Segmenting Medical Images: From UNet to Res-UNet and nnUNet
Authors:
Lina Huang,
Alina Miron,
Kate Hone,
Yongmin Li
Abstract:
This study provides a comparative analysis of deep learning models including UNet, Res-UNet, Attention Res-UNet, and nnUNet, and evaluates their performance in brain tumour, polyp, and multi-class heart segmentation tasks. The analysis focuses on precision, accuracy, recall, Dice Similarity Coefficient (DSC), and Intersection over Union (IoU) to assess their clinical applicability. In brain tumour…
▽ More
This study provides a comparative analysis of deep learning models including UNet, Res-UNet, Attention Res-UNet, and nnUNet, and evaluates their performance in brain tumour, polyp, and multi-class heart segmentation tasks. The analysis focuses on precision, accuracy, recall, Dice Similarity Coefficient (DSC), and Intersection over Union (IoU) to assess their clinical applicability. In brain tumour segmentation, Res-UNet and nnUNet significantly outperformed UNet, with Res-UNet leading in DSC and IoU scores, indicating superior accuracy in tumour delineation. Meanwhile, nnUNet excelled in recall and accuracy, which are crucial for reliable tumour detection in clinical diagnosis and planning. In polyp detection, nnUNet was the most effective, achieving the highest metrics across all categories and proving itself as a reliable diagnostic tool in endoscopy. In the complex task of heart segmentation, Res-UNet and Attention Res-UNet were outstanding in delineating the left ventricle, with Res-UNet also leading in right ventricle segmentation. nnUNet was unmatched in myocardium segmentation, achieving top scores in precision, recall, DSC, and IoU. The conclusion notes that although Res-UNet occasionally outperforms nnUNet in specific metrics, the differences are quite small. Moreover, nnUNet consistently shows superior overall performance across the experiments. Particularly noted for its high recall and accuracy, which are crucial in clinical settings to minimize misdiagnosis and ensure timely treatment, nnUNet's robust performance in crucial metrics across all tested categories establishes it as the most effective model for these varied and complex segmentation tasks.
△ Less
Submitted 5 July, 2024;
originally announced July 2024.
