Cilt Hastalıklarının Tanısında Ön İşlemli Görüntü Dönüştürücüler (Vision Transformers) ve Klasik Sınıflandırıcılar

Öz

Görüntü Dönüştürücüler (ViT'ler), tıpta son teknoloji derin öğrenme teknolojisidir. ViT'ler çok sayıda parametre gerektirir, bu nedenle öğrenme için nispeten büyük bir veri kümesine ihtiyaç duyarlar. Bu, şu anda sağlık hizmetlerinin dijitalleştirilmesi nedeniyle mümkündür. Karşılaştırma olarak, nispeten düşük sayıda giriş verisiyle karakterize edilen klasik sınıflandırıcıları da kullanıyoruz. Klinik uygulamada, dermoskopi, konfokal mikroskopi, yansıma konfokal mikroskopi ve Raman spektroskopisi gibi yüksek çözünürlüklü görüntüler cilt hastalıklarını teşhis etmek için kullanılır. ViT'lerin klinik uygulamada potansiyeli vardır. Modelin evrişimli sinir ağlarına göre avantajı, evrişimli işlemler kullanmamasıdır. Bir veri kümesinden önceden işlenmiş görüntüler, çeşitli boyutlarda beş ViT modeli ve ilgili klasik sınıflandırıcılar kullanılarak deneysel olarak sınıflandırıldı. Başka bir veri kümesinden önceden işlenmiş dermatoskopik görüntüler üzerinde de karşılaştırmalı deneyler yürütüldü. Bu makale, çeşitli cilt rahatsızlıklarını tanımlamak için bir yapay zeka yöntemini tanıtmaktadır. Veri seti, 5 kategoriye ayrılmış görüntüler içerir: normal, melanom, arsenik, sedef hastalığı ve egzama. Çalışma sırasında, cilt görüntüleri, önemli ayrıntıları ortaya çıkarmak için kontrastı artıran Uyarlamalı Histogram Eşitleme (AHE) tekniği kullanılarak ilk işleme tabi tutuldu. Bu ön işlemenin ardından, karmaşık görsel bilgileri yakalama yetenekleriyle ünlü ViT'ler kullanılarak görüntülerden özellikler elde edildi. Çıkarılan bu özellikler daha sonra geleneksel makine öğrenimi sınıflandırıcılarıyla birlikte kullanıldı ve incelenen cilt rahatsızlıklarının doğru bir şekilde teşhis edilmesiyle sonuçlandı. Bulgular, tıbbi görüntü sınıflandırmasıyla ilgili görevlerde ViT'leri klasik sınıflandırıcılarla birleştirmenin etkinliğini vurgulamaktadır.

Anahtar Kelimeler

• Cilt Hastalıkları
• Önişleme
• Uyarlanabilir Histogram Eşitleme
• Sınıflandırma
• Görme Dönüştürücüler

Preprocessed Vision Transformers and Classical Classifiers in Diagnosing Skin Diseases

Abstract

Vision Transformers (ViTs) are the state-of-the-art deep learning technology in medicine. ViTs require a large number of parameters, so they need a relatively large dataset for learning. This is currently possible due to the digitization of healthcare. As a comparison, we also use classical classifiers, which are characterized by a relatively low number of input data. In clinical practice, high-resolution images such as those from dermoscopy, confocal microscopy, reflectance confocal microscopy, and Raman spectroscopy are used to diagnose skin diseases. ViTs have potential in clinical practice. The advantage of the model over convolutional neural networks is that they do not use convolutional operations. Preprocessed images from a dataset were classified experimentally using five ViTs models of various sizes and respective classical classifiers. Comparative experiments were conducted also on preprocessed dermatoscopic images from another dataset. This article introduces an artificial intelligence method for identifying various skin conditions. The dataset contains images that are classified into 5 categories: normal, melanoma, arsenic, psoriasis, and eczema. During the study, skin images underwent initial processing using the Adaptive Histogram Equalization (AHE) technique, which enhanced the contrast to reveal important details. Following this preprocessing, features were obtained from the images using ViTs, renowned for their ability to capture intricate visual information. These extracted features were then utilized in conjunction with traditional machine learning classifiers, resulting in accurate diagnosis of the skin conditions being studied. The findings emphasize the effectiveness of combining ViTs with classical classifiers in tasks related to medical image classification.

