Abstract
Celiac disease is a common immune-mediate inflammatory disease of the small intestine triggered by an environmental agent (gluten) in genetically predisposed individuals. It affects approximately 1% of the population and is characterized by diarrhea, malabsorption, and villous atrophy of the intestinal mucosa. In the context of dynamics systems, entropy is a measure of uncertainty. In computer vision, entropy is a statistical measure of randomness that can be used to characterize the texture and complexity of the input image. The hypothesis of this proof-of-concept study was that the loss of physiological structural balance could be measured by entropy. Entropy was measured in 16 patients with celiac disease (57 biopsies, 7294 image patches) and compared with 20 cases of small intestine control (54 biopsies/whole-tissue sections, 15810 image patches). The series was expanded with 18 cases of unspecific duodenal inflammation (5966 image patches) and 5 cases of duodenal adenocarcinoma (3723 image patches). Analysis was performed using 224×224×3 input non overlapping image patches and Matlab 2026a entropy function. In comparison with control, celiac disease was characterized by lower histological entropy (6.04 ± 0.39 vs. 6.59 ± 0.40, respectively, P < 0.001). In conclusion, the histological evaluation of entropy as a topological analysis is useful for the differential diagnosis between normal intestinal mucosa control and celiac disease. Loss of randomness characterized celiac disease.