Research On Intelligent Detection And Segmentation Algorithms For Blood Cells Under Microscope

Human blood cell composition analysis is a commonly used technique for blood routine testing,which analyzes red blood cells,white blood cells,etc.in the human blood through the detection of medical instruments,and judges the patient’s health status from the analysis results.With the development of artificial intelligence technology,deep learning is increasingly being applied in healthcare,among which computer vision algorithms can be applied to blood cell detection and segmentation tasks,providing services for auxiliary healthcare.However,the current deep learning target detection model still faces issues of speed and accuracy compatibility,and due to the large number of blood cells observed under the microscope,there are cell overlap and adhesion phenomena.Therefore,achieving accurate detection and segmentation of blood cells is a very challenging task.This article mainly focuses on the detection and segmentation of blood cell images under a microscope,and studies and constructs a method that can achieve speed and accuracy compatibility.Blood cell detection and segmentation is one of the important tasks in medical image analysis,and in recent years,automated blood cell detection and segmentation technology has gradually been widely applied.However,the commonly used deep learning object detection algorithms for detecting and segmenting blood cells still have some shortcomings.This article addresses the issue of speed and accuracy incompatibility in current blood cell detection algorithms,and proposes a new lightweight algorithm model,Mobile YOLOv7,based on the currently performing object detection model YOLOv7.This article adopts methods such as introducing soft attention mechanism and adaptive spatial feature fusion to improve the accuracy of model detection.This article also adopts a lighter structure Mobile Vi Tv3 module to replace the original E-ELAN convolutional combination module in the backbone network of YOLOv7,reducing the number of parameters while also improving the detection speed of the model.Due to the fact that in clinical medicine,analyzing the composition of blood cells not only requires detection functions,but also sometimes requires blood cell segmentation.Therefore,this article also adds segmentation functions to the YOLOv7 object detection algorithm,making the model have stronger generalization ability.The above model has been extensively experimented and tested on three microscope blood cell datasets,and has achieved good performance indicators.