AI is highly effective in diagnosing TB, KIMS AI-drive chest X-ray study
A total of 16,675 CXRs of adult patients were analyzed retrospectively using an advanced AI tool, qXR, without direct clinician intervention.
Hyderabad: Artificial intelligence-driven tools can correctly rule non-tuberculosis (TB) cases with an accuracy of 97 per cent and identify TB cases, with an impressive sensitivity rate of 88.7 percent, as a recent AI-driven study taken up by Krishna Institute of Medical Sciences (KIMS) Hospital, Secunderabad, said.
A total of 16,675 CXRs of adult patients were analyzed retrospectively using an advanced AI tool, qXR, without direct clinician intervention.
The study focused on two key objectives, including evaluating the diagnostic accuracy of AI in TB detection and assessing its agreement with radiologists.
A significant aspect of the study was the agreement level between AI-driven assessments and expert radiologists’ interpretations. This is particularly significant given the global burden of TB and the challenges associated with its timely detection through traditional radiography, Dr. Latha Sarma, Head, Pulmonologist, KIMS Hospitals, said.
The ability of AI to assist in TB detection with such high accuracy is a game-changer, observed Dr. Chaithanya Isamalla, Senior Radiologist at KIMS.
