FPT Software’s experts have invented an AI algorithm to support the diagnosis of pneumothorax, and can be applied to detect lung-related diseases such as pneumonia, sarcoidosis, cystic fibrosis, Pulmonary tuberculosis, lung cancer, and idiopathic pulmonary fibrosis.

In 2019, FSOFTers invented the AI ​​algorithm (Residual Capsule), thereby building an AI model to help doctors detect pneumothorax.

From over 10.000 2D chest X-ray images, the AI algorithm conducts the evaluation, classification, and identification of images showing the risk of pneumothorax, which contributes to the rapid diagnosis or increases the diagnosis accuracy. According to Director of FPT Software’s Technology Development Division Tran Duc Tri Quang, the diagnostic results from this model reaches the accuracy of 83.92 percent.

“This model can detect signs of infectious diseases which affect the lungs or other organs. Our challenge is to collect data from new cases and “teach” the diagnostic model through X-ray images, instead of the nasopharyngeal mucus test”, said Mr. Quang.

“In the context of increasing respiratory-related diseases caused by global air pollution and climate change, the AI technology is expected to greatly support the healthcare segment, especially in diagnosing and handling a large number of cases”, emphasized Mr. Quang.

FPT Software is known as one of the pioneering companies in Vietnam in researching and applying the foundation technologies of the 4.0 Revolution, such as AI and IoT to bring the healthcare sector breakthrough changes. Recently, FPT Software’s AI model of pneumothorax detection has competed with more than 1.400 competitors to be ranked as TOP 5 at a competition organized by the US’ Osteoarthritis Institute. Also, FPT Software has submitted this patent to Japan.

Source: Cucumber

Related posts: