Validation of virtual colonoscopy in the detection of colorectal polyps and masses: rationale for proper study design

BACKGROUND

Colorectal cancer, the second-leading cause of cancer-related mortality, is a preventable malignancy in many cases. Despite the availability of several screening modalities, compliance with screening recommendations remains unacceptably low. Virtual colonoscopy is a novel, minimally-invasive technique with the potential to increase colorectal cancer screening rates, but its effectiveness must first be validated. Published studies comparing virtual colonoscopy to conventional colonoscopy have reported varying results. These discrepancies may be attributed to differences in bowel preparation and scanning techniques, as well as errors in endoscopic lesion measurement, endoscopic colonic segmental localization, and the ability of conventional colonoscopy to actually detect lesions. These methodological issues can affect scientific results and ultimately affect the public's perception of this emerging technique.

AIM

The goal of this report is to expose existing methodological shortcomings and propose solutions incorporated in this study design. This article describes the rationale, study design, and outcome definitions of a single-center, blinded, direct comparative trial aiming at assessing the ability of virtual colonoscopy to detect colorectal polyps and masses relative to the criterion standard, conventional colonoscopy. DESIGN FEATURES: Bowel preparation was standardized using oral sodium phosphate lavage, orally administered iodinated contrast, and controlled colonic insufflation. Segmental unblinding allowed a second-look when results were discrepant and polyp matching was performed using an algorithm based on segmental localization and lesion size determination.

CONCLUSIONS

This methodology could be applied to other studies assessing the accuracy of virtual colonoscopy in order to have uniformity of results.

Virtual endoscopy

Virtual endoscopy is a technique in which three-dimensional viewing of hollow structures is conducted through the utilization of high-resolution imaging and unique computer processing methods. The basic components of this technique and its applications for urology and other clinical disciplines are reviewed.

Evaluation of airway obstruction using virtual endoscopy

OBJECTIVES

This study examines the use of virtual endoscopy (VE) in the evaluation of patients with upper airway obstruction. The utility of VE compared with actual endoscopy was investigated with respect to accuracy of diagnosis and reproduction of endoscopic images.

STUDY DESIGN

A random cohort of 30 patients with various causes of airway obstruction was examined.

METHODS

The computed tomography (CT) data were reconstructed using a proprietary VE software program, FreeFlight, blind to the actual endoscopic findings. The cause of obstruction was identified and compared with actual endoscopic findings. This included 21 patients with airway stenoses, 8 patients with laryngotracheomalacia, 3 tracheal tumors, 2 glottic webs, 5 patients with innominate artery compression, 2 tracheal granulomas, and 7 patients with impaired true vocal cord mobility.

RESULTS

Virtual endoscopic evaluation was accurate in assessing stenosis width and length of fixed airway lesions. Correlation of stenosis shape and contour between actual endoscopy and VE was excellent. The stenosis-to-lumen ratios were compared between VE and actual endoscopy and were found to be within 10% (SD = 8). However, virtual endoscopic evaluation could not illustrate one of the glottic webs, half of the cases of tracheomalacia, or any of the cases of impaired true vocal cord mobility.

CONCLUSIONS

Virtual endoscopy was not as sensitive as actual endoscopy in detecting the cause of airway obstruction that was based on dynamic movement. However, VE was excellent for the measurement and definition of fixed airway lesions.

Automatic segmentation of the colon for virtual colonoscopy

Virtual colonoscopy is a minimally invasive technique that enables early detection of colorectal polyps and cancer. Normally, a patient's bowel is prepared with colonic lavage and gas insufflation prior to computed tomography scanning. An important step for 3D analysis of the image volume is segmentation of the colon. The high-contrast gas/tissue interface that exists in the colon lumen makes segmentation of the majority of the colon relatively easy; however, two factors inhibit automatic segmentation of the entire colon. First, the colon is not the only gas-filled organ in the data volume: lungs, small bowel, and stomach also meet this criterion. User-defined seed points placed in the colon lumen have previously been required to spatially isolate the colon. Second, portions of the colon lumen may be obstructed by peristalsis, large masses, and/or residual feces. These complicating factors require increased user interaction during the segmentation process to isolate additional colonic segments. To automate the segmentation of the colon, we have developed a method to locate seed points and segment the gas-filled lumen sections without user supervision. We have also developed an automated approach to improve lumen segmentation by digitally removing residual contrast-enhanced fluid. Experimental results with 20 patient volumes show that our method is accurate and reliable.

