of subspecialty training and had 3–6 years of colorectal experience. Senior trainees had been trained by the consultant surgeons and subsequently performed examinations on a regular basis. Trainee assessment in this study was performed without supervision.
Clinical assessment was performed a mean of 79 days ± 68.1 prior to surgery (range, same day to 405 days); a full history was obtained, and an examination was performed with the benefit of hospital notes and referral letters. Abdominal examination was followed by anorectal assessment with patients in the left lateral decubitus position. Inspection and palpation of the perineum was combined with digital examination of the anorectum and proctosigmoidoscopy to determine the course of any fistula track and any associated extensions. Fistula probes were not used for fear of creating additional tracks.
On the basis of this assessment, the clinician independently completed a fistula classification sheet, which was derived from a report by Parks et al (23) and well established in previous imaging studies of fistula in ano (4,8,13,16). With this sheet, the primary track of any fistula, if clinically present, was recorded as inter-, trans-, supra-, or extrasphincteric (23). Superficial fistulas (ie, those not thought to actually cross the anal sphincter muscle) were grouped with intersphincteric fistulas. The location of the enteric communication (ie, the internal opening) was recorded with respect to a clock face, per standard practice, and its level was recorded as either anal or rectal. The anatomic location of any abscess or extension arising from the primary fistula track was recorded as follows: superficial, intersphincteric, ischorectal, or supralevator (2,23). A horseshoe extension was defined as any extension from the primary track that appeared to extend to both sides of the internal opening (2,23). Any track that appeared to have a blind ending (ie, where an internal opening could not be defined clinically) was classified as a sinus rather than a fistula.
The study coordinator collected the classification sheet after the outpatient visit and sealed it in an envelope for later comparison against an outcome-derived reference standard. All patients underwent subsequent anal endosonography and MR imaging; physicians were blinded to the results of any other assessment.
Anal Endosonography
One of two experienced gastrointestinal radiologists (S.H., C.I.B.) or two surgical research fellows (G.N.B., A.B.W.) performed anal endosonography. All were experienced in this technique. The radiologists had more than 10 and 6 years of experience, respectively, with the technique at the commencement of study recruitment. Both surgical research fellows had more than 1 year of experience, each having been trained by the two radiologists and having subsequently performed examinations on a daily basis for that period. Anal endosonography performed by research fellows was unsupervised.
Scanning was performed a mean of 18 days ± 36.4 before surgery (range, same day to 183 days). A type 3535 (69 patients) or 2100 (35 patients) scanner (B-K Medical, Herlev, Denmark) was used. A 10-MHz transducer (type 6004; B-K Medical) was attached to a rotating endoprobe (type 1850; B-K Medical) and used in all patients. The transducer has a focal length of 5–45 mm, transverse resolution of less than 0.05 mm, lateral resolution of 0.5–1.0 mm, and beam thickness of 0.8 mm (22). The transducer was covered with a hard sonolucent plastic cone with a 1.7-cm diameter; the assembly was filled with degassed water for acoustic coupling, covered with a lubricated condom, and inserted into the anal canal (24). Men were placed in the left lateral position, and women were placed in the prone position.
Fistulas were identified by using established criteria, namely a hypoechoic track, occasionally containing focal hypoechoic areas representing gas within the fistula. Serial transverse images of the anal canal were obtained during slow probe withdrawal by the sonologist at two different magnifications and printed on laser film. The sonologist independently recorded the findings on a sheet that was identical to that used in the clinical examination by using the same definitions. The sonologist was blinded to the clinical findings in outpatients and was only aware that the patient potentially had a fistula. This sheet was then sealed by the sonologist in an envelope for later comparison against an outcome-derived reference standard. All scans were deemed technically adequate, and no patient was recalled for further imaging.
MR Imaging
After anal endosonography, all patients underwent 1.0-T MR imaging (Gyroscan T10-NT; Philips Medical Systems, Reigate, Surrey, England) in the supine position on the same day by using either a previously described and well-validated body protocol (13 patients) or a phased-array surface coil (91 patients) protocol (4,8,10). Usual contraindications to MR imaging applied. One of the gastrointestinal radiologists (S.H., C.I.B.) supervised MR imaging, and images were allocated such that the radiologist was blinded to any findings at prior anal endosonography (ie, anal endosonography had been performed by the nonsupervising radiologist or one of the fellows). Again, the radiologist was also unaware of any outpatient findings, except that the patients potentially had a fistula. Both radiologists had 3 years of experience with MR imaging for fistula in ano at the commencement of the study, and each had performed 300 examinations at this time.
