AI Breakthrough in Crohn’s Disease: Automated Imaging Enhances Diagnosis

According to a study published in Inflammatory Bowel Diseases, an advanced deep learning model for Crohn’s disease (CD) has demonstrated the ability to automatically segment lesions in computed tomography enterography (CTE) images, significantly enhancing diagnostic precision.

By combining artificial intelligence with radiomics, researchers developed machine learning classifiers to distinguish between active and inactive CD, a key challenge in managing the disease.

The study, which analyzed 277 CTE exams across 2 datasets, trained the nnU-Net neural network to segment CD lesions. It achieved a Dice similarity coefficient of 0.824, indicating high accuracy in lesion identification. Radiomics features from the segmented lesions were used to build 5 machine-learning classifiers, with logistic regression emerging as the best performer. This model achieved an area under the curve of 0.862, along with sensitivity, specificity, and accuracy scores of 0.697, 0.840, and 0.759, respectively.

“The automated segmentation model accurately segments CD lesions, and the machine learning classifier distinguishes CD activity well,” the researchers concluded. “This method can assist radiologists in promptly and precisely evaluating CD activity.”

Reference:

Gao Y, Zhang B, Zhao D, et al. Automatic Segmentation and Radiomics for Identification and Activity Assessment of CTE Lesions in Crohn's Disease. Inflamm Bowel Dis. 2024;30(11):1957-1964. doi:10.1093/ibd/izad285