Combined radiomics-clinical model improves AP for differentiating pulmonary tuberculosis from non-tuberculous lung lesions on CTNew CT model may help distinguish tuberculosis from other lung lesions in some patients

Frontiers in Medicine Published April 9, 2026 DOI ↗ Editorial oversight: Dr. Amelia Tan, PhD · Internal Medicine & Chronic Disease

AI-generated summary of the cited source, checked by automated accuracy review. How we work

Key Takeaway

Note that combined radiomics-clinical model AP declines in temporal validation, limiting immediate clinical utility.

This observational study assessed the feasibility of a combined model integrating radiomics and clinical-semantic features for differentiating pulmonary tuberculosis from non-tuberculous solid lung lesions using contrast-enhanced CT. The analysis included 900 patients enrolled prior to October 2017. The primary outcome measured average precision (AP) across three datasets: training, internal validation, and temporal validation.

In the training set, the combined model achieved an AP of 0.91, compared to 0.64 for the clinical-semantic model alone. In the internal validation set, the combined model maintained an AP of 0.85 versus 0.61 for the comparator. However, performance dropped significantly in the temporal validation set, where the combined model yielded an AP of 0.62 compared to 0.41 for the clinical-semantic model.

Safety data, including adverse events, discontinuations, and tolerability, were not reported. The study limitations are not explicitly detailed in the provided text, though the substantial drop in AP from internal to temporal validation sets implies potential overfitting or lack of generalizability over time. Funding sources and conflicts of interest were not reported.

The practice relevance remains uncertain given the observational design and the decline in model performance over time. Clinicians should interpret these results with caution, noting that the combined approach may not yet be ready for routine deployment without further external validation.

This study looked at whether a new computer model could better tell the difference between pulmonary tuberculosis and other solid lung lesions using contrast-enhanced CT scans. The team analyzed data from 900 patients who were enrolled before October 2017. They compared a combined model that used both radiomics and clinical-semantic features against a model that used only clinical-semantic features.

In the training set, the combined model achieved an average precision of 0.91, while the clinical-semantic model scored 0.64. In the internal validation set, the combined model scored 0.85 compared to 0.61 for the clinical-semantic model. However, in the temporal validation set, the combined model dropped to 0.62, and the clinical-semantic model fell to 0.41.

No safety concerns were reported because the study focused on diagnostic accuracy rather than treatment or patient outcomes. Readers should note that this was an observational study using historical data, which limits how much these results can be applied to current practice. The drop in performance over time suggests the model may need careful testing in real-world settings before it can be trusted for routine use.

What this means for you:

A combined CT model showed promise in early tests but needs more validation before clinical use.

Study Details

EvidenceLevel 5

PublishedMar 2026

View Original Abstract ↓

BackgroundTo explore the feasibility of a combined model integrating radiomics and clinical-semantic features for differentiating pulmonary tuberculosis (PTB) from non-tuberculous solid lung lesions on contrast-enhanced CT.MethodsIn this study, 900 patients enrolled before October 2016 were randomly partitioned into training and internal validation sets at a 3:1 ratio, while those recruited between October 2016 and October 2017 formed an independent temporal validation set. Clinical-semantic features were selected through univariate analysis followed by multivariate analysis, while predictive radiomics features were identified using analysis of variance, Spearman correlation analysis, least absolute shrinkage and selection operator regression. Binary logistic regression was then used to construct the clinical-semantic, radiomics, and combined models. Model performance was evaluated using average precision (AP) derived from the precision-recall curve, and differences between models were assessed using bootstrap resampling. Clinical utility was assessed using decision curve analysis.ResultsFollowing feature selection, two clinical-semantic and three radiomics features were incorporated into the combined model. This model achieved APs of 0.91, 0.85, and 0.62 in the training, internal validation, and temporal validation sets, respectively, outperforming the clinical-semantic model, which yielded APs of 0.64, 0.61, and 0.41 (p 

Combined radiomics-clinical model improves AP for differentiating pulmonary tuberculosis from non-tuberculous lung lesions on CTNew CT model may help distinguish tuberculosis from other lung lesions in some patients

Study Details

Late-onset rheumatoid arthritis shows higher DAS28 scores and lower remission rates compared to young-onset disease in a meta-analysis

Meta-analysis compares treatment outcomes in older versus younger rheumatoid arthritis patients

Clinical research that matters. Delivered to your inbox.

Combined radiomics-clinical model improves AP for differentiating pulmonary tuberculosis from non-tuberculous lung lesions on CTNew CT model may help distinguish tuberculosis from other lung lesions in some patients

More on Pulmonary Tuberculosis

Study Details

Late-onset rheumatoid arthritis shows higher DAS28 scores and lower remission rates compared to young-onset disease in a meta-analysis

Meta-analysis compares treatment outcomes in older versus younger rheumatoid arthritis patients

Clinical research that matters. Delivered to your inbox.

Related in Rheumatology

From Other Specialties