| Title |
Segmentation of Skin Cancer Lesions using ResUNet++ |
| Authors |
박준영(June-Young Park) ; 한영환(Young-Hwan Han) |
| DOI |
https://doi.org/10.5573/ieie.2022.59.2.95 |
| Keywords |
Deep learning; Semantic segmentation; Medical image; AI |
| Abstract |
In this paper, we propose a method of segmentation lesions using deep learning in skin cancer medical images. The segmentation model solved the segmentation problem using ResUNet++, which improved the loss function and optimization method. ResUNet++ is an improved version of U-Net model that has been mainly used to solve the problem of existing medical image segmentation. The ISIC Challenge dataset has the data necessary to segment the lesion. Therefore, a deep learning model is learned using the corresponding dataset. As a result of the experiment, the Dice Coefficient score, an evaluation index of the split model, was 0.9357, showing good results. |