MLP-UNet: Glomerulus Segmentation

Glomerulus segmentation in kidney tissue segments is a crucial nephropathology process used to diagnose renal diseases effectively. This study proposes novel and robust application of MLP (Multi-Layer Perceptron) based architectures for the glomeruli PAS (Periodic AcidSchiff) stained whole images effective diagnosis diseases. For challenge, proposed unique solution uses MLP-UNet Perceptron U-Net), design that evades using conventional convolution self-attention mechanisms. Additionally, compares various approaches, including U-Net, first time, trains TransUNet model on WSI (Whole Slide Image) dataset. Dice Score Loss were training these models as metric loss function. Results showed MLP-based provide comparable results (89.96%) pre-trained like (90.58%) with effectively 20% lesser parameters no pre-training, also produce superior scores across 5-fold cross-validation learn more quickly than U-Net architectures.