The long axis of the anal canal was identified by using a midline sagittal-localizing sequence and subsequent transverse and coronal short inversion time inversion-recovery sequences planned with respect to the anal canal axis by using the following parameters: repetition time msec/echo time msec, 1500/15; field of view, 375 mm; matrix, 256 x 256; section thickness, 4 mm; intersection gap, 1 mm; four signals accquired. Sagittal T2-weighted images were used in approximately 20% of patients at the discretion of the supervising radiologist when a presacral extension was suspected from the transverse and/or coronal short inversion time inversion-recovery images, by using the following parameters: 4563/150; field of view, 350 mm; matrix, 256 x 256; section thickness, 6 mm; intersection gap, 0.6 mm; four signals accquired. No intravenous contrast material or endorectal and/or endoanal receiver coil was used.
Established criteria for MR diagnosis of a fistula were used, namely a high-signal track related to the sphincter complex (10). The site of the internal opening was inferred from the presence of sepsis in the intersphincteric space (10). The supervising radiologist independently recorded the MR imaging findings on a fistula sheet identical to that used for the clinical and anal endosonography assessments, again noting his (S.H., C.I.B.) opinion of fistula classification, internal opening, and any associated extensions. The radiologist then sealed this sheet in an envelope, which was subsequently collected by the study coordinator for later comparison against an outcome-derived reference standard.
Examination with Anesthesia
Examination with anesthesia was performed by one of nine colorectal consultant surgeons with 9–30 years of colorectal experience in 92 patients and by 10 colorectal trainees with 3–7 years of colorectal experience who were in their last year of subspecialist training in 12. Surgical research fellows did not perform surgery. No attempt was made to allocate patients according to whether the surgeon had performed the outpatient clinical assessment. The outpatient clinical assessment previously recorded in the notes was available at the time of surgery, per usual clinical practice.
Examination with anesthesia was performed in a standard fashion (2) by using fistula probes and hydrogen peroxide to search for internal openings, as deemed appropriate. Because previous studies (4–8,17–21) have established that an examination with anesthesia alone is imperfect, MR imaging results were available at the time of the examination to maximize the surgeons’ chance of detecting fistula tracks and extensions. Examination with anesthesia was performed with MR imaging at the outset in 14 patients. The examination was initially performed while blinded to the results of MR imaging in the remaining 90 patients as part of an ongoing trial (8) to determine the therapeutic benefit of MR imaging at examination with anesthesia; although we initially wanted to perform a randomized trial, ethical permission could not be obtained to withhold MR imaging results. In these 90 patients, the study coordinator revealed the MR imaging results after the initial examination with anesthesia. Surgical reexploration and laying open and/or coring out of fistula tracks was performed as deemed necessary; setons (ie, surgical threads used to drain regions of infection) (2) were placed if appropriate, until the surgeon was satisfied that both assessment and treatment were complete. A fistula classification sheet was completed after examination with anesthesia that included any additional information obtained with MR imaging. This was independently documented by the operating surgeon on a fistula sheet identical to that used for preoperative assessment. This sheet was retained by the study coordinator and used subsequently for the outcome-derived reference standard.
Outcome-derived Reference Standard
Because it is well established that the most common cause of fistula recurrence is infection missed at examination with anesthesia (1,2,4–6,8), patients were followed up to determine clinical outcome and to identify patients in whom infection had been missed. This procedure was especially important because not all surgeons chose to act on MR imaging findings that indicated they had potentially missed occult infection during surgical exploration. Patients were followed up for a mean period of 23 months ± 11 (range, 3–46 months) and assessed in the outpatient department, with special attention paid to clinical evidence of fistula healing. Follow-up was undertaken by the surgical team responsible for the patient’s care; at this stage, the results of the clinical examination, MR imaging, anal endosonography, and surgical findings were available. The frequency and duration of follow-up was not altered for the purpose of this study. The study coordinator documented if any patients required further unplanned surgery because of failure to heal or further recurrence; however, further surgery was planned in some patients to lay open a track after a period of seton drainage or to remove a seton. The study coordinator documented the findings at any subsequent unplanned surgery.
