Purpose: To assess the role of muscle composition and radiomics in predicting allograft rejection in lung transplant.
Material and methods: The last available HRCT before surgery of lung transplant candidates referring to our tertiary center from January 2010 to February 2020 was retrospectively examined. Only scans with B30 kernel reconstructions and 1 mm slice thickness were included. One radiologist segmented the spinal muscles of each patient at the level of the 11th dorsal vertebra by an open-source software. The same software was used to extract Hu values and 72 radiomic features of first and second order. Factor analysis was applied to select highly correlating features and then their prognostic value for allograft rejection was investigated by logistic regression analysis (level of significance p < 0.05). In case of significant results, the diagnostic value of the model was computed by ROC curves.
Results: Overall 200 patients had a HRCT prior to the transplant but only 97 matched the inclusion criteria (29 women; mean age 50.4 ± 13 years old). Twenty-one patients showed allograft rejection. The following features were selected by the factor analysis: cluster prominence, Imc2, gray level non-uniformity normalized, median, kurtosis, gray level non-uniformity, and inverse variance. The radiomic-based model including also Hu demonstrated that only the feature Imc2 acts as a predictor of allograft rejection (p = 0.021). The model showed 76.6% accuracy and the Imc2 value of 0.19 demonstrated 81% sensitivity and 64.5% specificity in predicting lung transplant rejection.
Conclusion: The radiomic feature Imc2 demonstrated to be a predictor of allograft rejection in lung transplant.
Keywords: Computed tomography; Lung transplant; Muscle composition; Radiomics.
© 2023. The Author(s).