Open AccessClassification of Ischemic Stroke Lesions Based on Cascaded Branch Compression Neural Network
Author(s) -
Fengbing Jiang
Publication year - 2019
Publication title -
iop conference series. materials science and engineering
Language(s) - English
Resource type - Journals
eISSN - 1757-899X
pISSN - 1757-8981
DOI - 10.1088/1757-899x/563/4/042003
Subject(s) - upsampling , computer science , convolutional neural network , artificial intelligence , artificial neural network , image (mathematics) , pattern recognition (psychology) , cascade , network architecture , ischemic stroke , computer vision , medicine , ischemia , engineering , computer network , cardiology , chemical engineering
Automatic differentiation of disease images with machine vision technology is of great significance for medical diagnosis. This paper proposes a deep neural network architecture to achieve the classification of ischemic stroke lesions. The proposed architecture uses a cascading approach. The image obtained by down-sampling each image and the original image are respectively input into two convolutional neural networks of different depths and are associated with each other through a cascade structure to solve the defect that the single network cannot balance the local features and the global features. In addition, the use of smaller images obtained after downsampling can effectively improve the operational efficiency of deep networks. The experimental results show that the proposed algorithm can balance the accuracy and timeliness well.