Keywords

• Skin Diseases
• Preprocessing
• Adaptive Histogram Equalization
• Classification
• Vision Transformers

Kaynakça

1. [1] M. A. Richard, C. Paul, T. Nijsten, P. Gisondi, C. Salavastru, C. Taieb, ... & EADV Burden of Skin Diseases Project Team. “Prevalence of most common skin diseases in Europe: a population‐based study,” Journal of the European Academy of Dermatology and Venereology, vol. 36, no. 7, pp. 1088-1096, 2022.
2. [2] E. T. Anwar, N. Gupta, O. Porwal, A. Sharma, R. Malviya, A. Singh & N. K. Fuloria, “Skin diseases and their treatment strategies in sub-saharan african regions,” Infectious Disorders-Drug Targets Disorders), vol. 22, no. 2, pp. 41-54, 2022.
3. [3] J. A. Rossow, F. Queiroz-Telles, D. H. Caceres, K. D. Beer, “A One Health Approach to Combatting Sporothrix brasiliensis: Narrative Review of an Emerging Zoonotic Fungal Pathogen in South America,” Journal of Fungi, vol. 6, no. 4, p. 247, 2020.
4. [4] D. M. Elston, “Occupational skin disease among health care workers during the coronavirus (COVID-19) epidemic,” Journal of the American Academy of Dermatology, vol. 82, no. 5, p. 1085, 2020.
5. [5] V. R. Balaji, S. T. Suganthi, R. Rajadevi, V. K. Kumar, “Skin disease detection and segmentation using dynamic graph cut algorithm and classification through Naive Bayes classifier,” Measurement, vol. 163, p. 107922, 2020.
6. [6] Y. Liu, A. Jain, C. Eng, D. H. Way, K. Lee, P. Bui, et al., “A deep learning system for differential diagnosis of skin diseases,” Nature medicine, vol. 26, no. 6, pp. 900-908, 2020.
7. [7] S. Inthiyaz, B. R. Altahan, S. H. Ahammad, V. Rajesh, R. R. Kalangi, L. K. Smirani and A. N. Z. Rashed, “Skin disease detection using deep learning,” Advances in Engineering Software, vol. 175, p. 103361, 2023.
8. [8] N. Hameed, A. M. Shabut, and M. K. Ghosh, “Multi-class multi-level classification algorithm for skin lesions classification using machine learning techniques,” Expert Systems with Applications, vol. 141, p. 112961, 2020.

