Glioblastoma and radiotherapy: A multicenter AI study for Survival Predictions from MRI (GRASP study).

BACKGROUND: The aim was to predict survival of glioblastoma at 8 months after radiotherapy (a period allowing for completing a typical course of adjuvant temozolomide), by applying deep learning to the first brain MRI after radiotherapy completion. METHODS: Retrospective and prospective data were collected from 206 consecutive glioblastoma, isocitrate dehydrogenase -wildtype patients diagnosed between March 2014 and February 2022 across 11 UK centers. Models were trained on 158 retrospective patients from 3 centers. Holdout test sets were retrospective (n = 19; internal validation), and prospective (n = 29; external validation from 8 distinct centers). Neural network branches for T2-weighted and contrast-enhanced T1-weighted inputs were concatenated to predict survival. A nonimaging branch (demographics/MGMT/treatment data) was also combined with the imaging model. We investigated the influence of individual MR sequences; nonimaging features; and weighted dense blocks pretrained for abnormality detection. RESULTS: The imaging model outperformed the nonimaging model in all test sets (area under the receiver-operating characteristic curve, AUC P = .038) and performed similarly to a combined imaging/nonimaging model (P > .05). Imaging, nonimaging, and combined models applied to amalgamated test sets gave AUCs of 0.93, 0.79, and 0.91. Initializing the imaging model with pretrained weights from 10 000s of brain MRIs improved performance considerably (amalgamated test sets without pretraining 0.64; P = .003). CONCLUSIONS: A deep learning model using MRI images after radiotherapy reliably and accurately determined survival of glioblastoma. The model serves as a prognostic biomarker identifying patients who will not survive beyond a typical course of adjuvant temozolomide, thereby stratifying patients into those who might require early second-line or clinical trial treatment.

Constructing bilayer and volumetric atrial models at scale.

To enable large in silico trials and personalized model predictions on clinical timescales, it is imperative that models can be constructed quickly and reproducibly. First, we aimed to overcome the challenges of constructing cardiac models at scale through developing a robust, open-source pipeline for bilayer and volumetric atrial models. Second, we aimed to investigate the effects of fibres, fibrosis and model representation on fibrillatory dynamics. To construct bilayer and volumetric models, we extended our previously developed coordinate system to incorporate transmurality, atrial regions and fibres (rule-based or data driven diffusion tensor magnetic resonance imaging (MRI)). We created a cohort of 1000 biatrial bilayer and volumetric models derived from computed tomography (CT) data, as well as models from MRI, and electroanatomical mapping. Fibrillatory dynamics diverged between bilayer and volumetric simulations across the CT cohort (correlation coefficient for phase singularity maps: left atrial (LA) 0.27 ± 0.19, right atrial (RA) 0.41 ± 0.14). Adding fibrotic remodelling stabilized re-entries and reduced the impact of model type (LA: 0.52 ± 0.20, RA: 0.36 ± 0.18). The choice of fibre field has a small effect on paced activation data (less than 12 ms), but a larger effect on fibrillatory dynamics. Overall, we developed an open-source user-friendly pipeline for generating atrial models from imaging or electroanatomical mapping data enabling in silico clinical trials at scale (https://github.com/pcmlab/atrialmtk).