SkinCNN: Classification of Skin Cancer Lesions with A Novel CNN Model

Abstract

Recently, there has been an increase in the number of cancer cases due to causes such as physical inactivity, sun exposure, environmental changes, harmful drinks and viruses. One of the most common types of cancer in the general population is skin cancer. There is an increase in exposure to the sun's harmful rays due to reasons such as environmental changes, especially ozone depletion. As exposure increases, skin changes occur in various parts of the body, especially the head and neck, in both young and old. In general, changes such as swelling in skin lesions are diagnosed as skin cancer. Skin cancers that are frequently seen in the society are known as actinic keratosis (akiec), basal cell carcinoma (bcc), bening keratosis (bkl), dermatofibroma (df), melanoma (mel), melanocytic nevi (nv), and vascular (vasc) types. It is not possible to consider all possible skin changes as skin cancer. In such a case, the development of a decision support system that can automatically classify the specified skin cancer images will help specialized healthcare professionals. For these purposes, a basic model based on MobileNet V3 was developed using the swish activation function instead of the ReLU activation function of the MobileNet architecture. In addition, a new CNN model with a different convolutional layer is proposed for skin cancer classification, which is different from the studies in the literature. The proposed CNN model (SkinCNN) achieved a 97% success rate by performing the training process 30 times faster than the pre-trained MobileNet V3 model. In both models, training, validation and test data were modelled by partitioning according to the value of cross-validation 3. MobileNet V3 model achieved F1 score, recall, precision, and accuracy metrics of 0.87, 0.88, 0.84, 0.83, 0.84, and 0.83, respectively, in skin cancer classification. The SkinCNN obtained F1 score, recall, precision, and accuracy metrics of 0.98, 0.97, 0.96, and 0.97, respectively. With the obtained performance metrics, the SkinCNN is competitive with the studies in the literature. In future studies, since the SkinCNN is fast and lightweight, it can be targeted to run on real-time systems.

Keywords

• MobileNet V3
• CNN
• Skin cancer
• Transfer learning
• Classification

References

1. [1] R. Perroy, “World population prospects,” United Nations, vol. 1, no. 6042, pp. 587–592, 2015.
2. [2] D. Pimentel et al., “Ecology of Increasing Diseases: Population Growth and Environmental Degradation,” Hum. Ecol. Interdiscip. J., vol. 35, no. 6, pp. 653–668, 2007, doi: 10.1007/s10745-007-9128-3.
3. [3] M. Plummer, C. de Martel, J. Vignat, J. Ferlay, F. Bray, and S. Franceschi, “Global burden of cancers attributable to infections in 2012: a synthetic analysis,” Lancet Glob. Heal., vol. 4, no. 9, pp. e609–e616, 2016.
4. [4] H. Younis, M. H. Bhatti, and M. Azeem, “Classification of Skin Cancer Dermoscopy Images using Transfer Learning,” in 2019 15th International Conference on Emerging Technologies (ICET), Dec. 2019, pp. 1–4. doi: 10.1109/ICET48972.2019.8994508.
5. [5] U.-O. Dorj, K.-K. Lee, J.-Y. Choi, and M. Lee, “The skin cancer classification using deep convolutional neural network,” Multimed. Tools Appl., vol. 77, no. 8, pp. 9909–9924, 2018, doi: 10.1007/s11042-018-5714-1.
6. [6] A. J. McMichael and T. McMichael, Planetary overload: global environmental change and the health of the human species. Cambridge University Press, 1993.
7. [7] P. Martens and A. J. McMichael, Environmental change, climate and health: issues and research methods. Cambridge University Press, 2009.
8. [8] R. L. McKenzie, L. O. Björn, A. Bais, and M. Ilyasd, “Changes in biologically active ultraviolet radiation reaching the Earth’s surface,” Photochem. Photobiol. Sci., vol. 2, no. 1, pp. 5–15, 2003.

