KiU-Net: Overcomplete Convolutional Architectures for Biomedical Image and Volumetric Segmentation

Most methods for medical image segmentation use U-Net or its variants as they have been successful in most of the applications. After a detailed analysis of these "traditional" encoder-decoder based approaches, we observed that they perform poorly in detecting smaller structures and are unable to segment boundary regions precisely... (read more)

PDF Abstract

Results from the Paper


TASK DATASET MODEL METRIC NAME METRIC VALUE GLOBAL RANK RESULT BENCHMARK
Ultrasound Brain Anatomy US KiU-Net Dice 89.43 # 1
Liver Segmentation LiTS2017 KiU-Net 3D IoU 89.46 # 1
Dice 94.23 # 1
Medical Image Segmentation RITE KiU-Net Dice 75.17 # 1
Jaccard Index 60.37 # 1

Methods used in the Paper


METHOD TYPE
ReLU
Activation Functions
Max Pooling
Pooling Operations
3D Convolution
Convolutions
Convolution
Convolutions
Concatenated Skip Connection
Skip Connections
U-Net
Semantic Segmentation Models