Revisiting the Shape-Bias of Deep Learning for Dermoscopic Skin Lesion Classification

被引:2
|
作者
Lucieri, Adriano [1 ,2 ]
Schmeisser, Fabian [1 ]
Balada, Christoph Peter [2 ]
Siddiqui, Shoaib Ahmed [2 ]
Dengel, Andreas [1 ,2 ]
Ahmed, Sheraz [2 ]
机构
[1] Tech Univ Kaiserslautern, Dept Comp Sci, Erwin Schrodinger Str 52, D-67663 Kaiserslautern, Germany
[2] German Res Ctr Artificial Intelligence GmbH DKFI, Smart Data & Knowledge Serv SDS, Trippstadter Str, D-67663 Kaiserslautern, Germany
关键词
Dermatology; Digital dermatoscopy; Skin lesion analysis; Spectral analysis; Robustness; Deep learning; ABCD RULE; COLOR; DERMATOSCOPY; RECOGNITION; MELANOMAS; DIAGNOSIS; IMAGES;
D O I
10.1007/978-3-031-12053-4_4
中图分类号
TP39 [计算机的应用];
学科分类号
081203 ; 0835 ;
摘要
It is generally believed that the human visual system is biased towards the recognition of shapes rather than textures. This assumption has led to a growing body of work aiming to align deep models' decision-making processes with the fundamental properties of human vision. The reliance on shape features is primarily expected to improve the robustness of these models under covariate shift. In this paper, we revisit the significance of shape-biases for the classification of skin lesion images. Our analysis shows that different skin lesion datasets exhibit varying biases towards individual image features. Interestingly, despite deep feature extractors being inclined towards learning entangled features for skin lesion classification, individual features can still be decoded from this entangled representation. This indicates that these features are still represented in the learnt embedding spaces of the models, but not used for classification. In addition, the spectral analysis of different datasets shows that in contrast to common visual recognition, dermoscopic skin lesion classification, by nature, is reliant on complex feature combinations beyond shape-bias. As a natural consequence, shifting away from the prevalent desire of shape-biasing models can even improve skin lesion classifiers in some cases.
引用
收藏
页码:46 / 61
页数:16
相关论文
共 50 条
  • [1] Deep Ensemble Learning for Skin Lesion Classification from Dermoscopic Images
    Shahin, Ahmed H.
    Kamal, Ahmed
    Elattar, Mustafa A.
    [J]. 2018 9TH CAIRO INTERNATIONAL BIOMEDICAL ENGINEERING CONFERENCE (CIBEC), 2018, : 150 - 153
  • [2] SKIN LESION CLASSIFICATION FROM DERMOSCOPIC IMAGES USING DEEP LEARNING TECHNIQUES
    Lopez, Adria Romero
    Giro-i-Nieto, Xavier
    Burdick, Jack
    Marques, Oge
    [J]. 2017 13TH IASTED INTERNATIONAL CONFERENCE ON BIOMEDICAL ENGINEERING (BIOMED), 2017, : 49 - 54
  • [3] A customized deep learning framework for skin lesion classification using dermoscopic images
    Sahoo, Sandhya Rani
    Dash, Ratnakar
    Mohapatra, Ramesh Kumar
    [J]. COMPUTER ANIMATION AND VIRTUAL WORLDS, 2023, 34 (05)
  • [4] Deep Learning-Based Dermoscopic Image Classification System for Robust Skin Lesion Analysis
    Thamizhamuthu, Rajamanickam
    Maniraj, Subramanian Pitchiah
    [J]. TRAITEMENT DU SIGNAL, 2023, 40 (03) : 1145 - 1152
  • [5] Deep Learning Approaches Towards Skin Lesion Segmentation and Classification from Dermoscopic Images - A Review
    Baig, Ramsha
    Bibi, Maryam
    Hamid, Anmol
    Kausar, Sumaira
    Khalid, Shahzad
    [J]. CURRENT MEDICAL IMAGING, 2020, 16 (05) : 513 - 533
  • [6] Skin Lesion Segmentation in Dermoscopic Images With Ensemble Deep Learning Methods
    Goyal, Manu
    Oakley, Amanda
    Bansal, Priyanka
    Dancey, Darren
    Yap, Moi Hoon
    [J]. IEEE ACCESS, 2020, 8 : 4171 - 4181
  • [7] Radiomic and deep learning analysis of dermoscopic images for skin lesion pattern decoding
    Wang, Zheng
    Wang, Chong
    Peng, Li
    Lin, Kaibin
    Xue, Yang
    Chen, Xiao
    Bao, Linlin
    Liu, Chao
    Zhang, Jianglin
    Xie, Yang
    [J]. SCIENTIFIC REPORTS, 2024, 14 (01):
  • [8] Classification of Skin Lesion Images with Deep Learning Approaches
    Bayram, Buket
    Kulavuz, Bahadir
    Ertugrul, Berkay
    Bayram, Bulent
    Bakirman, Tolga
    Cakar, Tuna
    Dogan, Metehan
    [J]. BALTIC JOURNAL OF MODERN COMPUTING, 2022, 10 (02): : 241 - 250
  • [9] Multimodal skin lesion classification using deep learning
    Yap, Jordan
    Yolland, William
    Tschandl, Philipp
    [J]. EXPERIMENTAL DERMATOLOGY, 2018, 27 (11) : 1261 - 1267
  • [10] Melanoma Is Skin Deep: A 3D Reconstruction Technique for Computerized Dermoscopic Skin Lesion Classification
    Satheesha, T. Y.
    Satyanarayana, D.
    Prasad, M. N. Giri
    Dhruve, Kashyap D.
    [J]. IEEE JOURNAL OF TRANSLATIONAL ENGINEERING IN HEALTH AND MEDICINE, 2017, 5