NTT DATA, a leading IT services provider, announced today that the Artificial Intelligence (AI) diagnosis support solution of its US subsidiary NTT DATA Services was able to detect 56 emphysema cases while normal diagnosis without AI detect 17 cases during a recent Proof of Concept (PoC) conducted at the Deenanath Mangeshkar Hospital and Research Center in Pune, India.
The solution from NTT DATA Services is designed to help prioritize doctors’ workloads and expedite their reporting processes by using AI technology to analyze patients’ medical images and provide radiologists with diagnosis support.
When used to identify critical illnesses, CT and MRI scans can produce hundreds of images per study, the analysis of which can be very complex and can take radiologists considerable time. Market reports suggest the use of scanning equipment continues to grow globally and often too few radiologists are available to review the resulting volume of images. Efficient diagnosis with the assistance of AI has, therefore, become a key opportunity, which NTT DATA’s AI solution will help address.
Patients’ physical conditions and the diseases they most commonly suffer differ widely between countries due to several factors, including variations in climate, living conditions and eating habits. These physical differences are a potential challenge to the global deployment of AI in imaging, so the PoC was conducted to evaluate the capability to handle differences in patient populations. After an initial exercise in the United States to evaluate the solutions’ performance, NTT DATA carried out further tests using an anonymized annotated dataset at Deenanath Mangeshkar Hospital and Research Center in India with a help of NTT DATA partner company in India, DeepTek. The PoC utilized the data of 389 patients, comparing the results of the AI diagnosis support solution with diagnoses made by a team of radiologists.
Through the current PoC, it was also confirmed that with AI more emphysema cases than originally reported can be discovered. Investigating the cause of this an interesting observation was that most cases were mild or moderate levels of emphysema, rather than severe, which is the emphysema severity degree usually mentioned in radiology reports. Thus, it became apparent the current AI algorithm can contribute to early findings for early treatment or tracking of disease progression for risk management.
“While technology has driven considerable change in the medical industry, a shortage of qualified doctors has impacted the speed and efficiency of conventional diagnoses,” said Hiroyuki Kazama, Head of R&D HQs, NTT DATA. “With this AI diagnosis imaging solution, we are confident we can enhance the quality and efficiency of detection and diagnosis.”