Skin Lesion Segmentation Based on Improved U-net

Melanoma is one of the most common and dangerous skin cancers, accounting for 75% of deaths associated with skin cancer. Detection of melanoma in early stages can significantly improve the survival rate. Automatic segmentation of melanoma is an important and essential step for accurate detection of melanoma. Many existing works based on traditional segmentation methods and deep learning methods have been proposed for high-resolution dermoscopy images. However, due to the intrinsic visual complexity and ambiguity among different skin conditions, automatic melanoma segmentation is still a challenging task for existing methods. Among these methods, the deep learning methods have obtained more attention recently due to its high performance by training an end-to-end framework, which needs no human interaction. U-net is a very popular deep learning model for medical image segmentation. In this paper, we propose an efficient skin lesion segmentation based on improved U-net model. Experiments conducted on the 2017 ISIC Challenge dataset towards melanoma detection shows that the proposed method can obtain state-of-the-art performance on skin lesion segmentation task.

[1]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[2]  Jian Sun,et al.  Delving Deep into Rectifiers: Surpassing Human-Level Performance on ImageNet Classification , 2015, 2015 IEEE International Conference on Computer Vision (ICCV).

[3]  Vladlen Koltun,et al.  Multi-Scale Context Aggregation by Dilated Convolutions , 2015, ICLR.

[4]  Ming Chao,et al.  Improving Dermoscopic Image Segmentation With Enhanced Convolutional-Deconvolutional Networks , 2017, IEEE Journal of Biomedical and Health Informatics.

[5]  A. Jemal,et al.  Cancer statistics, 2019 , 2019, CA: a cancer journal for clinicians.

[6]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[7]  Hao Chang Skin cancer reorganization and classification with deep neural network , 2017, ArXiv.

[8]  Scott Schaefer,et al.  Image deformation using moving least squares , 2006, ACM Trans. Graph..

[9]  Qaisar Abbas,et al.  An Improved Segmentation Method for Non-melanoma Skin Lesions Using Active Contour Model , 2014, ICIAR.

[10]  Matt Berseth,et al.  ISIC 2017 - Skin Lesion Analysis Towards Melanoma Detection , 2017, ArXiv.

[11]  Abder-Rahman Ali,et al.  Melanoma detection using fuzzy C-means clustering coupled with mathematical morphology , 2014, 2014 14th International Conference on Hybrid Intelligent Systems.

[12]  Roberto Cipolla,et al.  SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[13]  Mohammad H. Jafari,et al.  Skin lesion segmentation in clinical images using deep learning , 2016, 2016 23rd International Conference on Pattern Recognition (ICPR).

[14]  Mohammad H. Jafari,et al.  Set of descriptors for skin cancer diagnosis using non-dermoscopic color images , 2016, 2016 IEEE International Conference on Image Processing (ICIP).

[15]  Jinshan Tang A multi-direction GVF snake for the segmentation of skin cancer images , 2009, Pattern Recognit..

[16]  Noel C. F. Codella,et al.  Skin lesion analysis toward melanoma detection: A challenge at the 2017 International symposium on biomedical imaging (ISBI), hosted by the international skin imaging collaboration (ISIC) , 2016, 2018 IEEE 15th International Symposium on Biomedical Imaging (ISBI 2018).