The TB Portals contain de-identified data from tuberculosis (TB) patient cases that have been contributed by multiple institutions from different clinical and research contexts. They have been collected as part of routine practice in TB clinics, research studies, and clinical trials. Some of the data originate from historical records, while others are being collected prospectively.

TB Portals radiological data include chest X-rays, CT studies, radiology reports, radiologist image annotations, and AI-generated image annotations. The majority of CT studies come from TB patients in Belarus, supplemented by those from patients in Moldova, Romania, Georgia, and Azerbaijan. Chest X-rays come from institutions in India, Belarus, Romania, Georgia, and Azerbaijan, with the largest proportion from Belarus. AI-generated image annotations are only included for a portion of cases.

The majority of the institutions contributing data have focused on collecting and submitting mostly drug-resistant TB cases. Currently, 75% of cases in the TB Portals are multi- or extensively drug-resistant.

Often, the data collected by contributing institutions have been used in studies of Mycobacterium tuberculosis (Mtb) genomic variability, comparing within and across epidemics, between different types of TB disease and anatomical localization, and during treatment. Because drug-resistant TB is more difficult to treat than drug-sensitive TB and a significant concern in many countries from which TB Portals data originates, many of these studies specifically focus on drug-resistant TB cases. In addition to analyses of region specific or treatment specific patterns of drug resistance, these studies investigate whether predicted resistance based on genomic analysis is in line with standard testing.


The TB Portals Program actively collects international TB patient case data, including clinical, imaging, and bacterial genomic information, from both drug-sensitive and resistant cases.

These de-identified and integrated data are made publicly available after quality checks and physician-verification. Data within the TB Portals are also linked to case-associated sputum samples stored locally. These samples can be used in additional research studies.

Of note, TB cases from the following research studies and institutions can be found within the TB Portals:

Cases with surgical samples of Mtb 

An emphasis of research being carried out by National Center for Tuberculosis and Lung Diseases (NCTLD), Ministry of Health, in Tbilisi, Georgia is how Mtb isolated from the lungs differ from those isolated from sputum. To answer this question, they began in 2017 to collect sputum and surgical lung samples from patient cases that have been deemed successfully cured by WHO criteria, yet present later with lung pathology. At the NCTLD, surgical removal of tuberculomas is done as a standard of care in these types of patients, and they are contributing clinical records, Mtb genomic sequences, and radiological images from these patients to the TB Portals. Active Mtb have been found in some of the surgical samples from these patient cases.

Drug-resistant relapse cases

A focus of research at the Institute of Phthisiopneumology, Ministry of Health, in Moldova is on TB reoccurrence, using genomics to differentiate between endogenous TB reactivation and exogenous TB reinfection. Longitudinal data (clinical records and Mtb genomic sequences from sputum samples) from cases classified as “relapse” by WHO classification have been contributed to the TB Portals by this group.

AI-generated image annotations

United Institute of Informatics Problems of the National Academy of Sciences of Belarus is working on a number of research projects, primarily with a focus on using machine learning methods to characterize and identify different clinical and genomic types of TB. One project is using artificial intelligence to recognize and digitally annotate lesions on chest CT studies and then compute a severity score based on these annotations. Many of these annotated CT studies are included in the TB Portals.

Cases on treatment with new TB drugs 

The Republican Scientific and Practical Center for Pulmonology and Tuberculosis (RSPCPT) in Minsk, Belarus is interested in identifying genomic variants related to emerging drug resistance in new TB drugs. The TB Portals contain a number of drug-resistant TB patient cases from this Center being treated with bedaquiline and delaminid. RSPCPT is also contributing patient cases that are receiving stem cell therapy in addition to standard drug treatment as a means to heal lung damage.

Extrapulmonary TB cases 

Institutes in Azerbaijan (Scientific Research Institute of Lung Diseases, Ministry of Health, Baku) and Kazakhstan (National Science Center of Phthisiopulmonology, Almatyare contributing laboratory-confirmed cases of extrapulmonary TB to the TB Portals. In their analyses, they seek to compare Mtb isolated from extrapulmonary sites to those from pulmonary sites.

Sensitive TB cases 

Institutes from South Africa (Perinatal HIV Research Unit, Chris Hani Baragwanath Hospital, Johannesburg), India (College of Pharmaceutical Sciences, Manipal), Republic of Congo (Fondation Congolaise pour la Recherche Medicale), China (The Chinese University of Hong Kong, Hong Kong), and Mali (University of Sciences, Techniques, and Technologies of Bamakoare primarily contributing drug-sensitive TB patient cases to the TB Portals. These are being collected prospectively, with a focus on genomic characterization of Mtb samples from sputum. Chest X-rays are also available from these sites. TB cases from Perinatal HIV Research Unit at Chris Hani Baragwanath Hospital in Johannesburg are associated with longitudinal sputum samples on standard (6-month) and shortened (4-month) treatment regimens for drug-sensitive TB.