Automatic skin disease diagnosis using deep learning from clinical image and patient information

Abstract

Background: Skin diseases are the fourth most common cause of human illness which results in enormous non-fatal burden in daily life activities. They are caused by chemical, physical and biological factors. Visual assessment in combination with clinical information is the common diagnostic procedure for diseases. However, these procedures are manual, time-consuming, and require experience and excellent visual perception. Objectives: In this study, an automated system is proposed for the diagnosis of five common skin diseases by using data from clinical images and patient information using deep learning pre-trained mobilenet-v2 model. Methods: Clinical images were acquired using different smartphone cameras and patient's information were collected during patient registration. Different data preprocessing and augmentation techniques were applied to boost the performance of the model prior to training. Results: A multiclass classification accuracy of 97.5%, sensitivity of 97.7% and precision of 97.7% has been achieved using the proposed technique for the common five skin disease. The results demonstrate that, the developed system provides excellent diagnosis performance for the five skin diseases. Conclusion: The system has been designed as a smartphone application and it has the potential to be used as a decision support system in low resource settings, where both the expert dermatologist and the means are limited.