Department of Radiology, University of Iowa Hospitals and Clinics, 200 Hawkins Drive, Iowa City, IA, 52242, USA.
College of Engineering, University of Iowa, Iowa City, IA, USA.
Eur Radiol. 2021 Nov;31(11):8703-8713. doi: 10.1007/s00330-021-07845-6. Epub 2021 Apr 23.
Despite the robust diagnostic performance of MRI-based radiomic features for differentiating between glioblastoma (GBM) and primary central nervous system lymphoma (PCNSL) reported on prior studies, the best sequence or a combination of sequences and model performance across various machine learning pipelines remain undefined. Herein, we compare the diagnostic performance of multiple radiomics-based models to differentiate GBM from PCNSL.
Our retrospective study included 94 patients (34 with PCNSL and 60 with GBM). Model performance was assessed using various MRI sequences across 45 possible model and feature selection combinations for nine different sequence permutations. Predictive performance was assessed using fivefold repeated cross-validation with five repeats. The best and worst performing models were compared to assess differences in performance.
The predictive performance, both using individual and a combination of sequences, was fairly robust across multiple top performing models (AUC: 0.961-0.977) but did show considerable variation between the best and worst performing models. The top performing individual sequences had comparable performance to multiparametric models. The best prediction model in our study used a combination of ADC, FLAIR, and T1-CE achieving the highest AUC of 0.977, while the second ranked model used T1-CE and ADC, achieving a cross-validated AUC of 0.975.
Radiomics-based predictive accuracy can vary considerably, based on the model and feature selection methods as well as the combination of sequences used. Also, models derived from limited sequences show performance comparable to those derived from all five sequences.
• Radiomics-based diagnostic performance of various machine learning models for differentiating glioblastoma and PCNSL varies considerably. • ML models using limited or multiple MRI sequences can provide comparable performance, based on the chosen model. • Embedded feature selection models perform better than models using a priori feature reduction.
尽管先前的研究报告称，基于 MRI 的放射组学特征在区分胶质母细胞瘤（GBM）和原发性中枢神经系统淋巴瘤（PCNSL）方面具有强大的诊断性能，但最佳序列或各种机器学习管道中序列和模型性能的最佳组合仍未确定。在此，我们比较了多种基于放射组学的模型区分 GBM 和 PCNSL 的诊断性能。
我们的回顾性研究包括 94 名患者（34 名 PCNSL 和 60 名 GBM）。使用各种 MRI 序列，通过 45 种可能的模型和特征选择组合，对 9 种不同序列排列进行了模型性能评估。使用五重交叉验证进行预测性能评估，重复 5 次。比较最佳和最差表现的模型以评估性能差异。
使用个体和组合序列的预测性能在多个表现最佳的模型中相当稳健（AUC：0.961-0.977），但在表现最佳和最差的模型之间存在相当大的差异。表现最佳的个体序列与多参数模型具有相当的性能。我们研究中的最佳预测模型使用 ADC、FLAIR 和 T1-CE 的组合，获得了最高 AUC 为 0.977，而排名第二的模型使用 T1-CE 和 ADC，获得了交叉验证 AUC 为 0.975。
基于模型和特征选择方法以及使用的序列组合，基于放射组学的预测准确性差异很大。此外，从有限的序列中得出的模型与从所有五个序列中得出的模型具有可比的性能。
用于区分胶质母细胞瘤和 PCNSL 的各种机器学习模型的基于放射组学的诊断性能差异很大。
基于所选模型，使用有限或多个 MRI 序列的 ML 模型可以提供可比的性能。
嵌入式特征选择模型比使用先验特征减少的模型表现更好。
