Data Analysis and Interpretation Strategies


Data Analysis and Interpretation Strategies

After the CT examination is complete, the image data are transferred via a network or via data storage media from the CT scanner to a workstation, where the datasets are converted into 2D and 3D images for evaluation.

Several different workstations nowadays feature specialized CT colonography software designed for image postprocessing and data interpretation. Various hardware and software solutions are also available from CT manufacturers and independent software producers. Along with 2D axial and multiplanar image reconstructions, all of these systems also offer interactive, manual, automated, or semiautomated 3D “fly-throughs” of the colonic lumen which simulate an endoscopic examination (hence the commonly used term “virtual colonoscopy”). Various 3D evaluation and functional applications are now available. CT colonography datasets are routinely analyzed with a combination of 2D and 3D visualizations, using either 2D or 3D views to search for lesions and then the alternate view (3D or 2D) to characterize the detected findings.

2D Evaluation

Axial Images

The data obtained at CT colonography may be evaluated on 2D axial images wherever there is a dedicated workstation or a PACS console. Evaluation is straightforward and time-efficient, as there is no need for further assessment of complex 2D or 3D reconstructions. Additional hardware or software is unnecessary, so this approach is also inexpensive.

Two-dimensional axial views are the primary standard images that are reconstructed from raw data. The evaluation of only axial views obtained with CT colonography offer high sensitivity and specificity for detection of large polyps (1 cm in size or more). By scrolling through a stack of axial CT images, the entire colonic lumen can be visualized continuously, slice by slice.

Slice thickness. Thin collimations of 0.6–0.75 mm with reconstructed slice thicknesses of 1 mm are preferred and are now considered standard. The reconstructed slice thickness should not exceed 2.5 mm.

Full screen views. Images should be enlarged to a full-screen size, because the target structures (e.g., polyps measuring about 1 cm in diameter) are very small in relation to the total amount of imaging information contained in a CT colonography dataset.

Lumen tracking. In 2D evaluations, the use of a technique known as “lumen tracking” is standard. Lumen tracking allows the investigator to follow the air-distended colon from one end to the other by scrolling with the mouse, concentrating on the cross-section of a single bowel segment (Fig. 3.1). Special care should be taken when assessing the colonic flexures: Here, the superior and inferior walls of the colon are sliced tangentially, which may reduce the conspicuity of polypoid lesions. Lumen tracking allows a continuous and systematic assessment of the entire colon.

Lumen tracking. In lumen tracking, the colonic lumen is tracked continuously in 2D cine mode from the rectum to the cecum, always focusing on just the cross-section of a single colon segment.

Avoid supposedly time-saving overview assessments of multiple bowel segments on a single view, especially coronal views.

The target lesion—a polyp 10 mm in size or greater—is a relatively small and often difficult-to-detect structure that can occur anywhere on the colon wall and may be located anywhere in the dataset. The lesion is easily missed if the examiner fails to concentrate on each individual colon segment separately. Investigators should ignore the remainder of the CT dataset in the first instance, only returning to evaluate it later in a second step.

Evaluation of internal structure and window settings. Two-dimensional axial images enable direct evaluation of the internal structure of a lesion in terms of homogeneity and attenuation. For a detailed evaluation of the colon, interactively adjusting the window setting from wide to narrow can be helpful.

Two-dimensional interpretation of CT colonography data generally uses two window settings: a wide window, similar to the classic lung window (width: 1500 HU, level: –150 HU), and a narrow soft-tissue window (width: 400 HU, level: 10 HU).

Wide window settings. Wide settings, such as lung or bone windows (the setting recommended above is between lung and bone window), provide a strong contrast between the lumen and the wall structures of the colon. This enables imaging of more of the bowel wall, such as the semilunar folds, which are not visualized on narrow window settings. Use of a wide window setting thus facilitates the detection of polypoid structures. Depiction of the internal structure of detected lesions is poorer, however.

Narrow window settings. Soft-tissue window settings are better suited for displaying the internal structure of a lesion with regard to tissue density and homogeneity (soft-tissue density polyp versus a lipoma or residual feces). However, a disadvantage of narrow window settings is the loss of detail of the bowel wall. On narrow window settings, the thin semilunar folds or the walls of two adjacent colon segments, small lesions, and pericolic air (e.g., in perforation of the bowel) are often barely distinguishable, if at all. Narrow window settings are better for depicting flat lesions in which often only a slight soft-tissue thickening of the wall is detectable (Fig. 3.2).

Individualized window settings. For proper evaluation of the colon on CT colonography, it is necessary to image the entire bowel wall as well as details of the structure and thickness of the wall and of potential lesions. For this reason, it is essential that any detected lesions are further evaluated using various window settings adjusted as required. This enables detection of air, areas of increased attenuation, and fatty tissue. The choice of window setting also influences manual 2D polyp measurement. Narrow window settings make polyps appear smaller; wide windows are best for showing their actual size (Fig. 3.3).

