Open-source PET-CT finds early signals of progressing tuberculosis in contacts

Tuberculosis remains a complex infection because people exposed to Mycobacterium tuberculosis can show a wide range of responses, from no disease to progressive, active tuberculosis. Positron Emission Tomography-Computed Tomography (PET-CT) has emerged as a powerful research tool to visualise this variability in the body, but its broader use depends on consistent ways to take, read and analyse images. Corresponding author Jee Whang Kim and colleagues set out to make PET-CT more standardised and reproducible for studies of TB contact persons. They tested a semi-automated approach built around open-source tools to reduce differences between human readers and to adjust for normal background uptake in organs. By focusing on people who had positive infection tests but were contacts of active TB cases, the team hoped to better characterise early infection changes and to lay groundwork for comparing results across different research groups.

The team measured 18 F-Fluorodeoxyglucose (FDG) uptake in intrathoracic lymph nodes (ITLNs) from eleven QuantiFERON-TB Gold-Plus positive TB contacts enrolled in a prospective observational study, eight of whom had serial PET-CT scans. Imaging was analysed using 3D Slicer, an open-source medical imaging platform that provides semi-automated segmentation and automated liver quantification. Extracted metrics included ITLN maximum standardised uptake value (SUVmax), mean standardised uptake value (SUVmean), metabolic volume (MV), total lesion glycolysis (TLG), and liver SUVmean. The semi-automated 3D Slicer workflow was compared to clinical software, both with and without standardising to liver uptake as a physiological reference, and operator variability was assessed. The authors report strong agreement between the open-source and clinical evaluation platforms, that semi-automated PET quantification reduced inter-operator variability, and that standardising to reference organs increased the sensitivity of the imaging analysis. Importantly, increasing metabolic activity on serial PET-CT in early infection was associated with detectable Mtb at the anatomical site of FDG uptake, a pattern consistent with a progressive phenotype of infection.

These results suggest practical steps toward making PET-CT a more reliable research tool for studying TB infection. By using an open-source platform and semi-automated steps, researchers can lower the chance that different operators will produce inconsistent results and can normalise images to a reference organ like the liver to improve sensitivity. The reported agreement with clinical software indicates that open-source workflows can be trustworthy while also promoting wider sharing and comparison of data across cohorts. The observed link between rising FDG activity on serial scans and the presence of Mtb at the same anatomical site supports the idea that PET-CT can reveal early, progressive changes in infection. The study describes preliminary findings that need further validation, but it points to a workflow that could help future prospective studies better detect and characterise who among TB contacts may be moving toward active disease.