Search

Simple Summary: Studies comparing performances of machine learning (ML) methods in building predictive models of toxicity in RT are rare. Thanks to the availability of a large cohort (n = 1314) of breast cancer patients homogeneously treated with tangential fields, different ML approaches could be compared. This work shows how more complex models typically achieve higher performances. At the same time, for this test case, the importance is given mainly by a few variables, and toxicity can be predicted by simpler models with similar performances. The availability of more individually characterizing features (here partially missing) is expected to have a likely much higher impact than the choice of the best-performing ML/DL approach. Purpose. Different ML models were compared to predict toxicity in RT on a large cohort (n = 1314). Methods. The endpoint was RTOG G2/G3 acute toxicity, resulting in 204/1314 patients with the event. The dataset, including 25 clinical, anatomical, and dosimetric features, was split into 984 for training and 330 for internal tests. The dataset was standardized; features with a high p-value at univariate LR and with Spearman ρ > 0.8 were excluded; synthesized data of the minority were generated to compensate for class imbalance. Twelve ML methods were considered. Model optimization and sequential backward selection were run to choose the best models with a parsimonious feature number. Finally, feature importance was derived for every model. Results. The model's performance was compared on a training–test dataset over different metrics: the best performance model was LightGBM. Logistic regression with three variables (LR3) selected via bootstrapping showed performances similar to the best-performing models. The AUC of test data is slightly above 0.65 for the best models (highest value: 0.662 with LightGBM). Conclusions. No model performed the best for all metrics: more complex ML models had better performances; however, models with just three features showed performances comparable to the best models using many (n = 13–19) features. [ABSTRACT FROM AUTHOR]

Purpose: Radiomics may aid in predicting prognosis in patients with colorectal liver metastases (CLM). Consistent data is available on CT, yet limited data is available on MRI. This study assesses the capability of MRI-derived radiomic features (RFs) to predict local tumor progression-free survival (LTPFS) in patients with CLMs treated with microwave ablation (MWA)., Methods: All CLM patients with pre-operative Gadoxetic acid-MRI treated with MWA in a single institution between September 2015 and February 2022 were evaluated. Pre-procedural information was retrieved retrospectively. Two observers manually segmented CLMs on T2 and T1-Hepatobiliary phase (T1-HBP) scans. After inter-observer variability testing, 148/182 RFs showed robustness on T1-HBP, and 141/182 on T2 (ICC > 0.7).Cox multivariate analysis was run to establish clinical (CLIN-mod), radiomic (RAD-T1, RAD-T2), and combined (COMB-T1, COMB-T2) models for LTPFS prediction., Results: Seventy-six CLMs (43 patients) were assessed. Median follow-up was 14 months. LTP occurred in 19 lesions (25%).CLIN-mod was composed of minimal ablation margins (MAMs), intra-segment progression and primary tumor grade and exhibited moderately high discriminatory power in predicting LTPFS (AUC = 0.89, p  = 0.0001). Both RAD-T1 and RAD-T2 were able to predict LTPFS: (RAD-T1: AUC = 0.83, p  = 0.0003; RAD-T2: AUC = 0.79, p  = 0.001). Combined models yielded the strongest performance (COMB-T1: AUC = 0.98, p  = 0.0001; COMB-T2: AUC = 0.95, p  = 0.0003). Both combined models included MAMs and tumor regression grade; COMB-T1 also featured 10 th percentile of signal intensity, while tumor flatness was present in COMB-T2., Conclusion: MRI-based radiomic evaluation of CLMs is feasible and potentially useful for LTP prediction. Combined models outperformed clinical or radiomic models alone for LTPFS prediction.