You are currently viewing Artificial Intelligence Based Chest X-Ray for screening of Tuberculosis (AI-TB)

Artificial Intelligence Based Chest X-Ray for screening of Tuberculosis (AI-TB)

In this study, we integrated a rapid cough surveillance approach along with symptom screening for the presumptive case identification at the hospital settings. We selected suspected TB patients visiting OPD from both study sites, who were screened for TB. Each study participant received posterior-anterior CXR using digital X-ray machines. Each CXR were classified as “abnormal” if any pulmonary abnormality was detected by human readers. Similarly, radiologists were blinded of the results generated by machine learning for TB cases. However, the final confirmation of the test result was considered based on GeneXpert diagnosis of the sputum sample. GeneXpert diagnosis was therefore be used as the reference for the study.

AI-TB project was a six-month duration research project, funded by the International Organization of Migration(IOM), Nepal. The research project aimed to evaluate the diagnostic accuracy of AI-based CXR by comparing its results with the human reading (radiologist), and using GeneXpert-MTB RIF Assay as a reference standard for Tuberculosis case detection. The study was conducted at two study sites in Nepal: Dhulikhel Hospital Kathmandu University Hospital (DHKUH) and Nobel Medical College and Hospital Private Limited. 

Team:
InvestigatorsDr. Rajeev Shrestha, Dr. Pushpanjali Adhikari, Ruby Shrestha

Project Team: Ruby Shrestha, Manish Rajbansi, Kusum Shilpakar, Shrishak Shahi, Prashamsa Bhandari

Radiologist: Dr. Ram Chandra Paudel, Dr. Asish Bahadur Sigh

Microbiologist: Madhu