Skip to main content
. 2023 Sep 2;13:14433. doi: 10.1038/s41598-023-41359-z

Figure 1.

Figure 1

Sample characteristics, distribution of public schizophrenia MRI datasets and the preprocessing pipeline. (A) Acquisition parameters of the T1W MRI scans and the patient demographic information of each dataset. In the BrainGluSchi, COBRE, and NMorphCH datasets, normal scans consisted of whole head structural T1W MR images obtained from healthy control subjects and schizophrenia scans consisted of whole head structural T1W MR images obtained from schizophrenia and schizoaffective disorder patients. (B) Data preprocessing pipeline to generate the input. For each structural MRI, we process the T1W 3D volume through a standardized pipeline consisting of three steps: (1) whole head T1W affine registration to the MNI152 template space, (2) skull stripping, and (3) whole brain affine registration to the MNI152 template space. The preprocessing of structural T1W MR data is necessary to remove unwanted artifacts and transform the data into a standard format before training the deep learning models.