TOPLINE:

Chest CT-based radiomic analysis revealed 16 distinct features that could differentiate tumourlets (TLs) from granulomas, with flatness showing a specificity of 100% and long-run high grey-level emphasis achieving a sensitivity of 92.3% in identification.

METHODOLOGY:

• Researchers conducted a retrospective observational study at Sant'Andrea University Hospital, Rome, Italy, between 2013 and 2021 and analysed 55 patients who underwent lung surgery (32 with TLs and 23 with granulomas).
• Two experienced thoracic radiologists independently manually segmented pulmonary lesions on unenhanced preoperative CT images of the patients to delineate the regions of interest.
• Investigators utilised an open-source 3D Slicer Radiomics software to extract 107 radiomic features, including shape and texture metrics.
• Statistical analysis evaluated the performance of these features in differentiating TLs from granulomas by calculating the area under the curve (AUC), sensitivity, and specificity.

TAKEAWAY:

• Significant differences in 16 out of 107 radiomic features were identified between TLs and granulomas, which included shape, first-order statistics, and matrix measurements.
• The flatness feature demonstrated high discriminative performance (AUC, 0.903; sensitivity, 76.9%; specificity, 100%; P < .001).
• Long-run high grey-level emphasis showed strong differentiation capability (AUC, 0.896; sensitivity, 92.3%; specificity, 76.5%; P < .001).
• Small-area low grey-level emphasis and low grey-level zone emphasis also exhibited notable differentiation performance (AUC, 0.898; sensitivity, 92.3%; specificity, 70.6%; P < .001 and AUC, 0.894; sensitivity, 76.9%; specificity, 88.2%; P < .001, respectively).

IN PRACTICE:

"Radiomics may serve as a non-invasive tool to support the characterization of small lung nodules, helping to differentiate between TL and granulomas, and potentially aiding clinical decision-making by identifying nodules that may require closer monitoring or further investigation," the authors wrote.

"Further prospective studies involving larger patient cohorts are warranted to validate these preliminary findings," they added.

SOURCE:

This study was led by Alessandra Siciliani, Sant'Andrea University Hospital, Rome, Italy. It was published online on December 27, 2025, in the Journal of Clinical Medicine.

LIMITATIONS:

The relatively small sample size and retrospective design restricted the statistical power and generalisability. The analysis relied solely on univariate statistics without multivariate predictive modelling due to the limited number of cases, potentially leading to optimistic AUC values without internal validation procedures. Additionally, this study was limited by the potential subjectivity in manual lesion segmentation, variability in CT acquisition parameters, and the lack of an independent validation cohort.

DISCLOSURES:

This study did not receive any external funding. The authors reported having no relevant conflicts of interest.

This article was created using several editorial tools, including AI, as part of the process. Human editors reviewed this content before publication.