The total population of the Philippines back in 2020 totalled approximately 110 million people. That year, there were an estimated 591,000 people who developed TB. Among them, 73,000 were children. Moreover, there were 334,459 missing people with TB, of which 54,551 were children.
In 2018, Delft Imaging collaborated with FIND and the International Organization for Migration (IOM) on using the CAD4TB artificial intelligence software to detect TB-related abnormalities on chest X-rays automatically.
Later, in 2021, we delivered CAD4TB to eight sites as part of the large iNTP project, funded by USAID and supported by the Stop TB Partnership. The project came with installation, training and long-term maintenance services.
In 2022, Delft Imaging delivered additional CAD4TB software to the Philippines to be used by Medecins Sans Frontieres (MSF). The project again featured installation, training and long-term maintenance services.
Making a difference
In a study published and presented during the Union World Conference on Lung Health in 2015, CAD4TB was utilised in the Palawan provincial areas of the Philippines. The study then used CAD4TB 4 (an older version of CAD4TB than what is currently available), and the software achieved a sensitivity of 90% and a specificity of 80%. The study concluded that computerised reading (artificial intelligence) provides high sensitivity and specificity and may assist human readers in active case-finding programs, thus improving screening throughput.
In a study published in 2019, researchers looked at automated chest X-ray readings for tuberculosis in the Philippines to improve case detection. The study examined 10,755 individuals, of which 2,534 had a positively assessed chest X-ray and 298 Xpert-positive cases. The publication noted that based on the radiological reference, the physician performed slightly worse than the CAD4TB software, although it was not found to be statistically significant at that time. The study concluded that the performance of automated chest X-ray reading is comparable to that of attending physicians, and its use as a second reader could increase TB case detection. Note that the study used an older version of CAD4TB than currently available.
