70th anniversary of Hanoi's Liberation Day Vietnam - Asia 2023 Smart City Summit Hanoi celebrates 15 years of administrative boundary adjustment 12th Vietnam-France decentrialized cooperation conference 31st Sea Games - Vietnam 2021 Covid-19 Pandemic
Aug 05, 2019 / 17:36

Vietnam develops diagnosis software through X-ray images

The software is built based on open data from actual radiographs of people with and without disease.

The research team of the Big Data Research Institute owned by conglomerate Vingroup has successfully developed a trial version of an auto-diagnosis software based on the input data of X-ray images, with accuracy of up to 90%, local media reported.

Usually, medical imaging diagnostics (such as X-ray images, cut-offs, magnetic resonance, among others) are performed manually by imaging diagnostic doctors. This approach has certain limitations on accuracy. Doctors can sometimes miss minor but important details because of visual restriction, time pressure, and work intensity.
 
The Big Data Research Institute signs a Memorandum of Understanding with with the Information Technology Agency under the Ministry of Health to prepare input data. Photo: Tienphong
The Big Data Research Institute signs a Memorandum of Understanding with with the Information Technology Agency under the Ministry of Health to prepare input data. Photo: Tienphong
The restrictions will be overcome when artificial intelligence (AI) is applied to image diagnosis. This approach has been successfully implemented by the Big Data Research Institute for chest radiographic diagnosis. This is a routine diagnosis and in great demand, accounting for 70-80% of the imaging diagnoses in hospitals in Vietnam and in the world.

The software is built based on open data from actual radiographs of people with and without disease. The results show that based on the health sector’s criteria, both the sensitivity and specificity of the algorithm, its diagnostic accuracy would reach over 90%, even over 95% for some diseases. If based on processing speed criteria, to read a standard X-ray image, doctors will need 5-10 minutes while it takes the diagnosis reader only 2-5 seconds, depending on the hardware configuration.

According to Dr. Nguyen Quy Ha, head of the medical image analysis team, although positive results have been initially achieved, the algorithm is being built based on a source of foreign open data. If it is possible to collect medical image data of Vietnamese people, the software accuracy and applicability will be higher.

In preparation for the input data, on July 23, the institute signed a memorandum of understanding with the Information Technology Agency under the Ministry of Health. The two sides agreed to coordinate in developing regulations on exchanging medical examination and treatment data of people. 

To put software into use, besides collecting Vietnamese data, legal issues and licensing procedures will be the next steps to be taken. "We will develop the diagnostic capacity for some dangerous and common diseases in Vietnam such as cancers or cardiovascular, neurological and diabetic diseases, with a view to meeting international standards and becoming useful to domestic hospitals," Director of the Big Data Research Institute Vu Ha Van said.