Extraction method of neuroimaging biomarkers based on interpretable integrated 3DCNN (Three Dimensional Neural Neural Network)

• Main Authors: YANG BAOYAO, ZENG AN, PAN DAN
• Format: Patent
• Language: Chinese; English
• Published: 16.09.2022
• Subjects: CALCULATING; COMPUTER SYSTEMS BASED ON SPECIFIC COMPUTATIONAL MODELS; COMPUTING; COUNTING; DIAGNOSIS; HUMAN NECESSITIES; HYGIENE; IDENTIFICATION; IMAGE DATA PROCESSING OR GENERATION, IN GENERAL; MEDICAL OR VETERINARY SCIENCE; PHYSICS; SURGERY

Aiming at the limitation of the prior art, the invention provides an explainable integrated 3DCNN (three-dimensional convolutional neural network)-based neural image biomarker extraction method. According to the method, a new neural image biomarker P-score is derived from a prediction result obtained from an integrated three-dimensional convolutional neural network model. According to the scheme, scientific research personnel can be helped to research longitudinal track change of structural magnetic resonance imaging (sMRI) in the progress process of the Alzheimer's disease, and the relation between the longitudinal track change and neurodegenerative change of the Alzheimer's disease subject is analyzed; the extracted neuroimaging biomarker can provide a basis for predicting an intervention sequence of a brain area in neurodegenerative change of a patient suffering from the Alzheimer's disease and an imminent clinical symptom. 本发明针对现有技术的局限性，提出了一种基于可解释集成3DCNN（三维卷积神经网络）的神经影像学生物标志物的提取方法。本发明从集成三维卷积神经网络模型获得的预测结果，衍