Machine learning models could help create a more standardized, reproducible, and efficient method to rank the severity of small intestinal Crohn's disease (CD) based on CT images.
New research published in academic radiology We compare the use of a machine learning model and the performance of two radiologists in assessing CD severity in a cohort of computed tomography enterography (CTE) scans, and find that a hybrid model improves this often subjective task. We discovered that it may hold clinical value in streamlining the process.
“Advances in artificial intelligence, computer vision, and machine learning provide new approaches for standardized assessment and explainable CD quantification,” write Ashish P. Wasnik, M.D., of the University of Michigan's Department of Radiology, and colleagues. . “Previous studies have shown that automated methods can efficiently extract and quantify traditional CD imaging features defined by radiologists and employed in clinical practice, thereby making automated This allows for reproducible evaluations.”
In this study, experts compared the severity scores of two radiologists who interpreted 236 CTE cases to severity scores generated by a hybrid machine learning model that combines deep learning, 3D CNN, and a random forest model. I compared it with Each investigator was tasked with classifying disease severity in each distal and distal ileum minisegment. Performance was compared using accuracy, sensitivity, weighted Cohen score, and precision.
