SchizoGoogLeNet: The GoogLeNet-Based Deep Feature Extraction Design for Automatic Detection of Schizophrenia

Table 6

The comparison of the proposed method with other methods for the same database.

Authors	Methods	ACC (%)	SEN (%)	SPE (%)
Khare et al. [35]	Empirical wavelet transformation with SVM	88.70	91.13	89.29
Siuly et al. [2]	EMD-based features with EBT	89.59	89.76	89.32
Guo et al. [38]	ERP features with RF	81.10	NA	NA
Khare and Bajaj [37]	F-TQWT-based scheme	91.39	92.65	93.22
Guo et al. [38]	Electrical marker with CNN	92.00	NA	NA
Khare et al. [38]	RVMD-based OELM method	92.93	97.15	91.06
Khare and Bajaj [39]	SPWVD-based TFR and CNN model	93.36	94.25	92.03
Proposed method	GoogLeNet-based deep features with an SVM model	98.84	99.02	98.58

NA = not available.Bold values represent the highest performance.