Diagnostic model for Crohn�s disease based on magnetic resonance enterography layered bowel wall enhancement and apparent diffusion coefficient (Preprint)

BACKGROUND Human imaging research transitions from mapping local effects to developing predictive models that integrate information distributed across various MRI modalities. OBJECTIVE We hypothesised that based on magnetic resonance enterography (MRE) layered bowel wall enhancement and apparent diffusion coefficient (ADC), measured in the affected parts of the intestine, one would be able to effectively differentiate active and chronic phases of Crohn’s disease. The aim of this study was to create a multidimensional diagnostic model for differentiating between Crohn’s disease phases. METHODS This study included 125 patients, 55 women (44.0%; aged 19 to 66 years) and 70 men (56.0%; aged 12 to 67 years), who underwent MRE and ADC measurement for the first time. No patients had been previously treated for Crohn’s disease. The group of potential explanatory variables comprised 11 variables, including the thickness of the occupied section, length of the occupied section, number of lymph nodes present, layered bowel wall enhancement, total transitions on fat tissue, features of restricted diffusion in diffusion-weighted imaging (DWI), and ADC values. RESULTS The final discrimination model was based on only two variables, namely ADC (A) and layered bowel wall enhancement (W). Active Crohn’s disease was defined as -6.339 + 4.747 × W + 0.008 × A, while chronic Crohn’s disease was defined as -11.365 + 2.812 × W + 0.012 × A. CONCLUSIONS The predictive model described here could identify the active form of Crohn’s disease with a probability of 93.06% and the chronic form with a probability of 75.57%. The use of classic MRE layered bowel wall enhancement and a DWI-based ADC metric eliminates the main shortcomings of both approaches. CLINICALTRIAL Not applicable