The Design And Implementation Of Experimental Platform For Medical Image Segmentation Algorithm

Medical image segmentation aims to automatically outline the areas corresponding to human structures or lesions in medical images.It is of great significance for disease diagnosis,quantification of human structure areas,and visualization of medical images.It is a research topic that has received much attention in the field of computer vision.Medical images have multiple imaging modalities and can be collected from multiple organs.There are significant visual differences between different images,which leads to the tedious and complicated research process of medical image segmentation algorithms for different imaging modalities and organs.Since there a large number of modules that can be reused in different medical image segmentation algorithms,this paper starts from the perspective of speeding up the algorithm design and develops the design and implementation of the medical image segmentation algorithm experimental platform,which aims to integrate the cumbersome data processing process and network structure.The construction process is modularized,and the threshold of algorithm design is reduced to better promote the development and progress of medical image segmentation algorithm design.Specifically,the main work of this article is as follows:First,an experimental platform for medical image segmentation algorithm is designed.After investigating lots neural networks used in medical image segmentation,the design rules of the network were summarized and an experimental platform for medical image segmentation algorithms was developed.The platform is designed to meet the needs of scientific research and implements modules such as data analysis,data expansion,network structure design,network training,and result acquisition.It can support researchers to choose the type of data storage and perform automatic image analysis,and choose expansion methods to automatically expand data.It can support researchers to choose the basic network and advanced modules supported by the platform to design a segmented neural network structure for specific tasks,and carry out related training and testing to obtain more satisfactory results.The test analysis of this paper proves the rationality and practicability of the designed platform.Secondly,the medical image segmentation algorithm experimental platform designed in this paper is applied to the construction of medical image segmentation neural network.The work includes two aspects:First,we studied the construction of a dedicated segmentation network suitable for neuron images,fundus retina images,abdomen images,and skin images,achieved good results in the corresponding field data sets;Then,researched the construction of a general neural network for medical image segmentation to solve the problem of reduced performance when the network migrated to other image types,experiments show that the constructed ACE-Net structure shows outstanding performance on two different data sets.The above experiments have verified the effectiveness of the algorithm experimental platform designed in this paper.