Development and Assessment of a Computed Tomography Radiomics–Based Model to Distinguish Crohn’s Disease From Intestinal Tuberculosis

Although the clinical and routine imaging features of Crohn's disease (CD) and intestinal tuberculosis (ITB) share many similarities, the therapeutic approaches for these diseases differ dramatically, and misdiagnosis can result in inappropriate treatment, which may cause further harm. Simple and reliable methods are therefore needed to differentiate CD from ITB, and this study developed a novel clinically applicable predictive model to help in their differentiation.

In this retrospective study, 87 CD patients and 80 ITB patients were divided into training and testing cohorts. Clinical indicators were first evaluated to construct clinical models. Then, slice‐by‐slice regions of interest were manually delineated on arterial and venous phase CT images, and radiomics features were extracted and screened to establish a radiomics model. Finally, a comprehensive predictive model based on radiomics features and clinical indicators was established. Receiver operating curve analysis was used to evaluate the differential diagnostic efficacy of the different models, and decision curve analysis was used to assess their clinical benefit.

A comprehensive predictive model was established based on clinical indicators (age, T‐cell enzyme–linked immunospot, and perianal lesions) and radiomics score. The model's area under the receiver operating curve was 0.980 (95% confidence interval [CI]: 0.961–0.999) in the training cohort and 0.975 (95% CI: 0.936–1.000) in the testing cohort. Decision curve analysis and clinical impact curves further confirmed its clinical practicality.

The comprehensive predictive diagnostic model, combining both CT radiomics features and clinical indicators, showed high values in the differentiation of CD from ITB.