다중 작업 학습 기반의 피부 병변 분류 방법

Abstract

Skin lesions have a high misdiagnosis rate due to a wide variety of forms. Recently, a deep learning based skin lesion classification method is difficult to classify due to hair and fuzzy boundaries of skin lesions. In this paper, we propose a network for classifying skin lesions and segmenting skin lesion regions using a multitask learning method. Experimentally, the result shows that the performance of our method has been improved by 2.48 % over the previous method.