TABLE 2.

Performance evaluation of the state‐of‐the‐art methods and the proposed method using the BRATS database

Segmentation Method	Dice score (%)	Details
Pereira 2016 [7]	88	Patch‐based segmentation net with a reduced filter kernel
Havaie 2017 [3]	88	A multi‐scale CNN for patch‐based segmentation
Kamnitsas 2017 [20]	90.1	3D multi‐scale CNN for patch‐based segmentation with optimized softmax layer
Chang 2019 [21]	80	CNN combined with a fully connected CRF as a mixture model to introduce the global context information
Ding SRNET 2019 [14]	83	Stack multi‐connection simple reducing‐net (SMCSRNet)
Khan 2020 [22]	81	Handcrafted features including LBP and HOG are combined with CNN to achieve pixel classification
Alkassar et al. 2019 [10]	89	VGG16 is utilized for brain tumour segmentation along with transfer learning
Proposed (LBTS‐Net16)	91	Lightweight VGG‐16 with reduced number of convolution filters and depth‐wise convolution
Proposed (LBTS‐Net19)	91.5	Lightweight VGG‐19 with reduced number of convolution filters and depth‐wise convolution