Studies in differentiating psoriasis from other dermatoses using small data set and transfer learning

EURASIP Journal on Image and Video Processing

Research

Anna Woźniacka¹  Sebastian Patrzyk¹  Mariusz Nieniewski²  Leszek J. Chmielewski³ 

[1] Department of Dermatology and Venereology, Medical University of Lodz, plac Hallera 1, 90-647, Lodz, Poland;Faculty of Mathematics and Informatics, University of Lodz, ul. Banacha 22, 90-238, Lodz, Poland;Institute of Information Technology, Warsaw University of Life Sciences, SGGW, ul. Nowoursynowska 159, 02-775, Warsaw, Poland;

关键词: Psoriasis;    Convolutional neural networks;    Transfer learning;    Papulosquamous skin diseases;    Deep learning;   

DOI  :  10.1186/s13640-023-00607-y

 received in 2022-06-02, accepted in 2023-03-30,  发布年份 2023

来源: Springer

【 摘 要 】

Psoriasis is a common skin disorder that should be differentiated from other dermatoses if an effective treatment has to be applied. Regions of Interests, or scans for short, of diseased skin are processed by the VGG16 (or VGG19) deep convolutional neural network operating as a feature extractor. 1280 features related to a given scan are passed to the Support Vector Machine (SVM) classifier using Radial Basis Functions (RBF) kernels. The main quality of the described setup is a very small number of 75 psoriasis patients and 75 non-psoriasis patients used in the teaching and testing sets taken together. For each patient, a variable number of clinical images are taken. Then, the scans of size 256×256\documentclass[12pt]{minimal}\usepackage{amsmath}\usepackage{wasysym}\usepackage{amsfonts}\usepackage{amssymb}\usepackage{amsbsy}\usepackage{mathrsfs}\usepackage{upgreek}\setlength{\oddsidemargin}{-69pt}\begin{document}$$256 \times 256$$\end{document} pixels are cropped from these images. There are 1988 scans of psoriasis patients and 1582 of non-psoriasis patients. The other quality of the described setup is the use of transfer learning for carrying over the neural network’s weights from non-medical domain (ImageNet) to clinical images of dermatoses. The next quality is that the input images are obtained with smart phone cameras without any special arrangements or equipment, so there is a great variability in working conditions, which hampers discriminative power of the classifier. The primary classification is carried out on individual scans, and then, majority voting is executed among the scans pertaining to an individual patient. The obtained recall (sensitivity) is 85.33%, and the precision is 82.58%. The 95% confidence interval for the accuracy of 80.08% is [77.14, 83.04]%. These numbers indicate that the described system can be useful for remote diagnosing of psoriasis, particularly in areas where access to dermatological personnel is limited.

【 授权许可】

CC BY   
© The Author(s) 2023

【 图 表 】

41116_2023_36_Article_IEq431.gif

41116_2023_36_Article_IEq430.gif

41116_2023_36_Article_IEq429.gif

Fig. 2

41116_2023_36_Article_IEq427.gif

41116_2023_36_Article_IEq426.gif

41116_2023_36_Article_IEq425.gif

Fig. 1

41116_2023_36_Article_IEq423.gif

41116_2023_36_Article_IEq422.gif

41116_2023_36_Article_IEq421.gif

41116_2023_36_Article_IEq420.gif

41116_2023_36_Article_IEq419.gif

41116_2023_36_Article_IEq418.gif

41116_2023_36_Article_IEq417.gif

41116_2023_36_Article_IEq416.gif

41116_2023_36_Article_IEq414.gif

41116_2023_36_Article_IEq413.gif

41116_2023_36_Article_IEq412.gif

41116_2023_36_Article_IEq408.gif

41116_2023_36_Article_IEq407.gif

41116_2023_36_Article_IEq404.gif

Fig. 3

41116_2023_36_Article_IEq401.gif

Fig. 2

41116_2023_36_Article_IEq399.gif

41116_2023_36_Article_IEq398.gif

41116_2023_36_Article_IEq397.gif

41116_2023_36_Article_IEq396.gif

Fig. 1

41116_2023_36_Article_IEq394.gif

41116_2023_36_Article_IEq393.gif

41116_2023_36_Article_IEq392.gif

41116_2023_36_Article_IEq391.gif

41116_2023_36_Article_IEq390.gif

41116_2023_36_Article_IEq389.gif

41116_2023_36_Article_IEq388.gif

41116_2023_36_Article_IEq387.gif

41116_2023_36_Article_IEq386.gif

41116_2023_36_Article_IEq385.gif

41116_2023_36_Article_IEq384.gif

41116_2023_36_Article_IEq383.gif

41116_2023_36_Article_IEq382.gif

Fig. 1

41116_2023_36_Article_IEq380.gif

41116_2023_36_Article_IEq378.gif

12938_2023_1111_Article_IEq24.gif

41116_2023_36_Article_IEq377.gif

41116_2023_36_Article_IEq375.gif

41116_2023_36_Article_IEq368.gif

41116_2023_36_Article_IEq365.gif

41116_2023_36_Article_IEq362.gif

Fig. 1

41116_2023_36_Article_IEq353.gif

41116_2023_36_Article_IEq351.gif

Fig. 3

41116_2023_36_Article_IEq347.gif

41116_2023_36_Article_IEq344.gif

41116_2023_36_Article_IEq341.gif