BrainNet: Optimal Deep Learning Feature Fusion for Brain Tumor Classification

Table 5

A comparative study of the proposed methodologies on the BraTS2018 dataset.

Research papers	Maximum achieved accuracy (%)	Maximum achieved execution-time speedup
PSO features + softmax [42]	92.50	NA
PLS features + ELM [43]	93.40	NA
Two-channel DNN [44]	93.69	NA
MANet [26]	94.91	NA
Proposed	96.7	25.5x

Bold represents the best values.