AI-assisted digital morphometry application for the diagnosis of a benign peritoneal fibroma in a dog

Abstract

This paper reports a rare benign peritoneal fibroma in a 7-year-old dog and demonstrates the use of artificial intelligence (AI)–assisted digital morphometry in veterinary pathology. The dog had presented with progressive abdominal distention. Upon examination, a well-encapsulated mass was found attached to the parietal peritoneum on a small vascular pedicle during exploratory laparotomy. A histopathological examination revealed interlacing bundles of spindle-shaped fibroblasts arranged within a stroma with abundant collagen. There was slight evidence of vascular proliferation, and no mitotic figures were observed, consistent with a benign fibromatous lesion. AI-based image analysis employed U-Net for segmentation and ResNet-based feature extraction to create quantitative morphometry parameters such that nuclear eccentricity, cytoplasmic area, and vascularization index were all consistent with a low proliferative rate and mostly organized stromal architecture observed on routine histopathology. Here, both AI-derived measures strengthened diagnostic confidence by reducing subjectivity and providing an objective foundation for evaluating canine soft-tissue tumors. The recovery after the operation was normal, and there was no recurrence noted at the follow-up visit 90 days later. This case demonstrates the benefit of introducing AI-based digital morphometry into routine histopathology to improve diagnostic accuracy, decrease interobserver variation, and facilitate data-driven characterization of soft tissue tumors in veterinary pathology.