Purpose: To evaluate the accuracy of computed tomography colonography (CTC) in differentiating chronic diverticular disease from colorectal cancer (CRC), using morphological and textural parameters.
Materials and methods: We included 95 consecutive patients with histologically proven chronic diverticular disease (n = 53) or CRC (n = 42) who underwent CTC. One radiologist, unaware of histological findings, evaluated CTC studies for the presence of potential discriminators including: maximum thickness, involved segment length, shouldering phenomenon, growth pattern, diverticula, fascia thickening, fat tissue edema, loco-regional lymph nodes, mucosal pattern. Another radiologist performed volumetric texture analysis on the involved segment.
Results: Several qualitative imaging parameters resulted to significantly correlated with colorectal cancer, including absence of diverticula in the affected segment, straightened growth pattern and shouldering phenomenon. A maximum wall thickness/involved segment length ratio < 0.1 had 98% specificity and 47% sensitivity in identifying diverticular disease. Regarding first-order texture analysis parameters, kurtosis resulted to be significantly different between the two groups.
Conclusions: Absence of diverticula, straightened growth pattern and shouldering phenomenon are significantly associated with CRC (71-91% sensitivity; 82-91%).
Keywords: Abdomen; Colonoscopy; Colorectal cancer; Computed tomography colonography; Diverticular disease; Texture analysis.
© 2021. Japan Radiological Society.