Design And Research Of Carotid Artery Auxiliary Diagnosis And Treatment System Based On Deep Learning

Carotid intima-media thickness,morphology of carotid atherosclerotic plaques,luminal diameter of common carotid artery,and degree of carotid artery stenosis,all of which are considered important information for clinical assessment of stroke risk,as well as carotid atherosclerosis.Traditional segmentation methods are easily affected by objective environmental factors and subjective judgments of doctors,and the segmentation quality is uneven.The segmentation method based on deep learning avoids some influencing factors,reduces the workload of medical workers,and improves the stability of segmentation and the efficiency of diagnosis.At the same time,it is difficult to quantify the results of manual segmentation,and the numerical data is very helpful for data analysis and mining.Regarding the issue above,we do research on the segmentation and calculation algorithms of carotid artery clinical data under ultrasound images,and implements a carotid artery auxiliary diagnosis and treatment system based on deep learning.The research contents of this thesis are as follows:1.We improve image segmentation accuracy.In this thesis,a segmentation model based on attention mechanism,Carotid-Net,is proposed to implement the segmentation algorithm of carotid artery clinical data under ultrasound images.The model proposed in this thesis has made two improvements on the basis of Attention UNet to improve the accuracy of model segmentation.One is to add a stripe pooling module,and the other is to add a dual attention module.At the same time,the loss function is optimized,and an appropriate optimizer is selected.The segmentation effect of the Carotid-Net model is better than that of the same type of medical image segmentation model on the test data set,and can be applied to the carotid artery auxiliary diagnosis and treatment system to improve the segmentation accuracy.2.We calculate the segmentation result.Based on the segmentation results of the Carotid-Net model,this thesis implements an algorithm for pixel-level measurement of the segmentation results.At the same time,the system converts the pixel measurement results into corresponding measurement units according to the requirements of clinical data,and quantifies the data to better display the patient’s carotid artery health.3.We implement the carotid artery auxiliary diagnosis and treatment system.The system is based on Web development and is divided into three modules: diagnosis and treatment business module,ultrasound image storage module,and clinical data segmentation and calculation module.The system covers the collection and storage of carotid ultrasound images,intelligent segmentation and calculation of carotid artery clinical data,and quality control of segmentation and calculation results.The core components of the system are deployed in a modular form,which reduces the coupling degree of the system,makes the operation simple and friendly,reduces the workload of doctors,and provides doctors with a full information of carotid artery status.