Volumetric Textural Analysis of Colorectal Masses at CT Colonography: Differentiating Benign versus Malignant Pathology and Comparison with Human Reader Performance

Abstract

Rationale and objectives: To (1) apply a quantitative volumetric textural analysis (VTA) to colorectal masses at CT colonography (CTC) for the differentiation of malignant and benign lesions and to (2) compare VTA with human performance.

Materials and methods: A validated, quantitative VTA method was applied to 63 pathologically proven colorectal masses (mean size, 4.2 cm; range, 3-8 cm) at noncontrast CTC in 59 adults (mean age, 66.5 years; range, 45.9-91.6 years). Fifty-one percent (32/63) of the masses were invasive adenocarcinoma, and the remaining 49% (31/63) were large benign adenomas. Three readers with CTC experience independently assessed the likelihood of malignancy using a 5-point scale (1 = definitely benign, 2 = probably benign, 3 = indeterminate, 4 = probably malignant, 5 = definitely malignant). Areas under the curve (AUCs) and accuracy levels were compared.

Results: VTA achieved optimal sensitivity of 83.6% vs 91.7% for human readers (P = .034), with specificities of 87.5% and 77.4%, respectively (P = .007). No significant difference in overall accuracy was seen between VTA and human readers (85.5% vs 84.7%, P = .753). The AUC for differentiating benign and malignant lesions was 0.936 for VTA and 0.917 for human readers. Intraclass correlation coefficient among the human readers was 0.76, indicating good to excellent agreement.

Conclusion: VTA demonstrates excellent performance for distinguishing benign from malignant colorectal masses (≥3 cm) at CTC, comparable yet potentially complementary to experienced human performance.

Keywords: CT colonography; colorectal cancer; texture analysis.

Fig. 1—. Large tubulovillous adenoma at CT colonography.

3D CTC image from asymptomatic 74-year-old woman shows a lobulated polypoid mass in the sigmoid colon that measures 3 cm. This corresponds to the middle 2D image on the bottom row in Fig. 3, and proved to be a tubulovillous adenoma.

Fig. 2—. Expansion of Haralick Model from 2D to 3D Space.

The left image illustrates Haralick’s original four directions (0°, 45°, 90°, 135° from the center reference pixel) used for calculation of textural measures in 2D space. Expansion to 3D space, illustrated on the right, allows for thirteen directions (from the center reference voxel) to be used in calculation of textural measures.

Fig. 3—. CT colonography images showing semi-automated segmentation.

Collage of CTC images showing the colorectal masses that were segmented semi-automatically for volumetric textural analysis. Four of these cases were adenocarcinoma (the top left image and the entire middle row). The remaining masses were all benign tubulovillous adenomas (TVA). Note the middle image of the bottom row corresponds to the lesion presented in Figure 1.

Fig. 4—. ROC curves for human readers versus volumetric textural analysis.

ROC curves demonstrating performance of volumetric textural analysis (VTA) versus individual and pooled human readers.