European Radiology: resting state fMRI machine learning to predict individualized hand movements in patients with PRC cortical tumors

Accurate brain function area mapping plays a vital role in preoperative planning.

Yang Wang

The purpose of this study is to use resting state functional magnetic resonance imaging (rs-fMRI) data to evaluate the effectiveness of neural network (NN) methods in predicting preoperative motor zone positioning.

Methods 109 patients with craniocerebral tumors underwent rs-fMRI and fMRI (TbfMRI) scan based on hand movement tasks.

_{NN_Act ICA_Act}

Schematic diagram of the processing flow of NN-ML prediction model and ICA motion network extraction

NN-ML prediction model and ICA motion extraction process flow schematic network NN-ML prediction model and ICA motion extraction process flow schematic network

First, use the resting state data of 98 human connection group project subjects for group ICA to obtain 32 groups of characteristics.

First, use the resting state data of 98 human connection group project subjects for group ICA to obtain 32 groups of characteristics.

Neural network (NN) machine learning process flowchart

Neural network (NN) machine learning process flowchart Neural network (NN) machine learning process flowchart

A neural network based on individual resting fMRI (RsfMRI) features and a healthy control group task fMRI activation map was used to train 20 models.

A neural network based on individual resting fMRI (RsfMRI) features and a healthy control group task fMRI activation map was used to train 20 models.

Compared with the MAP derived by ICA, the CC matrix of the MAP predicted by the neural network shows a higher diagonal value (p<0.

Regardless of whether there is movement disorder, DC _{NN_Act is} higher than DC _{ICA_Act} (P<0.

The neural network method can predict individual motor activation based on rs-fMRI data, and has a good clinical application prospect in patients with brain tumor anatomy and functional reconstruction.

The neural network method can predict individual motor activation based on rs-fMRI data, and has a good clinical application prospect in patients with brain tumor anatomy and functional reconstruction.

Original source

springer.

springer.

Leave a message here