Virtual bronchoscopy

The sophistication of three-dimensional (3-D) radiographic imaging from CT data has accelerated with the development and ongoing advances of helical CT. Virtual bronchoscopy (VB), the ability to create 3-D models of the airways and navigate through the tracheobronchial tree lumen in realtime simulated bronchoscopy, has gained popularity over the past 3 years. The ability of VB to image the airway and mediastinal structures simultaneously and in a 3-D format has helped revolutionize CT imaging. Unlike conventional bronchoscopy, VB can display the extent of narrowing caused by an airway lesion, the presence of patentcy beyond a stenosis, and the relationship between an airway lesion and the adjacent mediastinal structures.

Computing the centerline of a colon: a robust and efficient method based on 3D skeletons

We present a robust and efficient algorithm for calculating the centerline of a computer-generated colon model created from helical CT image data. The centerline is an essential aid for navigating through complex anatomy such as the colon. Our algorithm involves three steps. In the first step, we generate a 3D skeleton of the binary colon volume using a fast topological thinning algorithm. In the second step, we employ a graph search algorithm to remove extra loops and branches. These loops and branches are caused by holes in the object that are artifacts produced during image segmentation. In the final step, we compute a smooth representation of the centerline by approximating the skeleton with cubic B-splines. This final step is necessary because the skeleton contains many abrupt changes in direction due to the discrete nature of image data. The user supplies two endpoints for the centerline; otherwise, the algorithm is fully automated. Experimental results demonstrate that the algorithm is not only robust but also efficient.

Virtual bronchoscopy

Three-dimensional endoluminal tracheobronchial simulations can be derived successfully from thoracic helical CT scans, and can reproduce the appearances of major endobronchial abnormalities confirmed during FB. The prospects of ever-faster CT scanners (capable of submillimeter resolution) merged with greater computer power make it likely that current versions of virtual bronchoscopy images will seem primitive in the future. Initial descriptive reports suggest great potential, but the startling visual appeal of these 3-D portrayals of a patient's airway and mediastinal anatomy and the prospects of exploring this information in real time do not establish its clinical role. Such virtual bronchoscopy findings are generally predictable on the basis of currently available axial CT images alone. The extent to which these 3-D endobronchial renderings improve the already high predictive values of CT requires critical study. In their patients with lung cancer Cicero et al observed that neither the staging nor diagnosis was modified substantially, but virtual bronchoscopy contributed to enhanced understanding of the pathology of the neoplastic process. Whether this added perspective translates to tangible benefits for patients is an intriguing possibility that has yet to be proved. The unique 3-D endobronchial view may offer particular advantages in some individuals and contribute to the patient's noninvasive evaluation. Because of the already high yield of conventional CT, diagnostic yield alone is not likely to be the sole best measure of this evolving technology. Accordingly, future multidisciplinary research investigations will also need to prospectively address nuances of decision-making and measure appropriate patient outcomes. In these efforts the active dialogue between chest clinician and radiologist will remain essential to defining and realizing the true potential of virtual bronchoscopy.

Virtual colonoscopy

Virtual colonoscopy (VC) is an evolving technique that combines volume scanning and computer visualization methods to enable minimally invasive and effective colorectal cancer screening. Although VC offers significant clinical and public health advantages over conventional endoscopic screening, several issues confront VC's emergence into the medical marketplace.

Rectal imaging and cancer

Rectal imaging has evolved substantially during the past 25 years and now offers surgeons exquisite anatomic detail and physiologic information. Dynamic cystoproctography, helical computed tomography, endoscopic ultrasonography, endorectal magnetic resonance imaging, and immunoscintigraphy have become standards for the diagnosis of rectal disease, staging of neoplasia, and survey of therapeutic results. The indications, limitations, and relative costs of current imaging methods are reviewed, and advances in imaging technology that promise future benefits to colorectal surgeons are introduced.

Imaging of mediastinal lymph nodes: CT, MR, and FDG PET

The evaluation of mediastinal lymph nodes is an important aspect of staging in patients with non-small cell lung cancer. Anatomic imaging of lymph nodes with computed tomography (CT) and magnetic resonance (MR) imaging has been limited by the relatively low sensitivity and specificity of these techniques. Advances in physiologic imaging of mediastinal lymph nodes with 2-[fluorine-18] fluoro-2-deoxy-D-glucose (FDG) positron emission tomography (PET) have resulted in improved diagnostic accuracy in the determination of nodal status. Despite the limitations of CT, this technique still plays an important role by aiding in the selection of the most appropriate procedure for staging, by guiding biopsy, and by providing anatomic information for visual correlation with FDG PET images. At present, anatomic MR imaging of lymph nodes is primarily a problem-solving tool for cases with inconclusive CT results. Physiologic MR imaging with iron oxide is an exciting area of investigation, and the accuracy of this technique is being assessed in clinical trials. Anatomic and physiologic imaging techniques should be considered complementary rather than competitive imaging strategies.

