Diagnostic Accuracy of Artificial Intelligence-assisted Chest X-ray Interpretation Tools for Screening of Tuberculosis: A Systematic Review and Meta-analysis
Introduction: Tuberculosis (TB) continues to be a major worldwide health concern, which is the leading cause of death in nations like India. Despite various efforts to combat TB, effective screening and timely diagnosis remain challenging. The use of Artificial Intelligence (AI) in computer-aided in...
Saved in:
| Main Authors: | , , , , , , , |
|---|---|
| Format: | Article |
| Language: | English |
| Published: |
JCDR Research and Publications Private Limited
2025-08-01
|
| Series: | Journal of Clinical and Diagnostic Research |
| Subjects: | |
| Online Access: | https://jcdr.net/article_fulltext.asp?issn=0973-709x&year=2025&month=August&volume=19&issue=8&page=LC01-LC07&id=21293 |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Introduction: Tuberculosis (TB) continues to be a major worldwide health concern, which is the leading cause of death in nations like India. Despite various efforts to combat TB, effective screening and timely diagnosis remain challenging. The use of Artificial Intelligence (AI) in computer-aided interpretation of Chest X-Rays (CXRs) has demonstrated potential in improving TB detection.
Aim: To evaluate the diagnostic accuracy of AI-assisted CXR interpretation tools for TB screening.
Materials and Methods: Authors employed Systematic Review and Meta-analysis (SR/MA) using the Preferred Reporting Items for SR/MA of Diagnostic Test Accuracy (PRISMA-DTA) guideline. A comprehensive literature search was conducted for studies published between January 2019 and December 2023, focusing on AI-assisted software’s diagnostic accuracy in interpreting CXRs. Electronic databases such as Google Scholar, ScienceDirect, PubMed, and Institute of Electrical and Electronics Engineers Xplore (IEEE) were used. The raw diagnostic accuracy data, sensitivity, specificity, and Area Under the Curve (AUC) of the studies that met the inclusion criteria were examined and meta analysed to estimate pooled diagnostic accuracy measures.
Results: There were 1,825 records found in the database search. Ensuing screening and duplication removal, 170 full-text publications were assessed; 14 of them satisfied the requirements for inclusion in the SR/MA. The findings of SR highlight the important role of AI assisted diagnostic tools in faster and larger screening of patients. The meta-analysis revealed the overall sensitivity of AI assisted tools to be 92% (62.9-98.7%) while specificity was 98.2% (68.4-99.9%).
Conclusion: Although the large confidence interval questions the generalisability of the findings and consistency of the results, the present review signifies the important horizon that can be explored further for strengthening the TB elimination efforts. |
|---|---|
| ISSN: | 2249-782X 0973-709X |