18-04-2023 дата публикации
Номер: CN115985488A
Принадлежит:
The invention discloses a brain disease diagnosis method, device and system based on a dynamic super-network, and the method comprises the steps: collecting resting-state fMRI brain image data of a plurality of normal and healthy subjects and subjects suffering from mild cognitive impairment diseases, carrying out the preprocessing operation, and obtaining an ROI time sequence of the data of each subject; clustering each tested ROI time sequence by using an affine clustering algorithm, and adaptively grouping each tested brain region; constructing a dynamic brain network; performing feature selection to obtain different clustering coefficients; performing feature screening on the plurality of training objects according to different clustering coefficients; and finally, establishing a corresponding classification model. The sliding time window method is introduced to construct the brain network, so that a dynamic network model implied by the brain can be mined according to the time-varying ...
Подробнее