9. [9] D. M. Parkin, D. Mesher, and P. Sasieni, “13. Cancers attributable to solar (ultraviolet) radiation exposure in the UK in 2010,” Br. J. Cancer, vol. 105, no. 2, pp. S66–S69, 2011.
10. [10] R. J. Hay et al., “The global burden of skin disease in 2010: an analysis of the prevalence and impact of skin conditions,” J. Invest. Dermatol., vol. 134, no. 6, pp. 1527–1534, 2014.
11. [11] D. R. Bickers et al., “The burden of skin diseases: 2004: A joint project of the American Academy of Dermatology Association and the Society for Investigative Dermatology,” J. Am. Acad. Dermatol., vol. 55, no. 3, pp. 490–500, 2006.
12. [12] M. A. Morid, A. Borjali, and G. Del Fiol, “A scoping review of transfer learning research on medical image analysis using ImageNet,” Comput. Biol. Med., vol. 128, p. 104115, 2021, doi: https://doi.org/10.1016/j.compbiomed.2020.104115.
13. [13] F. Nachbar et al., “The ABCD rule of dermatoscopy: high prospective value in the diagnosis of doubtful melanocytic skin lesions,” J. Am. Acad. Dermatol., vol. 30, no. 4, pp. 551–559, 1994.
14. [14] D. B. Mendes and N. C. da Silva, “Skin lesions classification using convolutional neural networks in clinical images,” arXiv Prepr. arXiv1812.02316, 2018.
15. [15] C. Barata, M. Ruela, M. Francisco, T. Mendonça, and J. S. Marques, “Two systems for the detection of melanomas in dermoscopy images using texture and color features,” IEEE Syst. J., vol. 8, no. 3, pp. 965–979, 2013.
16. [16] I. Giotis, N. Molders, S. Land, M. Biehl, M. F. Jonkman, and N. Petkov, “MED-NODE: A computer-assisted melanoma diagnosis system using non-dermoscopic images,” Expert Syst. Appl., vol. 42, no. 19, pp. 6578–6585, 2015.
17. [17] D. Ruiz, V. Berenguer, A. Soriano, and B. Sánchez, “A decision support system for the diagnosis of melanoma: A comparative approach,” Expert Syst. Appl., vol. 38, no. 12, pp. 15217–15223, 2011.
18. [18] V. Pomponiu, H. Nejati, and N.-M. Cheung, “Deepmole: Deep neural networks for skin mole lesion classification,” in 2016 IEEE International Conference on Image Processing (ICIP), 2016, pp. 2623–2627. doi: 10.1109/ICIP.2016.7532834.
19. [19] D. A. Shoieb, S. M. Youssef, and W. M. Aly, “Computer-Aided Model for Skin Diagnosis Using Deep Learning,” J. Image Graph., vol. 4, no. 2, pp. 122–129, 2016, doi: 10.18178/joig.4.2.122-129.
20. [20] H. Çetiner, “Cataract disease classification from fundus images with transfer learning based deep learning model on two ocular disease datasets,” Gümüşhane Üniversitesi Fen Bilim. Enstitüsü Derg., vol. 13, no. 2, pp. 258–269, Jan. 2023, doi: 10.17714/gumusfenbil.1168842.
21. [21] H. Çetiner and B. Kara, “Recurrent Neural Network Based Model Development for Wheat Yield Forecasting,” J. Eng. Sci. Adiyaman Univ., vol. 9, no. 16, pp. 204–218, 2022, doi: 10.54365/adyumbd.1075265.
22. [22] A. Krizhevsky, I. Sutskever, and G. E. Hinton, “Imagenet classification with deep convolutional neural networks,” Adv. Neural Inf. Process. Syst., vol. 25, 2012.
23. [23] P. Tschandl, C. Rosendahl, and H. Kittler, “The HAM10000 dataset, a large collection of multi-source dermatoscopic images of common pigmented skin lesions,” Sci. data, vol. 5, no. 1, pp. 1–9, 2018.
24. [24] K. M. Hosny, M. A. Kassem, and M. M. Fouad, “Classification of Skin Lesions into Seven Classes Using Transfer Learning with AlexNet,” J. Digit. Imaging, vol. 33, no. 5, pp. 1325–1334, 2020, doi: 10.1007/s10278-020-00371-9.
25. [25] W. Bao, X. Yang, D. Liang, G. Hu, and X. Yang, “Lightweight convolutional neural network model for field wheat ear disease identification,” Comput. Electron. Agric., vol. 189, p. 106367, 2021, doi: https://doi.org/10.1016/j.compag.2021.106367.
26. [26] A. G. Howard et al., “Mobilenets: Efficient convolutional neural networks for mobile vision applications,” arXiv Prepr. arXiv1704.04861, 2017.
27. [27] M. Sandler, A. Howard, M. Zhu, A. Zhmoginov, and L.-C. Chen, “Mobilenetv2: Inverted residuals and linear bottlenecks,” in Proceedings of the IEEE conference on computer vision and pattern recognition, 2018, pp. 4510–4520.
28. [28] S. Qian, C. Ning, and Y. Hu, “MobileNetV3 for Image Classification,” in 2021 IEEE 2nd International Conference on Big Data, Artificial Intelligence and Internet of Things Engineering (ICBAIE), 2021, pp. 490–497. doi: 10.1109/ICBAIE52039.2021.9389905.
29. [29] S.-R.-S. Jianu, L. Ichim, D. Popescu, and O. Chenaru, “Advanced Processing Techniques for Detection and Classification of Skin Lesions,” in 2018 22nd International Conference on System Theory, Control and Computing (ICSTCC), Oct. 2018, pp. 498–503. doi: 10.1109/ICSTCC.2018.8540732.
30. [30] J. Kawahara, A. BenTaieb, and G. Hamarneh, “Deep features to classify skin lesions,” in 2016 IEEE 13th international symposium on biomedical imaging (ISBI), 2016, pp. 1397–1400. doi: 10.1109/ISBI.2016.7493528.
31. [31] Y. Li and L. Shen, “Skin lesion analysis towards melanoma detection using deep learning network,” Sensors, vol. 18, no. 2, p. 556, 2018.
32. [32] N. Codella, J. Cai, M. Abedini, R. Garnavi, A. Halpern, and J. R. Smith, “Deep Learning, Sparse Coding, and SVM for Melanoma Recognition in Dermoscopy Images,” in International workshop on machine learning in medical imaging, Springer, 2015, pp. 118–126. doi: 10.1007/978-3-319-24888-2_15.