Tuberculosis or lung cancer: the role of chest CT radiomics in diagnosis of lung cancer in developing nations

Purpose or Learning Objective To identify the CT radiomic features of both tuberculosis (TB) and lung cancer (LC) and to assess the potential role of radiomic in differentiating these two diseases. Methods or Background From March,2020 to September,2021, sixty patients with certain diagnosis of TB or LC histologically proven, who underwent chest CT evaluation were retrospectively enrolled. Exclusion criteria were: negative chest CT or only enhanced CT scans, previous surgical resection and CT severe motion artefacts. Radiomics features, extracted and selected from CT data by two radiologists in consensus, were evaluated and the results were compared with histopathology. The performance of chest CT radiomic features in differentiating LC from TB was tested by receiver operating characteristic (ROC) curves and the areas under the curve (AUCs), calculating sensitivity and specificity too. Results or Findings Forty patients were finally enrolled, 31 male, mean age 59±12.9years (SD), range 21-82; twenty-eight patients were affected by LC and twelve by TB. Radiomic features were extracted by chest CT scans. Significant differences were found in 61/107 radiomic features: 4 Shape, 13 First Order, 18 GreyLevelCo-occurrenceMatrix (GLCM), 8 GrayLevelDependenceMatrix (GLDM), 9 Grey-LevelRunLengthMatrix (GLRLM), 6 GrayLevelSizeZoneMatrix (GLSZM) and 2 NeighboringGrayToneDifferenceMatrix (NGTDM), all with P<0.05. LargeDependenceEmphasis (GLDM feature) and LargeAreaLowGrayLevelEmphasis (GLSZM feature) showed the best performance in discriminating LC from TB, (respectively AUC:0.92, sensitivity: 85.7%, specificity: 91.7%; AUC: 0.92, sensitivity: 75%, specificity: 100%, all P<0.0001). Conclusion Radiomics may be a non-invasive imaging tool in many low-income countries for differentiating LC from TB and may have a pivotal role in avoiding delayed diagnosis of LC, improving oncological patients management.