In Brief | Deep Learning Detection of Active Pulmonary Tuberculosis at Chest Radiography Matched the Clinical Performance of Radiologists

Published on MedED: 2 July 2023

Google's artificial intelligence (AI) technology may become an essential tool, in improving early tuberculosis (TB) diagnosis, particularly in resource-constrained environments where access to skilled radiologists is limited. 

The AI technology, known as DeepMind, has been trained to analyse chest X-rays and detect signs of TB accurately. Researchers of a recent study published in Radiology report that the technology model demonstrated a sensitivity of 94% and a specificity of 97% in detecting TB-related abnormalities, significantly outperforming human radiologists. 

The potential benefits of this AI-powered diagnostic tool are significant: by improving the speed and accuracy of diagnosis, it could facilitate early detection, enabling earlier treatment, ultimately leading to better patient outcomes and reduced transmission rates.

The researchers note that several challenges need to be addressed before wide-scale adoption is possible. Further studies and validation are required to ensure its effectiveness across different populations and healthcare settings. Additionally, integrating this technology into existing healthcare systems and ensuring accessibility in regions with limited resources will be crucial.

Kazemzadeh, S., Yu, J., Jamshy, S., Pilgrim, R., Nabulsi, Z., Chen, C., Beladia, N., Lau, C., McKinney, S. M., Hughes, T., Kiraly, A. P., Kalidindi, S. R., Muyoyeta, M., Malemela, J., Shih, T., Corrado, G. S., Peng, L., Chou, K., Chen, P. C., Liu, Y., … Prabhakara, S. (2023). Deep Learning Detection of Active Pulmonary Tuberculosis at Chest Radiography Matched the Clinical Performance of Radiologists. Radiology, 306(1), 124–137. https://doi.org/10.1148/radiol.212213