Lesion segmentation using U-Net network
This work addresses a domain-specific problem in medical imaging for skin lesion analysis, but it is incremental as it applies an existing method with minor modifications.
The paper tackled skin lesion segmentation for melanoma detection by training a U-Net network with adjustments to the loss function for class imbalance and post-processing for contour refinement, achieving results in the ISIC 2018 challenge.
This paper explains the method used in the segmentation challenge (Task 1) in the International Skin Imaging Collaboration's (ISIC) Skin Lesion Analysis Towards Melanoma Detection challenge held in 2018. We have trained a U-Net network to perform the segmentation. The key elements for the training were first to adjust the loss function to incorporate unbalanced proportion of background and second to perform post-processing operation to adjust the contour of the prediction.