A mobile application that uses deep learning to analyze skin lesion images can detect monkeypox with 91 percent accuracy, according to a new artificial intelligence (AI) research study. | Tech Rasta

[ad_1]

Recently monkeypox is spreading all over the world. As of August 2022, approximately 47,000 laboratory-confirmed cases of monkeypox have been reported worldwide.

It is difficult to diagnose monkeypox clinically in its early stages because monkeys are very similar to chickenpox and measles. PCR tests can be used to confirm detection, but PCR tests are not readily available. Therefore computational methods for detection of monkeypox will help in easier and faster detection of monkeypox in its early stages.

Although human monkeypox disease dates back to ancient times (1970s), computer vision-based studies for preliminary disease diagnosis have only recently begun. Currently there are very few studies on it. So computer intervention is required in this regard.

Smartphones are becoming increasingly important in healthcare monitoring and delivery. There are many benefits of using mobile healthcare applications. First, they can help people get information about their diseases quickly. Second, their therapists can follow them. Third, they can help people manage their diseases. Fourth, they can help people track their progress. Fifth, they can help motivate people.

A recently published study presents an Android mobile application that uses deep pre-trained networks to aid in the classification of human monkeypox from skin lesion images. The app collects video images through the mobile device camera. These images are then sent to a deep convolutional neural network running on the same device. The network then classifies the images as positive or negative to detect monkeypox. Images of skin lesions on people with monkeypox and other skin lesions were used to train the network.

A publicly accessible dataset and a deep transfer learning strategy were applied for this aim. The network classifies images as positive or negative to detect monkeypox. Images of skin lesions on people with monkeypox and other skin lesions were used to train the network. A publicly accessible dataset and a deep transfer learning strategy were applied for this aim. The whole training and testing process was done in Matlab with different pre-trained networks. The best-performing network was replicated and trained using TensorFlow. By switching to the TensorFlow Lite model, the TensorFlow model has been made suitable for mobile devices. A TensorFlow Lite model and library for Monkeypox detection are integrated into the mobile application.

The application was run on three different devices and inference times were collected during runtime. The averages of inference times are 197ms, 91ms and 138ms. Test results show that the system can accurately classify 91.11% of photos. A prescribed smartphone app can also be trained to make a basic diagnosis of other skin conditions.

People with physical injuries can easily determine a primary diagnosis using the system provided. Therefore, monkeypox patients may require immediate medical attention for a definitive diagnosis.

The method proposed in this research is a good solution to detect monkeypox because it is faster than clinical detection, more reliable and more readily available than PCR tests. This method can be extended to detect more skin diseases, making the process of diagnosing all skin diseases easier, faster and more reliable.

This Article is written as a research summary article by Marktechpost Staff based on the research paper 'Human Monkeypox Classification from Skin Lesion Images with Deep Pre-trained Network using Mobile Application'. All Credit For This Research Goes To Researchers on This Project. Check out the paper.
Please Don't Forget To Join Our ML Subreddit


Rishab Jain, Consulting Intern at MarkTechPost. Currently studying B.Tech in Computer Science at IIIT, Hyderabad. He is a machine learning enthusiast and has a keen interest in artificial intelligence and statistical methods in data analytics. He is passionate about developing better algorithms for AI.


[ad_2]

Source link