To support the integration of lung ultrasound as a diagnostic tool for lung infections, Noémie Boillat-Blanco collaborates with the LiGHT laboratory at EPFL and at the Harvard Chan School of Public Health, leveraging AI to facilitate lung ultrasound interpretation [1, 2].
Her research team demonstrated the exceptional performance of AI-guided lung ultrasound for tuberculosis diagnosis, achieving a sensitivity of 0.91 (95% CI 0.90-0.96) and specificity 0.85 (95% CI 0.74-0.88), meeting WHO criteria for sputum-free TB triage. Recently, as part of an international consortium, they were awarded an EDCTP Horizon grant to further validate the diagnostic performance of this tool and facilitate its implementation.
She is also exploring the potential of AI-guided lung ultrasound across various patient’s populations and for diagnosing different lung infections, including pneumonia in emergency departments (700-patient cohort) and nursing homes (200-patients cohort, analyses ongoing).
Through this research, she aims to leverage AI to expand access to healthcare and improve the implementation of efficient diagnostic tools, particularly in medical imaging, ultimately enhancing patient care and diagnostic accuracy.
Selected publications:
[1] Suttels V, Wachinou P, Du Toit J, Boillat-Blanco N*, Hartley MA*. Ultrasound for point-of-care sputum-free tuberculosis detection: Building collaborative standardized image-bank. EBioMedicine 2022; 81: 104078. DOI.
[2] Suttels V, Du Toit J, Fiogbe A, Wachinou P, Guendehou B, Alovokpinhou F, Toukoui P, Hada A, Sefou F, Vinasse P, Makpemikpa G, Capo-Chichi D, Garcia E, Brahier T, Keitel K, Ouattara K, Cissoko Y, Beyey S, Mans P, Agodokpessi G, Boillat-Blanco N*, Hartley MA*. Point-of-care ultrasound for tuberculosis management in Sub-Saharan Africa – a balanced SWOT analysis. Int J Infect Dis 2022; 123: 46. DOI.