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.
Computational competitions are the standard for benchmarking medical image analysis algorithms, but they typically use small curated test datasets acquired at a few centers, leaving a gap to the reality of diverse multicentric patient data. To this end, the Federated Tumor Segmentation (FeTS) Challenge represents the paradigm for real-world algorithmic performance evaluation. The FeTS challenge is a competition to benchmark (i) federated learning aggregation algorithms and (ii) state-of-the-art segmentation algorithms, across multiple international sites. Weight aggregation and client selection techniques were compared using a multicentric brain tumor dataset in realistic federated learning simulations, yielding benefits for adaptive weight aggregation, and efficiency gains through client sampling. Quantitative performance evaluation of state-of-the-art segmentation algorithms on data distributed internationally across 32 institutions yielded good generalization on average, albeit the worst-case performance revealed data-specific modes of failure. Similar multi-site setups can help validate the real-world utility of healthcare AI algorithms in the future.
The International Subarachnoid Aneurysm Trial (ISAT) showed superiority for endovascular treatment of ruptured aneurysms and technology has since moved on rapidly. Many approaches and technology now exist for the endovascular management of ruptured and unruptured intracranial aneurysms, which reflects their varied nature – there is no one-size-fits-all technique. We aim to provide an overview of the various classes of device and the major developments over the past decade. Coiling is the oldest of the technology and continues to demonstrate high levels of occlusion and acceptable risks, making it the default treatment choice, particularly in the acutely ruptured aneurysm setting. Advances on coiling include the use of adjuncts such as balloons, stents and fully retrievable temporary neck-bridging devices, which have facilitated the treatment of more complex aneurysms. Flow divertors have also revolutionised complex aneurysm treatment with small added risk in acute aneurysm treatment and seek to remodel the aneurysm–vessel interface without accessing the aneurysm sac. The latest development and most promising avenue appears to be intrasaccular flow disrupting devices like WEB, Contour and Neqstent that provide excellent opportunities to treat wide neck complex aneurysm with minimal mortality and morbidity and good occlusion rates and may in future replace a significant number of stent-assisted coiling too.
Objectives: Radiomics is a promising avenue in non-invasive characterisation of diffuse glioma. Clinical translation is hampered by lack of reproducibility across centres and difficulty in standardising image intensity in MRI datasets. The study aim was to perform a systematic review of different methods of MRI intensity standardisation prior to radiomic feature extraction. Methods: MEDLINE, EMBASE, and SCOPUS were searched for articles meeting the following eligibility criteria: MRI radiomic studies where one method of intensity normalisation was compared with another or no normalisation, and original research concerning patients diagnosed with diffuse gliomas. Using PRISMA criteria, data were extracted from short-listed studies including number of patients, MRI sequences, validation status, radiomics software, method of segmentation, and intensity standardisation. QUADAS-2 was used for quality appraisal. Results: After duplicate removal, 741 results were returned from database and reference searches and, from these, 12 papers were eligible. Due to a lack of common pre-processing and different analyses, a narrative synthesis was sought. Three different intensity standardisation techniques have been studied: histogram matching (5/12), limiting or rescaling signal intensity (8/12), and deep learning (1/12)—only two papers compared different methods. From these studies, histogram matching produced the more reliable features compared to other methods of altering MRI signal intensity. Conclusion: Multiple methods of intensity standardisation have been described in the literature without clear consensus. Further research that directly compares different methods of intensity standardisation on glioma MRI datasets is required. Key Points: • Intensity standardisation is a key pre-processing step in the development of robust radiomic signatures to evaluate diffuse glioma. • A minority of studies compared the impact of two or more methods. • Further research is required to directly compare multiple methods of MRI intensity standardisation on glioma datasets.
17 May 2022 in Physics and Imaging in Radiation Oncology
Magnetic Resonance Imaging (MRI) exhibits scanner dependent contrast, which limits generalisability of radiomics and machine-learning for radiation oncology. Current deep-learning harmonisation requires paired data, retraining for new scanners and often suffers from geometry-shift which alters anatomical information. The aim of this study was to investigate style-blind auto-encoders for MRI harmonisation to accommodate unpaired training data, avoid geometry-shift and harmonise data from previously unseen scanners. A style-blind auto-encoder, using adversarial classification on the latent-space, was designed for MRI harmonisation. The public CC359 T1-w MRI brain dataset includes six scanners (three manufacturers, two field strengths), of which five were used for training. MRI from all six (including one unseen) scanner were harmonised to common contrast. Harmonisation extent was quantified via Kolmogorov-Smirnov testing of residual scanner dependence of 3D radiomic features, and compared to WhiteStripe normalisation. Anatomical content preservation was measured through change in structural similarity index on contrast-cycling (δSSIM). The percentage of radiomics features showing statistically significant scanner-dependence was reduced from 41% (WhiteStripe) to 16% for white matter and from 39% to 27% for grey matter. δSSIM < 0.0025 on harmonisation and de-harmonisation indicated excellent anatomical content preservation. Our method harmonised MRI contrast effectively, preserved critical anatomical details at high fidelity, trained on unpaired data and allowed zero-shot harmonisation. Robust and clinically translatable harmonisation of MRI will enable generalisable radiomic and deep-learning models for a range of applications, including radiation oncology treatment stratification, planning and response monitoring.
Glioblastoma (GBM) has the typical radiological appearance (TRA) of a centrally necrotic, peripherally enhancing tumor with surrounding edema. The objective of this study was to determine whether the developing GBM displays a spectrum of imaging changes detectable on routine clinical imaging prior to TRA GBM. Patients with pre-operative imaging diagnosed with GBM (1 January 2014–31 March 2022) were identified from a neuroscience center. The imaging was reviewed by an experienced neuroradiologist. Imaging patterns preceding TRA GBM were analyzed. A total of 76 out of 555 (14%) patients had imaging preceding TRA GBM, 57 had solitary lesions, and 19 had multiple lesions (total = 84 lesions). Here, 83% of the lesions had cortical or cortical/subcortical locations. The earliest imaging features for 84 lesions were T2 hyperintensity/CT low density (n = 18), CT hyperdensity (n = 51), and T2 iso-intensity (n = 15). Lesions initially showing T2 hyperintensity/CT low density later showed T2 iso-intensity. When CT and MRI were available, all CT hyperdense lesions showed T2 iso-intensity, reduced diffusivity, and the following enhancement patterns: nodular 35%, solid 29%, none 26%, and patchy peripheral 10%. The mean time to develop TRA GBM from T2 hyperintensity was 140 days and from CT hyperdensity was 69 days. This research suggests that the developing GBM shows a spectrum of imaging features, progressing through T2 hyperintensity to CT hyperdensity, T2 iso-intensity, reduced diffusivity, and variable enhancement to TRA GBM. Red flags for non-TRA GBM lesions are cortical/subcortical CT hyperdense/T2 iso-intense/low ADC. Future research correlating this imaging spectrum with pathophysiology may provide insight into GBM growth patterns.