IMPROVED DEEP LEARNING MODEL-BASED SKIN DISEASE IMAGE SEGMENTATION ALGORITHM
Keywords:
Deep learning, Skin disease image segmentation, Coordinate attention mechanism, Spatial perception module, Feature fusionAbstract
Aiming at the problems of complex background interference, low segmentation accuracy of small targets, blurred lesion boundaries and large scale differences in skin disease image segmentation, this paper proposes two improved deep learning models, namely the CAU-Net model fused with the coordinate attention mechanism and the CSwinTransU-Net model integrated with the spatial perception module. Experimental results on the ISIC2018 skin disease dataset show that the mean Intersection over Union (mIoU), mean Dice coefficient (mDice) and average accuracy (aAcc) of the CAU-Net model reach 85.50%, 92.06% and 93.69% respectively, and the corresponding indicators of the CSwinTransU-Net model are 84.84%, 91.66% and 93.49%. Both models outperform the basic models, effectively improving the accuracy and robustness of skin disease image segmentation, and providing technical support for the early diagnosis of skin cancer.References
[1] Qiu Y, Li Q Q, Wang Y. Application of Deep Learning Technology in Medical Image Segmentation. Computer Knowledge and Technology, 2022, 18(10): 74-75.
[2] Gu M J. Research on Intelligent Segmentation and Recognition of Skin Lesions Based on Deep Learning. Nanjing University of Posts and Telecommunications, 2023.
[3] Wang A L. Research on Melanoma Image Segmentation Based on Deep Learning. Lanzhou University of Technology, 2023.
[4] Song P F. Research on Medical Image Segmentation Algorithm Based on Convolutional Neural Network and Transformer. Shandong Technology and Business University, 2023.
[5] Sun Z L. Research on Semi-Supervised Learning Based Segmentation Algorithm for Skin Disease Medical Images. Jilin University, 2023.
[6] Yin W, Zhou D M, Fan T, et al. A Skin Lesion Image Segmentation Method Based on Dense Atrous Spatial Pyramid Pooling and Attention Mechanism. Journal of Biomedical Engineering, 2022, 39(06): 1108-1116.
[7] Liang L M, Zhou L S, Yin J, et al. Skin Lesion Segmentation Algorithm Fused with Multi-Scale Transformer. Journal of Jilin University (Engineering and Technology Edition), 2024, 54(04): 1086-1098.
[8] Chen Y X. Research on Segmentation and Classification of Skin Disease Images Based on Deep Learning. Jiangxi University of Science and Technology, 2022.
[9] Shelhamer E, Long J, Darrell T. Fully Convolutional Networks for Semantic Segmentation. IEEE Transactions on Pattern Analysis and Machine Intelligence, 2017, 39(4): 640-651.
[10] Valanarasu J M J, Patel V M. UNeXt: MLP-based Rapid Medical Image Segmentation Network. Proceedings of the 25th International Conference on Medical Image Computing and Computer Assisted Intervention– MICCAI 2022. MICCAI 2022. Lecture Notes in Computer Science, 2022, 13435: 23-33. DOI: /10.1007/978-3-031-16443-9_3.
[11] Bozorgpour A, Sadegheih Y, Kazerouni A, et al. DermoSegDiff: A Boundary-Aware Segmentation Diffusion Model for Skin Lesion Delineation. PRIME 2023. Lecture Notes in Computer Science, 2023, 14277: 146-158. DOI: 10.1007/978-3-031-46005-0_13.
[12] Hu J, Shen L, Sun G. Squeeze-and-Excitation Networks. Squeeze-and-Excitation Networks, 2018: 7132-7141. DOI: 10.1109/CVPR.2018.00745.