Performance Comparison of Machine Learning-Based Classification of Skin Diseases from Skin Lesion Images
- 31 Downloads
Skin is one of the main parts of the human body. At the same time, skin will be easily infected and damaged by various kinds of skin diseases. Skin disease is a major health hazard across the globe. Nowadays, many people are suffering from skin diseases. It is tedious and time consuming for doctors to manually diagnose them. Recently, machine learning techniques have been successful in the detection and recognition of different types of objects in the images which have been applied to recognize various types of diseases from the medical images. Various machine learning techniques have been used to recognize and classify skin diseases from the images. Here, three machine learning techniques support vector machine (SVM), VGGNet and Inception-ResNet-v2 have been implemented to classify seven types of skin diseases from skin lesion images. Performance of these models has been evaluated and compared by using precision and recall values. Inception-ResNet-v2 has been found to be superior based on the classification performance among these three models.
KeywordsSkin lesion VGG16 SVM Inception-ResNet-v2
This research is funded by the Information and Communication Technology (ICT), Division of Ministry of Posts, Telecommunications and Information Technology, Government of People’s Republic of Bangladesh.
- 2.Azmi, N.F.M., Sarkan, F.M., Haslina, M.S., Yazriwati, Y., Suriayati, C.: ABCD rules segmentation on malignant tumor and benign skin lesion images. In: 3rd International Conference on Computer and Information Sciences (ICCOINS), pp. 66–70. IEEE (2016) Google Scholar
- 3.Alessandro, B.D., Atam, P.D., Nizar, M.: Computer-aided analysis of epi-illumination and transillumination images of skin lesions for diagnosis of skin cancers. In: Annual International Conference on Engineering in Medicine and Biology Society (EMBC), pp. 3434–3438. IEEE (2011)Google Scholar
- 8.Maria, M.V.J., Rosado, L., Ferreira, M.: A new risk assessment methodology for dermoscopic skin lesion images. In: International Symposium on Medical Measurements and Applications (MeMeA), pp. 570–575. IEEE (2015)Google Scholar
- 9.Szegedy, C., Loffe, S., Vanhoucke, V., Alemi, A.A.: Inception-v4, inception-resnet and the impact of residual connections on learning. In AAAI, 4, p. 12 (2017)Google Scholar