Gomaa Ahmed, Huang Yixing, Hagag Amr, Schmitter Charlotte, Höfler Daniel, Weissmann Thomas, Breininger Katharina, Schmidt Manuel, Stritzelberger Jenny, Delev Daniel, Coras Roland, Dörfler Arnd, Schnell Oliver, Frey Benjamin, Gaipl Udo S, Semrau Sabine, Bert Christoph, Hau Peter, Fietkau Rainer, Putz Florian
Department of Radiation Oncology, University Hospital Erlangen, Friedrich-Alexander-Universität Erlangen-Nürnberg, Erlangen, Germany.
Comprehensive Cancer Center Erlangen-EMN (CCC ER-EMN), Erlangen, Germany.
Neurooncol Adv. 2024 Jul 11;6(1):vdae122. doi: 10.1093/noajnl/vdae122. eCollection 2024 Jan-Dec.
This research aims to improve glioblastoma survival prediction by integrating MR images, clinical, and molecular-pathologic data in a transformer-based deep learning model, addressing data heterogeneity and performance generalizability.
We propose and evaluate a transformer-based nonlinear and nonproportional survival prediction model. The model employs self-supervised learning techniques to effectively encode the high-dimensional MRI input for integration with nonimaging data using cross-attention. To demonstrate model generalizability, the model is assessed with the time-dependent concordance index (Cdt) in 2 training setups using 3 independent public test sets: UPenn-GBM, UCSF-PDGM, and Rio Hortega University Hospital (RHUH)-GBM, each comprising 378, 366, and 36 cases, respectively.
The proposed transformer model achieved a promising performance for imaging as well as nonimaging data, effectively integrating both modalities for enhanced performance (UCSF-PDGM test-set, imaging Cdt 0.578, multimodal Cdt 0.672) while outperforming state-of-the-art late-fusion 3D-CNN-based models. Consistent performance was observed across the 3 independent multicenter test sets with Cdt values of 0.707 (UPenn-GBM, internal test set), 0.672 (UCSF-PDGM, first external test set), and 0.618 (RHUH-GBM, second external test set). The model achieved significant discrimination between patients with favorable and unfavorable survival for all 3 datasets (log-rank 1.9 × 10, 9.7 × 10, and 1.2 × 10). Comparable results were obtained in the second setup using UCSF-PDGM for training/internal testing and UPenn-GBM and RHUH-GBM for external testing (Cdt 0.670, 0.638, and 0.621).
The proposed transformer-based survival prediction model integrates complementary information from diverse input modalities, contributing to improved glioblastoma survival prediction compared to state-of-the-art methods. Consistent performance was observed across institutions supporting model generalizability.
本研究旨在通过在基于Transformer的深度学习模型中整合磁共振成像（MR）图像、临床和分子病理数据，改善胶质母细胞瘤的生存预测，解决数据异质性和性能可推广性问题。
我们提出并评估了一种基于Transformer的非线性和非比例生存预测模型。该模型采用自监督学习技术，通过交叉注意力有效地对高维MRI输入进行编码，以便与非成像数据集成。为了证明模型的可推广性，使用3个独立的公共测试集在2种训练设置中，用时依一致性指数（Cdt）对模型进行评估：宾夕法尼亚大学胶质母细胞瘤数据集（UPenn-GBM）、加州大学旧金山分校多模态胶质母细胞瘤数据集（UCSF-PDGM）和里奥霍特加大学医院胶质母细胞瘤数据集（RHUH-GBM），每个数据集分别包含378例、366例和36例。
所提出的Transformer模型在成像数据和非成像数据方面均取得了良好的性能，有效地整合了两种模态以提高性能（UCSF-PDGM测试集，成像Cdt为0.578，多模态Cdt为0.672），同时优于基于3D-CNN的最新晚期融合模型。在3个独立的多中心测试集中观察到了一致的性能，Cdt值分别为0.707（UPenn-GBM，内部测试集）、0.672（UCSF-PDGM，第一个外部测试集）和0.618（RHUH-GBM，第二个外部测试集）。该模型在所有3个数据集中对生存良好和不良的患者都实现了显著的区分（对数秩分别为1.9×10、9.7×10和1.2×10）。在第二种设置中，使用UCSF-PDGM进行训练/内部测试，使用UPenn-GBM和RHUH-GBM进行外部测试，也获得了类似的结果（Cdt分别为0.670、0.638和0.621）。
所提出的基于Transformer的生存预测模型整合了来自不同输入模态的互补信息，与现有方法相比，有助于改善胶质母细胞瘤的生存预测。在不同机构中观察到了一致的性能，支持模型的可推广性。
