Development of an automatic blood cell counting device using deep learning algorithms

The amount of blood cells in the blood is the basic building blocks of blood that provides information about the general health status of individuals or their diseases, if any. Detection and determination of the amounts of red, white blood cells and platelets in the blood are very important for human health. In the management of all these processes, basic factors such as the complexity of cell structures, loss of time, and the necessity of expert opinion make the realization of these processes quite complicated. In this study, 37 different current object detection algorithms consisting of Yolo and Detectron2 algorithms were used to detect and count blood cells automatically, to detect them quickly and to determine their amounts. The test and training results of the 37 algorithms used were examined comparatively. It was determined that the Yolo11-l model is the most suitable model for a fully automatic blood counting device with its high performance rate, low margin of error, and integration speed to an electronic card. With the deep learning-based fully automatic blood cell detection and counting device produced, laboratory costs can be reduced, the need for expert personnel can be reduced, and it has been observed that early and accurate diagnosis of diseases can be achieved by preventing personal comments.