Although machine learning (ML) has shown promise across disciplines, out-of-sample generalizability is concerning. This is currently addressed by sharing multi-site data, but such centralization is challenging/infeasible to scale due to various limitations. Federated ML (FL) provides an alternative paradigm for accurate and generalizable ML, by only sharing numerical model updates. Here we present the largest FL study to-date, involving data from 71 sites across 6 continents, to generate an automatic tumor boundary detector for the rare disease of glioblastoma, reporting the largest such dataset in the literature (n = 6, 314). We demonstrate a 33% delineation improvement for the surgically targetable tumor, and 23% for the complete tumor extent, over a publicly trained model. We anticipate our study to: 1) enable more healthcare studies informed by large diverse data, ensuring meaningful results for rare diseases and underrepresented populations, 2) facilitate further analyses for glioblastoma by releasing our consensus model, and 3) demonstrate the FL effectiveness at such scale and task-complexity as a paradigm shift for multi-site collaborations, alleviating the need for data-sharing.

Glioblastoma (GBM) is the most common primary brain tumor. Despite aggressive treatment, GBM almost always recurs. The current standard-of-care for treatment of newly diagnosed GBM has remained relatively unchanged since 2005: maximal safe resection followed by concomitant chemoradiation (CRT) with temozolomide (TMZ), and subsequent adjuvant TMZ. In 2011, the first-generation tumor treating fields (TTF) device, known at the time as the NovoTTF-100A System (renamed Optune), was approved by the Food and Drug Administration (FDA) for treatment of recurrent GBM. The TTF device was subsequently approved as an adjuvant therapy for newly-diagnosed GBM in 2015. The following is a review of the TTF device, including evidence supporting its use and limitations.

Stereotactic radiosurgery (SRS) has become a well-established first-line therapy for limited brain metastases after multiple phase 3 randomized clinical trials of SRS with and without whole-brain radiotherapy (WBRT) collectively demonstrated no overall survival advantage with the addition of WBRT to SRS despite the superior central nervous system (CNS) control observed with WBRT.^1-5 The absence of an overall survival advantage to justify the toxic effects of WBRT on cognitive function and quality of life^1,4,6 made SRS alone the preferred treatment for limited brain metastases in most settings.⁷ However, patients with small cell lung cancer were excluded from the landmark randomized clinical trials^2-5,8 that established SRS alone as a first-line strategy, making small cell lung cancer an exception where WBRT has remained the standard of care for limited and even solitary brain metastases.⁹ Historical reservations regarding SRS alone in small cell lung cancer have included concerns for short interval CNS progression that could potentially lead to a decrease in overall survival with WBRT omission, as well as the paucity of data on first-line SRS in this setting.¹⁰

Glioblastoma, the most common and most rapidly progressing primary malignant tumor of the central nervous system, continues to portend a dismal prognosis, despite improvements in diagnostic and therapeutic strategies over the last 20 years. The standard of care radiographic characterization of glioblastoma is magnetic resonance imaging (MRI), which is a widely utilized examination in the diagnosis and post-treatment management of patients with glioblastoma. Basic MRI modalities available from any clinical scanner, including native T1-weighted (T1w) and contrast-enhanced (T1CE), T2-weighted (T2w), and T2-fluid-attenuated inversion recovery (T2-FLAIR) sequences, provide critical clinical information about various processes in the tumor environment. In the last decade, advanced MRI modalities are increasingly utilized to further characterize glioblastomas more comprehensively. These include multi-parametric MRI sequences, such as dynamic susceptibility contrast (DSC), dynamic contrast enhancement (DCE), higher order diffusion techniques such as diffusion tensor imaging (DTI), and MR spectroscopy (MRS). Significant efforts are ongoing to implement these advanced imaging modalities into improved clinical workflows and personalized therapy approaches. Functional MRI (fMRI) and tractography are increasingly being used to identify eloquent cortices and important tracts to minimize postsurgical neuro-deficits. A contemporary review of the application of standard and advanced MRI in clinical neuro-oncologic practice is presented here.