Multi-time window resting state functional magnetic resonance image data feature analysis method

• Main Authors: HE ZILONG, WANG JINGYI, HAO XIAOKE, MA MINGMING, WANG XIAOFANG, ZHOU CHAOFEI, SHI SHUO
• Format: Patent
• Language: Chinese; English
• Published: 12.04.2024
• Subjects: CALCULATING; COMPUTING; COUNTING; PHYSICS

The invention relates to a multi-time-window resting state functional magnetic resonance image data feature analysis method, which comprises the following steps of: for all training samples, dividing a time sequence of resting state functional magnetic resonance image data into a plurality of continuous time windows, and carrying out multi-time-window joint feature selection by using sparse consistency learning. According to the method, an L1 norm is used for carrying out sparse constraint on feature weights in a single time window, and an L21 norm is used for carrying out consistency constraint on feature weights of all time windows. Besides, according to a learned feature weight matrix, a sample-time window-brain region-brain connected four-order tensor is constructed, then tensor decomposition is used to obtain factor matrixes of the four dimensions, and finally a multi-core support vector machine is used to carry out fusion classification on the selected features. According to the method, the factor matri