Interpretation of abdominal CT: analysis of errors and their causes

PURPOSE

Our goal was to analyze those factors contributing to the error rate in the interpretation of abdominal CT scans at an academic medical center.

METHOD

From a total of 694 consecutive patients (329 male, 365 female), we evaluated the error rates of interpreting abdominal CT studies. The average patient age was 54 years. All abdominal CT studies were reviewed by three to five CT faculty radiologists on the morning after the studies were performed. The error rate was correlated with reader variability, the number of cases read per day, the presence of a resident, inpatient versus outpatient, organ systems, etc. The chi 2-test was used for statistical analysis.

RESULTS

A total of 56 errors were found in the reports of 53 patients (overall error rate = 7.6%). Of these errors, 19 were judged to be clinically significant and 7 affected patient management. A statistically significant difference in error rates was noted among the five faculty radiologists (3.6-16.1%, p = 0.00062). No significant correlates between error rates and any of the other variables could be established.

CONCLUSION

The primary determinant of error rates in body CT is the skill of the interpreting radiologist.

Virtual colonoscopy

A new method for colon examination, virtual colonoscopy, has been created by combining spiral computed tomography and virtual reality computer technology. If virtual colonoscopy proves equal to or superior to traditional colonoscopy in sensitivity and specificity for cancer screening, it would yield significant clinical and public health benefits because of its potential for increased acceptance among patients as well as its considerably lower cost and risks.

Tracheobronchial tree: three-dimensional spiral CT with bronchoscopic perspective

The purpose of this report is to describe an original 3D technique that permits bronchoscopic perspective simulations of the tracheobronchial tree on an affordable workstation. Software based on surface shading was developed to permit navigation through the central airways using spiral CT data. Multiplanar reformations are displayed simultaneously with the virtual bronchoscopy simulations to give information about tissues surrounding the airways. Two clinical examples are reported to illustrate the value of this advanced computer technique.

Virtual bronchoscopy. Relationships of virtual reality endobronchial simulations to actual bronchoscopic findings

Advances in computer technology have permitted development of virtual reality images of the tracheobronchial tree using data sets derived from helical CT of the chest. To determine the relevance of these images to actual bronchoscopic findings, we compared "virtual bronchoscopy" images with videotaped bronchoscopy results in 20 patients who had undergone both helical chest CT and fiberoptic bronchoscopy during clinical evaluation of their thoracic problems. Suboptimal endobronchial simulations in ten patients identified important, readily-addressed technical requirements for this imaging procedure. In the ten patients with technically suitable renderings of the airway, virtual bronchoscopy simulations accurately demonstrated endobronchial obstructions by tumor in five, airway distortion and/or ectasia in four, and accessory bronchi in another. These preliminary observations suggest that virtual bronchoscopy simulations accurately represent major endobronchial anatomic findings. This technique may have a role in prebronchoscopy planning, endoscopy training, and/or endobronchial therapy, and merits further study.

Receiver operating characteristic curves: a basic understanding

Receiver operating characteristic (ROC) is one form of an objective measurement that can be used to compare newer imaging technologies against human observer performance (the ability of the expert radiologist). Use of ROC curves allows one to account for a continuum of radiologic interpretations when calculating sensitivity and specificity for an imaging modality and avoids the inaccuracies that arise from assuming that imaging findings are absolutely normal or abnormal. An ROC curve is generated by plotting sensitivity on the y axis as a function of [1-specificity] on the x axis for a continuum of diagnostic criteria. ROC curves allow visual analysis of the trade offs between the sensitivity and the specificity of a test with regard to the variable diagnostic criteria used by radiologists. Because ROC curve analysis is gaining wide acceptance in medical literature, an explanation of ROC methods with the use of simple examples is necessary to increase the knowledge and understanding of practicing radiologists.

Childhood odontoid fractures evaluated with computerized tomography. Case report

A case is presented that illustrates the use of computerized tomography for the evaluation of an odontoid fracture in a child. Such a "fracture" may actually represent a "synchondrotic slip" of the odontoid and C-2 vertebral body between the neural arches of C-2. Treatment should almost always be conservative. Complete healing of the injury is expected.