An Alzheimer's disease region detection method based on a convolutional neural network

Abstract

The invention discloses an Alzheimer's disease region detection method based on a convolutional neural network. A focus detection task is divided into three steps of feature extraction, focus positioning and focus classification. The method comprises the following steps: training an improved AlexNet neural network model to extract the characteristics of an original brain nuclear magnetic resonanceimage and generate a characteristic pattern; Then training an area generation network model to generate propose used for positioning a lesion area of the brain nuclear magnetic resonance image; And then sending the feature map and propodal into a subsequent classification network for classification training, so as to classify which disease stage the lesion area is in. The result shows that the method has a good detection effect on the Alzheimer's disease region of the brain nuclear magnetic resonance image.

Description

technical field [0001] The invention belongs to the technical field of medical image processing. Specifically, a method for detecting Alzheimer's lesion areas based on convolutional neural networks for the purpose of detecting Alzheimer's lesion areas in brain magnetic resonance images is designed. Background technique [0002] Alzheimer's disease is a typical senile degenerative disease, commonly known as senile dementia, clinically manifested as memory loss and loss of language ability. Currently, the number of Alzheimer's disease patients in China ranks first in the world. Therefore, early diagnosis of Alzheimer's disease has become very urgent. MRI images are a particularly useful means of assessing brain lesions clinically. Accurate detection of brain lesion areas is not only important for assisting doctors in treatment planning, but also for subsequent follow-up evaluations. However, the manual detection method is very time-consuming, and is easily affected by subj...

AI Technical Summary

Problems solved by technology

However, these above methods can only classify a single MRI image, and cannot detect the lesion area in the MRI image, which is still insufficient for the diagnosis of Alzheimer's disease

Moreover, the methods proposed by many scholars simply use the existing traditional classic image recognition network models to classify Alzheimer's disease. Knowledge is applied to the model

Images