An algorithm for early detection of anemia in CNN Convolutional Neural Network
Annotatsiya
Anemia is one of the most common health problems among children and pregnant women. This disease is caused by a decrease in the number of red blood cells (erythrocytes) due to the deformation of the damaged hemoglobin (Hb) cells. This can lead to insufficient oxygen delivery to the body. The main causes of anemia are insufficient production of red blood cells, their quality is destroyed, or their mass is lost. In this study, a new model based on the improved YOLO 8 algorithm is proposed for early assessment and diagnosis of anemia. The YOLO 8 model is based on convolutional neural networks (CNN) and advanced object recognition techniques for blood cell classification and classification. Compared with other models, the proposed model differs in providing several advantages. (Adaptive histogram equalization, Optimized convolution layers, Discrete filter training, Deconvolution layer expansion, Improved bounding box prediction and non-maximal). With this improvement, YOLO 8 achieved an accuracy of 94.65%, which is significantly higher than traditional methods. The newly proposed model is characterized by accuracy and real-time performance. The three-dimensional model will be of great importance in the future.