ResUNet++: An Advanced Architecture for Medical Image Segmentation

Accurate computer-aided polyp detection and segmentation during colonoscopy examinations can help endoscopists resect abnormal tissue and thereby decrease chances of polyps growing into cancer. Towards developing a fully automated model for pixel-wise polyp segmentation, we propose ResUNet++, which is an improved ResUNet architecture for colonoscopic image segmentation. Our experimental evaluations show that the suggested architecture produces good segmentation results on publicly available datasets. Furthermore, ResUNet++ significantly outperforms U-Net and ResUNet, two key state-of-the-art deep learning architectures, by achieving high evaluation scores with a dice coefficient of 81.33%, and a mean Intersection over Union (mIoU) of 79.27% for the Kvasir-SEG dataset and a dice coefficient of 79.55%, and a mIoU of 79.62% with CVC-612 dataset.

PDF Abstract
Task Dataset Model Metric Name Metric Value Global Rank Result Benchmark
Medical Image Segmentation ASU-Mayo Clinic dataset ResUNet++ mIoU 0.8569 # 1
Recall 0.6534 # 1
Precision 0.4896 # 1
DSC 0.8743 # 1
Medical Image Segmentation CVC-ClinicDB ResUNet++ mean Dice 0.7955 # 34
Medical Image Segmentation CVC-VideoClinicDB ResUNet++ Dice 0.8798 # 3
mIoU 0.8730 # 2
Recall 0.7749 # 1
precision 0.6702 # 2
Medical Image Segmentation ETIS-LARIBPOLYPDB ResUNet++ mIoU 0.7534 # 2
mean Dice 0.6364 # 16
Medical Image Segmentation KvasirCapsule-SEG ResUNet+ DSC 0.9499 # 2
mIoU 0.9087 # 1
Medical Image Segmentation Kvasir-SEG ResUNet++ mean Dice 0.8133 # 44
Polyp Segmentation Kvasir-SEG ResUNet++ DSC 0.8133 # 2
mIoU 0.7927 # 7
Colorectal Polyps Characterization Kvasir-Sessile dataset ResUNet++ + TTA DSC 0.5042 # 1

Methods