Radiomics nomogram analysis of T2-fBLADE-TSE in pulmonary nodules evaluation

• Published in: Magnetic resonance imaging 2022-01, Vol.85, p.80-86
• Main Authors: Yang, Shuyi, Wang, Yida, Shi, Yuxin, Yang, Guang, Yan, Qinqin, Shen, Jie, Wang, Qingle, Zhang, Haoling, Yang, Shan, Shan, Fei, Zhang, Zhiyong
• Format: Article
• Language: English
• Subjects: Differential diagnosis; Humans; Lung Neoplasms - diagnostic imaging; Lung Neoplasms - pathology; MRI; Nomogram; Nomograms; Pulmonary nodule; Radiomics; Retrospective Studies; Tomography, X-Ray Computed
• ISSN: 0730-725X; 1873-5894
• DOI: 10.1016/j.mri.2021.10.010

To develop and validate a radiomics nomogram for differentiating between malignant pulmonary nodules and benign nodules. 56 benign and 51 malignant nodules from 96 patients were analyzed using manual segmentation of the T2-fBLADE-TSE, while the nodules signal intensity (SIlesion), lesion muscle ratio (LMR) and nodule size were all measured and recorded. The maximum relevance and minimum redundancy (mRMR) and the least absolute shrinkage and selection operator (LASSO) were used to select nonzero coefficients and develop the model in pulmonary nodules diagnosis. The radiomics nomogram was also developed. The clinical prediction value was determined by the decision curve analysis (DCA). The nodule size, SIlesion and LMR of the benign group were 1.78 ± 0.57 cm, 227.50 ± 81.39 and 2.40 ± 1.27 respectively, in contrast to the 2.00 ± 0.64 cm, 232.87 ± 82.21 and 2.17 ± 0.91, respectively, in the malignant group (P = 0.09, 0.60 and 0.579). A total of 13 radiomics features were retained. The Rad-score of the benign nodules group was lower than that of the malignant nodules group (P < 0.001 & 0.049, training & test set). The AUC of radiomics signature for nodules diagnosis was 0.82 (95% CI, 0.73–0.91) in the training set and 0.71 (95% CI, 0.51–0.90) in the test set. A nomogram, consisting of 13 radiomics features and nodule size, produced good prediction in the training set (AUC, 0.82; 95% CI, 0.73–0.91), which was significantly better than that of T2-based quantitative parameters (AUC, 0.62; 95% CI, 0.50–0.75, P = 0.003). In the test set, the performance of radiomics nomogram (AUC, 0.70; 95% CI, 0.51–0.90) was also better than that of T2-based quantitative parameters (AUC, 0.46; 95% CI, 0.25–0.67) (P = 0.145). The DCA showed that radiomics nomogram and T2-based quantitative parameter had overall net benefits, while the performance of nomogram was better. We constructed and validated a T2-fBLADE-TSE-based radiomics nomogram that can help to differentiate between malignant pulmonary nodules and benign nodules. •TSE with fast-BLADE, which also termed as PROPELLER, is used to depict and evaluate pulmonary nodules.•Rad-score was calculated by the mathematical model with the selected useful prediction radiomics features in pulmonary nodules diagnosis•We developed and validated the first radiomics nomogram based on the T2-fBLADE-TSE to differentiate between with malignant pulmonary nodules and benign nodules.