Mitotic cell detection in histopathological images of neuroendocrine tumors using improved YOLOv5 by transformer mechanism

Automatic analysis of pathological images is important for the diagnosis and treatment of diseases. The use of computerized systems in this field is becoming increasingly common. Due to evolving technology and the speed of information needed, it is desirable for computers to be able to recognize objects like humans. Deep learning methods, which are a subfield of artificial intelligence, and image processing algorithms that recognize objects from images have been used in many fields in recent years, including healthcare. The aim of this study is to detect the mitoses in the histopathological images of neuroendocrine tumors using image processing methods based on deep learning. In our study, You Only Look Once-v5 (YOLOv5), the most widely used object recognition method, was used by combining the YOLOv5 transform module. YOLOv5 recognized mitotic cells with an accuracy of 0.80, a recall of 0.67, and an F1 score of 0.73, while the YOLOv5 transformer model recognized mitotic cells with an accuracy of 0.89, a recall of 0.68, and an F1 score of 0.77. The acceleration of the process and the objective evaluation will contribute significantly to an accurate and fast diagnosis. Another advantage is the time saved for pathologists, who can concentrate on important cases. In summary, automatic mitotic cell detection will facilitate tumor grade determination, treatment, and patient monitoring.