An outcome-derived reference standard for fistula classification was determined for each patient on the basis of clinical outcome and findings at MR imaging and examination with anesthesia; any fistula or abscess found at subsequent unplanned surgery was incorporated in this standard. This was an extension of the methods of Schwartz et al (19), who found that a combination of at least two modalities was necessary to be confident that classification was correct. We believed it was important to include assessment of the clinical course after surgery, because fistula healing is the only definitive assurance that all infection has been identified and treated (1,2).
We used the following criteria: In 74 patients, the fistula classification at MR imaging and the final surgical classification at examination with anesthesia were in agreement. This was defined as the outcome-derived reference standard, because examination with anesthesia was followed by complete healing or control of infection at sites where loose setons were used in these patients. In the remaining 30 patients, there was some disagreement between the fistula classification at MR imaging and the final surgical classification at examination with anesthesia; thus, clinical outcome was used for arbitration. Of these 30 patients, 13 had a fistula track or perianal abscess at a site that was initially noticed at MR imaging but not found at examination with anesthesia. These patients underwent further unplanned surgery that confirmed infection at the site that was initially noticed at MR imaging; thus, the MR imaging classification was assumed to be correct. The remaining 17 patients did not need further unplanned surgery, but there was disagreement between findings at MR imaging and examination with anesthesia. In these patients, the study coordinator—in collaboration with other study investigators—reached a consensus reference standard by using the method of Schwartz et al (19), supplemented by knowledge that these patients had healed after examination with anesthesia. Thus, if MR imaging indicated an abscess that had not been found at examination with anesthesia and if healing followed this finding, the finding at MR imaging was assumed to have been incorrect. In this way, an outcome-derived reference standard for fistula classification was arrived at in all 104 patients.
Statistical Analysis
Fistula classification at outpatient clinical and digital assessment, anal endosonography, and MR imaging were compared with the outcome-derived reference standard to determine the level of agreement for each individual assessment. Fistula tracks were considered correctly classified when placed into the correct anatomic group, extensions (abscess and/or horseshoe) when described in the correct anatomic compartment and quadrant, and internal openings when described within the same quadrant and at the correct level (defined as anal or rectal) according to the reference standard. Significant differences in correct classification rates for categorical variables and the a priori hypothesis that clinical examination would fare worst and MR imaging best (with anal endosonography intermediate) was examined by using the 
2 test statistic for trend, with significance assigned at the 5% level. The accuracy of consultant versus trainee fistula assessment in outpatients was assessed by using the Fisher exact test. Agreement between the various categorical assessments and the reference standard was assessed by using the weighted test statistic, with 95% confidence intervals. Analysis was performed by using Arcus Quickstat Biomedical, version 1.2 (Research Solutions, Cambridge, England).
|
RESULTS |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
|
|
|
Recurrent Disease
Of the 76 patients with recurrent disease, the outcome-derived reference standard was used to establish 17 superficial or intersphincteric fistulas, 50 transsphincteric fistulas, three suprasphincteric fistulas, and four extrasphincteric fistulas. There were three patients with a sinus and three with no evidence of infection. Four patients had two tracks each, resulting in 80 possible classifications in this group of 76 patients. Of these possible classifications, 49 (61%) were correctly classified with clinical examination, 64 (80%) with anal endosonography, and 70 (88%) with MR imaging. There were 74 internal openings; 56 (76%) were correctly identified with clinical examination, 67 (91%) with anal endosonography, and 71 (96%) with MR imaging. Four of these internal openings were rectal, all of which were correctly identified with MR imaging; however, clinical examination and anal endosonography depicted only one internal opening each. There were 42 abscesses in this group, 11 (26%) of which were correctly classified with clinical examination, 34 (81%) with anal endosonography, and 37 (88%) with MR imaging (Fig 3). There were 16 horseshoe extensions in this group, seven (44%) of which were correctly classified with clinical examination, nine (56%) with anal endosonography, and 15 (94%) with MR imaging.
|
|
|
Agreement
Agreement between the modality investigated and the reference standard for classification of the primary track is shown in Table 2. values and 95% confidence intervals for digital assessment, anal endosonography, and MR imaging were 0.38 (0.25, 0.5), 0.68 (0.55, 0.81), and 0.84 (0.71, 0.97), which indicate fair, good, and very good agreement, respectively (25).
