Acnes Detection System Using Aiot Technologies

Abstract

Acne vulgaris is a common skin condition that can lead to serious consequences in severe cases. Traditionally, patients need to visit specialized healthcare facilities for examination and treatment, which can impact their time and work, especially considering the overload often experienced by medical systems. Therefore, machinery needs to be involved as an assistant in the phase of acne diagnosis. Many image analysis algorithms have been developed using images captured by mobile devices. Nevertheless, most of these algorithms are mainly based on antiquated features such as color-model or texture-based, which may result in poor performance when confronted with the intricate nature of acne lesions. Consequently, artificial intelligence (AI) models have been developed to alleviate these challenges for patients and reduce the burden on healthcare systems. However, due to the scarcity of high-quality acne image data, some of these models have not yet achieved optimal results. In an attempt to overcome these limitations, this paper proposes the ACNE10, an AIoT system implemented from the combination of IoT and AI, to accurately identify primary lesions, secondary lesions, severe clinical forms, and differential diagnoses of acne commonly encountered in clinical settings. The AI model is trained on a dataset containing 9440 labeled images encompassing various acne lesions and diagnostic differentiations. The result shows that the model achieved state-of-the-art performance with a mAP score of 0.774 across the total 12 types of acne lesions. The accuracy of detecting each type of acne object is impressively high and balanced between the classes despite the imbalance in the dataset caused by the unequal number of images in each acne category. The system is deployed using three different approaches: web application, Zalo Mini App, and IoT-based. With this study, ACNE10 is expected to provide medical support in the acne diagnosis processes, as well as the acne patient to understand their conditions for better treatment.