基于Faster R-CNN的疟疾血涂片检测改进算法

根据WHO发布的报告,每年疟疾的新发病例超过2亿,死亡人数仍居高不下.疟疾血涂片镜检法是疟疾检测的金标准,但由于人工评估所需的步骤繁琐,即使在经验丰富的医师手中,这种诊断方法也很耗时并且容易发生漏检和误检.此外疟原虫细胞形状、密度和颜色的变化以及某些细胞类的不确定性等因素,对疟原虫检测提出了重大挑战.基于深度学习的神经...

An Improved Algorithm for Malaria Blood Smear Detection Based on Faster R-CNN

According to the WHO,more than 200 million new malaria cases occur each year,and the number of deaths remains high.Malaria blood smear microscopy is the gold standard for malaria detection,but due to the tedious steps required for manual evaluation,it is time consuming and prone to missed and false tests,even by experienced physicians.In addition,factors such as changes in the shape,density,and color of plasmodium cells and the uncertainty of certain cell types pose major challenges for the detection of plasmodium.Neural network models based on deep learning have achieved great success in object detection,but the most advanced models have not been widely used in biological image data.Aiming at this problem,an improved Faster R-CNN algorithm based on deep learning is proposed to identify malaria blood smear cells and detect their infected stage.Based on the original Faster R-CNN,a convolution filter layer is added,a deep residual network with better extracted features is used,and the attributes of the anchor point are optimized to improve the problems of missed detection and false detection in the classification and detection of malaria blood smear cells.The experiment shows that the improved Faster R-CNN model has an average accuracy rate of 79.56%on the public blood smear dataset of plasmodium vivax(Malaria)infection,which is 8.84%higher than the original Faster R-CNN model.