Deep Learning Approach for Medical Image Analysis

Table 1

Segmentation performance (%) of the proposed model compared with previous models on medical images.

Authors	Method	Percentage of accuracy
Proposed system (enhanced UNET)	Skin lesions, brain MRI, and retina images	Above 90% mean accuracy and dice coefficient
Zhou et al.’s [21] CNN	Multiple organ detection	79% mean accuracy
Moeskops et al.’s [22] CNN	Brain MRI	Overall DSC 73.53%
Xu et al.’s [23] CNN	Skin, fibroglandular tissue, mass, and fatty tissue	80% accuracy
Roth et al.’s [24] FCN (3D U-Net)	Arteries, portal vein, liver, spleen, stomach, gallbladder, and pancreas	DICE score (89.3 ± 6.5) % during testing