|
values and 95% confidence intervals for digital assessment, anal endosonography, and MR imaging were 0.29 (0.2, 0.38); 0.64 (0.54, 0.74); and 0.88 (0.78, 0.98), which indicate fair, good, and very good agreement, respectively (25).
|
|
DISCUSSION |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
Because of these findings, we used examination with anesthesia guided by MR imaging to ensure our reference standard was as accurate as possible; however, this meant the reference standard was not truly independent of MR imaging. Indeed, our hospital ethical committee decreed that there was overwhelming evidence that the MR images had to be available in the operating room (4–8,17–21,26) and that nondisclosure was unethical. In any event, the preoperative MR classification had already been established by the radiologist and could not be changed as a consequence of findings at examination with anesthesia. Importantly, intraoperative disclosure of MR imaging findings allowed us to control for instances when infection might have otherwise been missed by the surgeon and resulted in a false-positive result for MR imaging. This type of procedure, whereby the reference standard is modified or enhanced by the modality being investigated, is increasingly used when the unmodified reference standard may occasionally be inaccurate. For example, in studies of CT colonography, reference colonoscopy is often modified by disclosure of the imaging results during the procedure.
We also used clinical follow-up to try to ensure that our reference standard was as robust as possible, because there were instances when the operating surgeon disagreed with the findings at MR imaging. For example, the surgeon occasionally diagnosed a sinus because he could not locate the internal opening that had been seen on MR images. Also, there were instances when the radiologist suggested that an abscess that could not be subsequently palpated was present; however, the operating surgeon was unwilling to blindly make an incision and search for it (8). The true fistula classification in such patients can only be resolved by means of the clinical course after surgery, because patients in whom infection has been overlooked will inevitably experience a relapse (1,2). Thus, our outcome-derived reference standard was as close as possible to the true fistula classification in each patient and afforded us the best means with which to judge the accuracy of the preoperative investigations. Most importantly, we wanted to define the relative position of anal endosonography, which was why this test was not used to influence the outcome-derived reference standard.
Our primary aim was to define the relative accuracy of anal endosonography when in experienced hands and with use of a modern 10-MHz transducer. Further, we used investigators experienced in both anal endosonography and MR imaging to eliminate technique bias. Our a priori hypothesis was that digital examination would fare the worst and that MR imaging would fare the best. In this we were proved correct; MR imaging was the most accurate modality in all comparisons made. Anal endosonography, although inferior to MR imaging, was always superior to clinical examination. Indeed, agreement for anal endosonography was generally "good" in the category immediately adjacent to MR imaging, which was "very good". Clinical examination tended to be "fair", which was separated from the category "good" by the category "moderate" (25).
Anal endosonography is therefore likely to be a worthwhile test when MR imaging is unavailable. For example, anal endosonography was used to correctly classify 81% of all primary tracks in our study, and our confidence intervals for anal endosonography and MR imaging overlapped. Anal endosonography was also especially good at allowing us to correctly predict the internal opening in 91% of patients, which approaches the 97% achieved with MR imaging. Identification of the internal opening is a major surgical aim, since failure to locate it is highly predictive of recurrence (8). Anal endosonography is well suited to the identification of the internal opening both rapidly and inexpensively.
Anal endosonography depicted fewer extensions than MR imaging, probably because of field-of-view limitations. Nevertheless, when considering abscess and horseshoe extensions together, anal endosonography still correctly depicted the majority of these abnormalities, in contrast to clinical examination. MR imaging, however, is undoubtedly the best preoperative modality to use for the detection of secondary tracks. Failure to identify and treat these abnormalities is the most common cause of relapse, which explains the prevalence of secondary tracks in patients with recurrent disease (1,2,8). Of the patients included in the study, 76 had recurrent disease, which reflects the tertiary referral nature of our unit; furthermore, it should be borne in mind that this case mix presents substantial difficulties for any imaging modality. Because of this, we hypothesized that any incremental benefit attributable to MR imaging in comparison to anal endosonography might be minimized in patients with primary disease, since extensions are less common and infection is more likely to be confined to the sphincter complex, thus allowing field-of-view problems to be overcome. Of the six abscesses in the 28 patients with primary disease, however, anal endosonography depicted only two and actually fared worse than clinical examination in this subgroup analysis, which was the only occasion it did so.