9. [9] N. Melnyk, I. Vlasova, W. Skowrońska, and A. Bazylko, “Current knowledge on interactions of plant materials traditionally used in skin diseases in Poland and Ukraine with human skin microbiota,” International Journal of Molecular Sciences, vol. 23, no. 17, p. 9644, 2022.
10. [10] H. Li, Y. Pan, J. Zhao, and L. Zhang, “Skin disease diagnosis with deep learning: A review,” Neurocomputing, vol. 464, pp. 364-393, 2021.
11. [11] SS Han, I Park, SE Chang, W Lim, MS Kim, “Intelligence dermatology: deep neural networks empower medical professionals in diagnosing skin cancer and predicting treatment options for 134 skin,” Dermatology, vol. 140, no. 9, pp. 1753-1761, 2020.
12. [12] P.N. Srinivasu, J.G. SivaSai, M.F. Ijaz, A.K. Bhoi, and W. Kim, “Classification of skin disease using deep learning neural networks with MobileNet V2 and LSTM,” Sensors, vol. 21, no. 8, p. 2852, 2021.
13. [13] N. Ghaffar Nia, E. Kaplanoglu, and A. Nasab, “Evaluation of artificial intelligence techniques in disease diagnosis and prediction,” Discover Artificial Intelligence, vol. 3, no. 1, p. 5, 2023.
14. [14] K. Das, C. J. Cockerell, A. Patil, and P. Pietkiewicz, “Machine learning and its application in skin cancer,” International Journal of Environmental Research and Public Health, vol. 18, no. 24, p. 13409, 2021.
15. [15] J. Jayashree, S. S. V. Nukala, and J. Vijayashree, “The rise of AI in the field of healthcare,” Cognitive Machine Intelligence, pp. 221-244, 2024.
16. [16] A. Mubeen and U. N. Dulhare, “Metaheuristic Algorithms for the Classification and Prediction of Skin Lesions: A Comprehensive Review,” Machine Learning and Metaheuristics: Methods and Analysis, pp. 107-137, 2023.
17. [17] A. Gomolin, E. Netchiporouk, and R. Gniadecki, “Artificial intelligence applications in dermatology: where do we stand?,” Frontiers in medicine, vol. 7, p. 100, 2020.
18. [18] T. B. Jutzi, E. I. Krieghoff-Henning, and T. Holland-Letz, “Artificial intelligence in skin cancer diagnostics: the patients' perspective,” Frontiers in medicine, vol. 7, p. 233, 2020.
19. [19] J. Kawahara and G. Hamarneh, “Visual Diagnosis of Dermatological Disorders: Human and Machine Performance,” arXiv preprint arXiv:1906.01256, 2019.
20. [20] S. Mishra, S. Chaudhury, H. Imaizumi, and T. Yamasaki, “Assessing Robustness of Deep learning Methods in Dermatological Workflow,” arXiv preprint arXiv:2001.05878, 2020.
21. [21] S. Chan, V. Reddy, B. Myers, Q. Thibodeaux et al., "Machine Learning in Dermatology: Current Applications, Opportunities, and Limitations,” Dermatology and therapy, vol. 10, pp. 365-386. 2020.
22. [22] D. Sonntag, F. Nunnari, and H. J. Profitlich, “The Skincare project, an interactive deep learning system for differential diagnosis of malignant skin lesions. Technical Report,” arXiv preprint arXiv:2005.09448, 2020.
23. [23] G. J. Chowdary, “Machine Learning and Deep Learning Methods for Building Intelligent Systems in Medicine and Drug Discovery: A Comprehensive Survey,” arXiv preprint arXiv:2107.14037, 2021.
24. [24] M. Qays Hatem, “Skin lesion classification system using a K-nearest neighbor algorithm,” Visual Computing for Industry, Biomedicine, and Art, vol. 5, no. 1, p. 7, 2022.
25. [25] Z. Li, K. Christoph Koban, T. Ludwig Schenck, R. Enzo Giunta et al., “Artificial Intelligence in Dermatology Image Analysis: Current Developments and Future Trends,” Journal of clinical medicine, vol. 11, no. 22, p. 6826, 2022.
26. [26] S. Khan, H. Ali, and Z. Shah, “Identifying the role of vision transformer for skin cancer—A scoping review,” Frontiers in Artificial Intelligence, vol. 6, 1202990, 2023.
27. [27] T. Lai, “Interpretable Medical Imagery Diagnosis with Self-Attentive Transformers: A Review of Explainable AI for Health Care,” BioMedInformatics, vol. 4, no. 1, pp. 113-126. 2023.
28. [28] A. Khan, Z. Rauf, A. Rehman Khan, S. Rathore et al., “A Recent Survey of Vision Transformers for Medical Image Segmentation,” arXiv preprint arXiv:2312.00634, 2023.
29. [29] V. Ravi, T. J. Alahmadi, T. Stephan, P. Singh and M. Diwakar, “DEEPSCAN: Integrating Vision Transformers for Advanced Skin Lesion Diagnostics,” The Open Dermatology Journal, vol. 18, no. 1, 2024.
30. [30] Q. Abbas, Y. Daadaa, U. Rashid and M. E. Ibrahim, “Assist-dermo: A lightweight separable vision transformer model for multiclass skin lesion classification,” Diagnostics, vol. 13, no. 15, p. 2531, 2023.
31. [31] E. G. Espinosa, J. S. R. Castilla and F. G. Lamont, F. G. “Skin Disease Pre-diagnosis with Novel Visual Transformers,” In Workshop on Engineering Applications (pp. 103-113). Cham: Springer Nature Switzerland, 2024.
32. [32] S. Aladhadh, M. Alsanea, M. Aloraini, T. Khan, S. Habib and M. Islam, “An effective skin cancer classification mechanism via medical vision transformer,” Sensors, vol. 22, no11, p. 4008, 2022.
33. [33] https://www.kaggle.com/datasets/sayedhossainjoba yer/skin-diseases-identification, Last Accessed On: 30.10.2024.
34. [34] A. Dosovitskiy, “An image is worth 16x16 words: Transformers for image recognition at scale,” arXiv preprint arXiv: 2010.11929, 2020.
35. [35] K. Han, Y. Wang, H. Chen, X., Chen, J. Guo, Z. Liu & D. Tao, “A survey on visual transformer,” arXiv preprint arXiv:2012.12556, 2020.
36. [36] M. M. Naseer, K. Ranasinghe, S. H. Khan, “Intriguing properties of vision transformers,” Advances in Neural Information Processing Systems, vol. 34, pp. 23296-23308, 2021.
37. [37] Y. H. Cao, H. Yu, and J. Wu, “Training Vision Transformers with Only 2040 Images,” European Conference on Computer Vision (pp. 220-237), 2022.
38. [38] A. Parvaiz, M. Anwaar Khalid, R. Zafar, H. Ameer et al., “Vision Transformers in Medical Computer Vision - A Contemplative Retrospection,” Engineering Applications of Artificial Intelligence, vol. pp. 122, 2022.
39. [39] S. Jelassi, M. E. Sander, and Y. Li, “Vision Transformers provably learn spatial structure,” Advances in Neural Information Processing Systems, vol. 35, pp. 37822-37836, 2022.
40. [40] V. A. Nguyen, K. Pham Dinh, L. Tung Vuong, T. T. Do et al., “Vision Transformer Visualization: What Neurons Tell and How Neurons Behave?,” arXiv preprint arXiv:2210.07646, 2022.
41. [41] S. A. Khan, S. Hussain, and S. Yang, “Contrast enhancement of low-contrast medical images using modified contrast limited adaptive histogram equalization,” Journal of Medical Imaging and Health Informatics, vol. 10, no. 8, pp. 1795-1803. 2020. [42] V. Banupriya and A. Kalaivani, “Improved retinal fundus image quality with hybrid image filter and enhanced contrast limited adaptive histogram equalization,” International Journal of Health Sciences, vol. I, pp. 9234-9246, 2022.