Abstract
Automated classification of skin lesions in dermoscopy images has the potential to significantly improve survival rates and reduce the risk of death for skin cancer patients. However, existing supervised learning models heavily depend on well-annotated dermoscopy training data, which is expensive and labor-intensive to obtain. This paper addresses this issue by proposing a semi-supervised framework called Noisy Consistent Pseudo Labeling (NCPL), which only utilizes less annotated images with many unlabeled raw data. The NCPL framework consists of two components: the Noisy-Consistent Sample Learning(NCSL) module to remove low-confidence images, and the Attentive Clustered Feature Integration (ACFI) module, incorporating an uncertainty-aware attention mechanism. Specifically, the NCSL module is introduced to filter and generate reliable pseudo-labels for unlabeled skin images, with excellent capability of removing noisy samples. Additionally, the ACFI module integrates high-dimensional representations of original lesion images in an attentive manner, assisted with the annotated data. By focusing the representative samples and removing noisy images, the NCPL approach performs outstanding experimental results, demonstrating the superiority of the NCPL framework in semi-supervised skin lesion classification task.
Q. Li—Supported by organization x.
Access this chapter
Tax calculation will be finalised at checkout
Purchases are for personal use only
References
Perez, M., Abisaad, J.A., Rojas, K.D., Marchetti, M.A., Jaimes, N.: Skin cancer: primary, secondary, and tertiary prevention. Part I. J. Am. Acad. Dermatol. 87(2), 255–268 (2022)
Gregoor, A.M.S., et al.: An artificial intelligence based app for skin cancer detection evaluated in a population based setting. NPJ Digital Med. 6(1), 90 (2023)
Yinhao, W., Chen, B., Zeng, A., Pan, D., Wang, R., Zhao, S.: Skin cancer classification with deep learning: a systematic review. Front. Oncol. 12, 893972 (2022)
Zhou, Y., et al.: Multi-site cross-organ calibrated deep learning (MUSCLD): automated diagnosis of non-melanoma skin cancer. Med. Image Anal. 84, 102702 (2023)
Nakai, K., Chen, Y.-W., Han, X.-H.: Enhanced deep bottleneck transformer model for skin lesion classification. Biomed. Signal Process. Control 78, 103997 (2022)
Mishra, S., Zhang, Y., Zhang, L., Zhang, T., Hu, X.S., Chen, D.Z.: Data-driven deep supervision for skin lesion classification. In: Wang, L., Dou, Q., Fletcher, P.T., Speidel, S., Li, S. (eds.) MICCAI 2022. LNCS, vol. 13431, pp. 721–731. Springer, Cham (2022). https://doi.org/10.1007/978-3-031-16431-6_68
Wang, L., Zhang, L., Shu, X., Yi, Z.: Intra-class consistency and inter-class discrimination feature learning for automatic skin lesion classification. Med. Image Anal. 85, 102746 (2023)
Lie, J., et al.: Flexible sampling for long-tailed skin lesion classification. In: Wang, L., Dou, Q., Fletcher, P.T., Speidel, S., Li, S. (eds.) MICCAI 2022. LNCS, vol. 13433, pp. 462–471. Springer, Cham (2022). https://doi.org/10.1007/978-3-031-16437-8_44
He, X., Tan, E.-L., Bi, H., Zhang, X., Zhao, S., Lei, B.: Fully transformer network for skin lesion analysis. Med. Image Anal. 77, 102357 (2022)
Song, L., Wang, H., Wang, Z.J.: Decoupling multi-task causality for improved skin lesion segmentation and classification. Pattern Recognition 133, 108995 (2023)
Wang, Y., Wang, Y., Cai, J., Lee, T.K., Miao, C., Wang, Z.J.: SSD-KD: a self-supervised diverse knowledge distillation method for lightweight skin lesion classification using dermoscopic images. Med. Image Anal. 84, 102693 (2023)
Zhou, S., et al.: FixMatch-LS: semi-supervised skin lesion classification with label smoothing. Biomed. Signal Process. Control 84, 104709 (2023)
Gao, Z., et al.: A semi-supervised multi-task learning framework for cancer classification with weak annotation in whole-slide images. Med. Image Anal. 83, 102652 (2023)
Zhou, S., et al.: ReFixMatch-LS: reusing pseudo-labels for semi-supervised skin lesion classification. Med. Biol. Eng. Comput. 61(5), 1033–1045 (2023)
Zeng, L., et al.: Advancements in nanoparticle-based treatment approaches for skin cancer therapy. Mol. Can. 22(1), 10 (2023)
Tschandl, P., Rosendahl, C., Kittler, H.: The ham10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions. Sci. Data 5(1), 1–9 (2018)
Chang, C.H., Wang, W.E., Hsu, F.Y., Chen, R.J., Chang, H.C.: AI HAM 10000 database to assist residents in learning differential diagnosis of skin cancer. In: 2022 IEEE 5th Eurasian Conference on Educational Innovation (ECEI), pp. 1–3. IEEE (2022)
He, K., Zhang, X., Ren, S., Sun, J.: Deep residual learning for image recognition. In: Proceedings of the IEEE Conference on Computer Vision and Pattern Recognition, pp. 770–778 (2016)
Bdair, T., Navab, N., Albarqouni, S.: FedPerl: semi-supervised peer learning for skin lesion classification. In: de Bruijne, M., et al. (eds.) MICCAI 2021. LNCS, vol. 12903, pp. 336–346. Springer, Cham (2021). https://doi.org/10.1007/978-3-030-87199-4_32
Van der Maaten, L., Hinton, G.: Visualizing data using t-SNE. J. Mach. Learni. Res. 9(11), 2579–2605 (2008)
Jung, H., Oh, Y.: Towards better explanations of class activation mapping. In: Proceedings of the IEEE/CVF International Conference on Computer Vision, pp. 1336–1344 (2021)
Author information
Authors and Affiliations
Corresponding author
Editor information
Editors and Affiliations
Rights and permissions
Copyright information
© 2023 The Author(s), under exclusive license to Springer Nature Switzerland AG
About this paper
Cite this paper
Zhu, Q., Li, S., Li, Z., Min, X., Li, Q. (2023). Noisy-Consistent Pseudo Labeling Model for Semi-supervised Skin Lesion Classification. In: Woo, J., et al. Medical Image Computing and Computer Assisted Intervention – MICCAI 2023 Workshops. MICCAI 2023. Lecture Notes in Computer Science, vol 14394. Springer, Cham. https://doi.org/10.1007/978-3-031-47425-5_22
Download citation
DOI: https://doi.org/10.1007/978-3-031-47425-5_22
Published:
Publisher Name: Springer, Cham
Print ISBN: 978-3-031-47424-8
Online ISBN: 978-3-031-47425-5
eBook Packages: Computer ScienceComputer Science (R0)