Classification of oestrus periods in rats with the YOLOv5 model using deep learning techniques

The aim of the present study was to accurately detect estrus phases in female rats using uterine images using deep learning techniques based on Convolutional Neural Networks. Since the 28-day menstrual cycle in humans ends in 4-5 days in rats, this animal species is preferred as a model organism in many studies, especially in studies related to the female reproductive system. In this study, images of Hematoxylin Eosin stained sections of the uterine tissue of female rats were taken under light microscope. With the images obtained, estrus periods in rats were classified histologically. After the examination, an artificial intelligence-based model was proposed for the classification of estrus periods in rats in the images obtained from uterine sections. In the study, the estrous period is classified into four phases: proestrus, estrus, metestrus and diestrus. In the proposed model, the classification success of sub-models of the YOLOv5 algorithm such as YOLOv5n, YOLOv5s, YOLOv5m were compared with histological results. With the YOLOv5m model, 98.3% accuracy, 99% precision, 98% recall and 98% F1-score values were obtained. The results show that the proposed model will provide a second opinion support to expert pathologists in the analysis of microscopic images.