|
||||||||||||||||||||||||||||||||||
|
Home
» Archive
» 2023
» Veterinary Session
Veterinary sessionKállai Petra - year 6 University of Veterinary Medicine Budapest, Department of Pathology Supervisor: Dr. Horváth Dávid Géza In dogs, soft tissue sarcomas are one of the most frequently diagnosed heterogeneous tumor groups. The expected recurrence and metastasis tendency of these tumors is closely related to the histological grade, which is traditionally determined by a pathologist based on the degree of tumor differentiation, the mitotic count and the extent of necrotic areas. The grade must be an accurate and reproducible prognostic factor, but at the same time there may be disparities due to the subjectivity between examiners. Due to the spread of digital pathology, the quantitative, objective and reproducible analysis of digitized sections has gained remarkable relevance. In our research, we digitized 50 hematoxylin-eosin-stained tissue slides from different dogs, and the mitotic count together with the extent of necrotic areas were determined using QuPath image analysis software. We evaluated the performance of the program by comparing the digitally corrected grade with the manually determined one. We had agreement between 39 of the 50 examined cases (78%), the kappa value was 0.669, which shows a substantial agreement between the software’s and the pathologist's classification. The significant accordance may have been due to the exclusive manual evaluation of the differentiation and the simpler evaluation of the necrotic areas. Although the recognition of mitotic cells proved to be the most complicated, significantly different numbers of dividing cells could not influence the grade to a large extent. Partially digitized grading systems can potentially make the diagnostic work faster, more objective and more reproducible, but the training of the programs cannot lack the cooperation of expert pathologists. List of lectures |