It is not clear to what degree our results for anal endosonography were influenced by improved transducer technology, since this occurred simultaneously with improved understanding of sonographic anal anatomy as a consequence of endoanal MR imaging (21,27,28). We did not, however, use endoanal MR receiver coils, because it is well recognized that their limited field of view makes them less suited to depiction of remote infection (27,29). There is no doubt, however, that the shorter focal length provided by a 10-MHz transducer provides better spatial resolution than that of older 7-MHz models, both within and adjacent to the anal sphincters (22).
The findings in the present study confirm the relative inaccuracy of fistula assessment when compared with imaging. It also suggests that digital assessment may not improve with experience, because we found no difference between the findings of consultants and trainees. It should be borne in mind, however, that trainees were in their last year of subspecialist training and the high proportion of recurrent cases means the group was particularly difficult to assess.
If MR imaging and expertise in its interpretation are available, our results suggest that the best option for preoperative assessment would be to progress straight to MR imaging. Although we found that MR imaging was used to incorrectly classify 10% of primary tracks when compared with the reference standard, this was predominantly due to minor differences in track description that made little real difference to treatment. For example, there were several instances where the track crossed the very distal fibers of the subcutaneous external sphincter, which is technically a transsphincteric fistula, but was often described as intersphincteric at MR imaging.
Our study does have some limitations. It has been suggested that hydrogen peroxide might improve anal endosonography (30–32). Anecdotally, we have not found this approach to be valuable, since it has many of the limitations that affect fistulography. The sonologists had differing levels of experience with anal endosonography, although all were trained and reported the examinations independently at the time this study was conducted. Similarly, the surgeons who performed the assessments had different levels of experience, although all were colorectal specialists, and any trainees were in their final year of training and had passed their subspecialist exit examination. The difficulty of defining a true reference standard for fistula in ano has been discussed previously. Anal endosonography was not used to influence the outcome-derived reference standard, which could potentially represent a source of bias. We believe that using clinical outcome as the final arbiter minimized potential biases.
In summary, MR imaging is the most accurate preoperative technique for classification of fistula in ano and performs best in the evaluation of the primary track and any extensions. While MR imaging is generally the preferred technique, anal endosonography is demonstrably superior to digital examination and is particularly adept in aiding detection of the internal opening. Anal endosonography may be used when MR imaging is unavailable or expertise in its interpretation is lacking.
|
FOOTNOTES |
|---|
Author contributions: Guarantor of integrity of entire study, S.H.; study concepts, S.H., C.I.B.; study design, S.H.; literature research, G.N.B., S.H.; clinical studies, all authors; data acquisition, all authors; data analysis/interpretation, G.N.B., S.H., C.I.B.; statistical analysis, G.N.B., S.H.; manuscript preparation, G.N.B.; manuscript definition of intellectual content and editing, S.H., C.I.B.; manuscript revision/review, all authors; manuscript final version approval, S.H.
|
REFERENCES |
|---|
|
TOPABSTRACTINTRODUCTIONMATERIALS AND METHODSRESULTSDISCUSSIONREFERENCES |
|---|
This article has been cited by other articles:
|
|
 |
|
  R. Greenhalgh, C R. Cohen, D. Burling, and S. A. Taylor Investigating perianal pain of uncertain cause BMJ, February 16, 2008; 336(7640): 387 - 389. [Full Text] [PDF]
|
|
|
|
 |
|
 A. Jhobta, R. Kaur, R. Jhobta, A. Dabra, and S. Kochhar Fistula in Ano: Role of Transperineal and Transvaginal Sonography Journal of Diagnostic Medical Sonography, November 1, 2006; 22(6): 375 - 381. [Abstract] [PDF]
|
|
|
|
 |
|
 Y. Hama, K. Makita, T. Yamana, and K. Dodanuki Mucinous adenocarcinoma arising from fistula in ano: MRI findings. Am. J. Roentgenol., August 1, 2006; 187(2): 517 - 521. [Abstract] [Full Text] [PDF]
|
|
|
|
 |
|
 S. Halligan and J. Stoker Imaging of fistula in ano. Radiology, April 1, 2006; 239(1): 18 - 33. [Abstract] [Full Text] [PDF]
|
|
| ||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||||
| HOME | HELP | FEEDBACK | SUBSCRIPTIONS | ARCHIVE | SEARCH | TABLE OF CONTENTS |
| RADIOLOGY | RADIOGRAPHICS | RSNA JOURNALS ONLINE |