Suboptimal bowel preparation and colonic distension. Despite certain limitations, 2D views enable evaluation of suboptimal datasets such as those obtained in patients with large amounts of residual fluid or stool. Two-dimensional views also enable imaging of poorly distended bowel segments, for which 3D views are not feasible.

Correlation of prone and supine images. Simultaneous display and correlation of axial images in the prone and supine positions allows immediate determination of whether a detected finding is mobile or not (Fig. 3.4). This helps to determine whether it is a true lesion that is adherent to the bowel wall or a pseudolesion that moves.

Window settings and differentiation of lesions. Effect of window settings on visualization of anatomic structure and differentiation of lesions. a Wide window settings allow good visualization of the semilunar folds and the bowel wall (*). Two polyps are also readily detectable, one in the transverse colon (arrow), the other one in the sigmoid colon (arrowhead). However, the internal structure of the two polyps cannot be adequately evaluated. b Narrow window settings (soft-tissue windows) result in a loss of anatomical detail at the semilunar folds and the bowel wall (*). However, this window setting shows the fat attenuation of the polypoid lesion (= lipoma) in the transverse colon (arrow) and the soft-tissue attenuation of the polyp in the sigmoid colon (arrowhead).
Window settings and polyp measurement. Measuring a polyp with an incorrect window setting can result in false categorization by size. a With (correct) wide window settings, a polyp size of 11 mm in diameter (arrowheads) is measured. b When measured with a soft-tissue window, the same polyp appears distinctly smaller (about 9 mm in diameter). This could result in false categorization of the lesion by size.
Correlation of prone and supine scans for evaluating the mobility of a sessile polyp. a On a prone axial 2D image, a small, broad-based polyp is seen on the ventral wall of the descending colon (arrow). b When the patient changes position the lesion does not move, suggesting that it is a true polyp (arrow). Supine/prone scan correlations increase not just the sensitivity but also the specificity of CT colonography.

2D Multiplanar Reformations

In addition to axial images, a stack of images in other planes may be reconstructed from the isotropic datasets. Most workstations come with one or two additional imaging planes in the form of coronal or sagittal multiplanar reformations.

2D Multiplanar reformation for problem solving. A typical use of 2D multiplanar reformation (2D-MPR) is for problem solving during primary evaluation using axial slices. A suspected lesion found on axial views may be further assessed and characterized on the corresponding coronal and sagittal views. This allows polypoid lesions to be easily differentiated from semilunar folds and complex fold structures (Fig. 3.5).

Combined 2D/3D evaluation. After a suspected lesion is detected on a 2D view with lumen tracking, the finding is correlated with the corresponding MPR and 3D views. The correlation of 2D views with a 3D endoluminal view increases diagnostic confidence and helps avoid misinterpretation. Here the diagnosis of a 1-cm pedunculated polyp is confirmed.
Polyp display time. Polyps are displayed for different lengths of time in 2D cine mode and in virtual colonoscopy. a In 2D cine mode, this small polyp is displayed on only two images. The lesion may therefore only be visible for a few fractions of a second when scrolling with the mouse through the CT dataset. b On the 3D endoluminal view, the bowel segment bearing the polyp is in the reader′s field of view for several seconds. Thus, it is easier to detect the polyp in the 3D view.

2D Multiplanar reformation and lumen tracking. Multiplanar reconstructions may also be used as an addition to axial views in lumen tracking. Cross-sectional evaluation of a bowel segment along the longitudinal axis is easier than interpreting tangential views. In tangentially sliced bowel segments, partial volume effects affecting the bowel wall are more common. This is why the axial plane is unsuitable for evaluating the superior and inferior walls of the colonic flexures, and the coronal plane is less useful for evaluating the anterior and posterior walls of the large bowel. In addition, a segment of the bowel that has been imaged in a tangenital plane is larger and thus more difficult to analyze than an axial cross-section. Therefore, to assess the transverse colon at the flexures, for example, it is helpful to switch from the axial to the sagittal view. This shows the transverse colon perpendicular to the lumen.

An automated variant of this evaluation strategy is to use an automated centerline flight path for visualizing the colon. In this approach, the 2D imaging plane is set perpendicular to the axis of the bowel segment.

Only gold members can continue reading. Log In or Register to continue

Stay updated, free articles. Join our Telegram channel

Jun 26, 2020 | Posted by in GASTROENTEROLOGY | Comments Off on Data Analysis and Interpretation Strategies

Full access? Get Clinical Tree

Get Clinical Tree app for offline access