APSR2026-947

 
Late-Breaking
Submitted
Oral Presentation
AO36-003
 
 
https://storage.unitedwebnetwork.com/files/1285/1241906-46903-APSR26_Oral Template_433551.pptx
Format: PPTX Only
 
Format: PDF Only
06. Tuberculosis
5

No more than 15 authors can be listed.
Please ensure the authors are listed in the right order.
If there are multiple affiliations, please use the "+" button next to "Select Country/Region" to add additional affiliation fields.

Pichaya Petborom - Srinakharinwirot university Internal medicine Bangkok Thailand
Tharit Khununyhaporn * Srinakharinwirot university Medicine Bangkok Thailand
Teeyapat Charoenwattananon - Srinakharinwirot university Medicine Bangkok Thailand
Thanathorn Limphitakphong - Srinakharinwirot university Medicine Bangkok Thailand
Thanutham Chaipanthaset - Srinakharinwirot university Medicine Bangkok Thailand
 
 
 
 
 
 
 
 
 
 
Presenting Author Details
Tharit
Khununyhaporn
Tharit.khu@gmail.com
+66875561130
Under 30
Thailand
Original Research
Evaluating the effectiveness of artificial intelligence in chest radiography interpretation for diagnosis in pulmonary tuberculosis compared with medical practitioner
Maximum 25 words
Abstract Content
Background and Aims: Tuberculosis (TB) remains a major public health challenge, particularly in high-burden countries such as Thailand. Delayed diagnosis contributes to ongoing transmission and poor clinical outcomes. Artificial intelligence (AI)-assisted chest radiograph interpretation has emerged as a promising tool for rapid TB screening. This study aimed to evaluate the diagnostic performance of an AI algorithm for pulmonary TB detection on chest radiographs compared with pulmonologists, radiologists, and sixth-year medical students.
Methods: A total of 365 chest radiographs obtained from patients treated at HRH Princess Maha Chakri Sirindhorn Medical Center between April 21 and May 21, 2024, were retrospectively analyzed. The dataset included patients with confirmed pulmonary TB and individuals without a history of TB. Chest radiographs were interpreted independently by pulmonologists, radiologists, sixth-year medical students, and the AI software Inspectra CXR. Human readers evaluated the images through an online platform, and their interpretations were compared with confirmed diagnoses. Diagnostic performance was assessed using accuracy, sensitivity, specificity, positive predictive value (PPV), and negative predictive value (NPV).
Results: The mean age of participants was 41.47 ± 17.22 years. Using a cut-off value of 50%, Inspectra CXR achieved an area under the receiver operating characteristic curve of 0.89, accuracy of 95.07% (95% CI, 92.74–97.38), sensitivity of 81.82% (95% CI, 77.86–85.77), and specificity of 95.48% (95% CI, 93.35–97.61). The PPV was 36.00% (95% CI, 31.08–40.92), while the NPV was 99.41% (95% CI, 98.63–100.00). Inspectra CXR demonstrated higher sensitivity than pulmonologists (36.36%), radiologists (54.55%), and sixth-year medical students (45.45%).
Conclusion: Inspectra CXR demonstrated excellent diagnostic performance for pulmonary TB screening, with high accuracy, specificity, and negative predictive value. Its superior sensitivity compared with human readers suggests that AI-assisted screening may serve as an effective tool for early TB detection and support TB control efforts in high-burden settings.
 
 
 
 
 
No
No
 
JPG or JPEG or PNG format ONLY
Tuberculosis, Artificial Intelligence, Chest X-ray
maximum 3 keywords and no abbreviations, separated by commas
 
319