Application of deep learning in detecting neurological disorders from magnetic resonance images: a survey on the detection of Alzheimer’s disease, Parkinson’s disease and schizophrenia

Brain Informatics

Table 3 Summary of DL-based studies for prediction and classification of PD from [s]-MRI

Ref.	Regions	DL Tech.	Pre-Proc.	Feature	Dataset	Size	Accuracy
[75]	Axial	CNN-RNN	–	CBFd	NTUA	55 PD, 23 PD Synd	98%
[57]	Sagittal, coronal, axial planes	3D-CNN	SST, DA	CNN based, age, sex	PPMI	452 PD, 204 HC	100%
[76]	Mild brain	CNN		CBF	NIMHANS	45 PD, 20 APS, 35 HC	80%\(^{5\alpha }\)
[77]	Lentiform nucleus	CNN-RNN		CNN based	NTUA	66176	98%
[58]	Whole brain	CNN	NM, F, SM	CBF	PPMI	100 PD, 82 HC	88.9%
[45]	Basal ganglia, mesencephalon	CNN	AC, BR, SN, SM	CNN based	PPMI		Control vs PD 94.5-96%, PD vs SWEDD 88.7%

Pre-Proc. pre-processing, Synd syndrome, \(n\alpha\)– n fold cross-validation, AC alignment correction, SWEDD scans without evidence for dopaminergic deficit, CBF CNN-based features