Presented at the Cognitive Models and Artificial Intelligence Conference (BMYZ2023), Ankara, Türkiye, Oct 26, 2023

SETSCI Conference Proceedings, 2023, 15, Page (s): 8-12 , https://doi.org/10.36287/setsci.6.1.006

Published Date: 29 December 2023    | 1220     4

Skin cancer is the abnormal growth of skin cells and one of the most common cancers of all. There are several types of skin cancer. Melanoma, a form of skin cancer, has increased 237% in Turkey in the last 30 years. While the U.S. adds about one million new cases of melanoma each year, this rate in Turkey is 1.9 per 100 thousand men and 1.3 per 100 thousand women. The death rate from skin cancer is 1 in 100 patients worldwide. As with all cancers, early diagnosis of this cancer is crucial, and artificial intelligence (AI) appears to be a promising technology for detecting early-stage skin cancer from dermoscopic images in recent years. AI-based studies for skin cancer classification are usually performed using three different types of images:
dermoscopic images, clinical images, and histopathological images. In this study, a new deep learning model called CNN-BM (Convolutional Neural Network-Based Model) is proposed for skin cancer diagnosis using dermoscopic images. In this context, HAM10000, a commonly employed public dataset consisting of 10015 dermoscopic images, is utilized. The proposed model not only increases the success rate in the training process, but also reduces the execution time. CNN-BM consists of convolution, max pooling, dropout, flatten, and activation layers. Relu and sigmoid functions are chosen as activation functions. CNNs are sensitive to the batch size values, which significantly affects the quality of the model. Unlike other deep learning models used in the literature for skin cancer diagnosis, the proposed model uses a small batch size to prevent overfitting and increase the regularization effect. Similarly, by incorporating a dropout layer and dense-sparse-dense training techniques into the model, overfitting is avoided and the success rate of the network is increased. To determine the most efficient values for the hyperparameters, a trial-and-error method is employed. Research findings indicate that the success of the model is superior to other studies in the literature with 86.48% accuracy and 85.13% precision rates.

Keywords - skin cancer, cancer detection, CNN, CNN-BM, deep learning

Koçak, Y. (2021), Cilt (Deri) Kanseri Belirtileri, Tedavisi ve Korunma Yöntemleri, Memorial, 1-1.

Goyal, M., et al. (2020), Artificial Intelligence-based Image Classification Methods for Diagnosis of Skin Cancer: Challenges and Opportunities, Computers in Biology and Medicine,. 127, 104065.

Society, A.C. (2018), Cancer Facts & Figures 2018, American Cancer Society, 1-1.

Nikolaou, V. and A.J.B.j.o.d. Stratigos (2014), Emerging Trends in the Epidemiology of Melanoma, Journal of Emerging Technologies and Innovative Research, 170(1), 11-19.

Society, A.C. (2008), Cancer Facts & Figures 2008, American Cancer Society, 1-1.

Haenssle, H.A., et al. (2018), Man Against Machine: Diagnostic Performance of a Deep Learning Convolutional Neural Network for Dermoscopic Melanoma Recognition in Comparison to 58 Dermatologists, Annals of oncology, 29(8), 1836-1842.

Tschandl, P., et al. (2019), Expert-level Diagnosis of Nonpigmented Skin Cancer by Combined Convolutional Neural Networks, JAMA dermatology, 155(1), 58-65.

Codella, N.C., et al. (2017), Deep Learning Ensembles for Melanoma Recognition in Dermoscopy Images, IBM Journal of Research and Development, 61(4/5), 1-5.

Codella, N., et al. (2018), Skin Lesion Analysis Toward Melanoma Detection 2018: A Challenge Hosted by the International Skin Imaging Collaboration, ISIC, 1-1.

Tschandl, P., C. Rosendahl, and H. Kittler (2018), The HAM10000 Dataset, a Large Collection of Multi-Source Dermatoscopic Images of Common Pigmented Skin Lesions, Scientific Data, 5(1), 180161.

Kousis, I.; Perikos, I.; Hatzilygeroudis, I.; Virvou, M. (2022), Deep Learning Methods for Accurate Skin Cancer Recognition and Mobile Application, Electronics, 11, 1294.

Al-Masni MA, Kim DH, Kim TS. (2020), Multiple Skin Lesions Diagnostics via Integrated Deep Convolutional Networks for Segmentation and Classification, Comput Methods Programs Biomed, 1-1.

Esteva, A., Kuprel, B., Novoa, R. et al. (2017), Dermatologist-level Classification of Skin Cancer with Deep Neural Networks, Nature, 542, 115–118.

Han, Song & Pool, Jeff & Narang, Sharan & Mao, Huizi & Tang, Shijian & Elsen, Erich & Catanzaro, Bryan & Tran, John & Dally, William (2016), DSD: Regularizing Deep Neural Networks with Dense-Sparse-Dense Training Flow, 1-1.

Al-Bander, Baidaa; Yas, Qahtan M.; Mahdi, Hussain; and Al-Hamd, Rwayda KH. S. (2021) "Benchmarking of deep learning algorithms for skin cancer detection based on ahybrid framework of entropy and VIKOR techniques," Turkish Journal of Electrical Engineering and Computer Sciences: Vol. 29: No. 8, Article 4.

Kandel, I., & Castelli, M. (2020). The effect of batch size on the generalizability of the convolutional neural networks on a histopathology dataset. ICT Express, 6(4), 312–315.