Deep-learning-based reconstruction of undersampled MRI to reduce scan times: a multicentre, retrospective, cohort study.

BACKGROUND: The extended acquisition times required for MRI limit its availability in resource-constrained settings. Consequently, accelerating MRI by undersampling k-space data, which is necessary to reconstruct an image, has been a long-standing but important challenge. We aimed to develop a deep convolutional neural network (dCNN) optimisation method for MRI reconstruction and to reduce scan times and evaluate its effect on image quality and accuracy of oncological imaging biomarkers. METHODS: In this multicentre, retrospective, cohort study, MRI data from patients with glioblastoma treated at Heidelberg University Hospital (775 patients and 775 examinations) and from the phase 2 CORE trial (260 patients, 1083 examinations, and 58 institutions) and the phase 3 CENTRIC trial (505 patients, 3147 examinations, and 139 institutions) were used to develop, train, and test dCNN for reconstructing MRI from highly undersampled single-coil k-space data with various acceleration rates (R=2, 4, 6, 8, 10, and 15). Independent testing was performed with MRIs from the phase 2/3 EORTC-26101 trial (528 patients with glioblastoma, 1974 examinations, and 32 institutions). The similarity between undersampled dCNN-reconstructed and original MRIs was quantified with various image quality metrics, including structural similarity index measure (SSIM) and the accuracy of undersampled dCNN-reconstructed MRI on downstream radiological assessment of imaging biomarkers in oncology (automated artificial intelligence-based quantification of tumour burden and treatment response) was performed in the EORTC-26101 test dataset. The public NYU Langone Health fastMRI brain test dataset (558 patients and 558 examinations) was used to validate the generalisability and robustness of the dCNN for reconstructing MRIs from available multi-coil (parallel imaging) k-space data. FINDINGS: In the EORTC-26101 test dataset, the median SSIM of undersampled dCNN-reconstructed MRI ranged from 0·88 to 0·99 across different acceleration rates, with 0·92 (95% CI 0·92-0·93) for 10-times acceleration (R=10). The 10-times undersampled dCNN-reconstructed MRI yielded excellent agreement with original MRI when assessing volumes of contrast-enhancing tumour (median DICE for spatial agreement of 0·89 [95% CI 0·88 to 0·89]; median volume difference of 0·01 cm3 [95% CI 0·00 to 0·03] equalling 0·21%; p=0·0036 for equivalence) or non-enhancing tumour or oedema (median DICE of 0·94 [95% CI 0·94 to 0·95]; median volume difference of -0·79 cm3 [95% CI -0·87 to -0·72] equalling -1·77%; p=0·023 for equivalence) in the EORTC-26101 test dataset. Automated volumetric tumour response assessment in the EORTC-26101 test dataset yielded an identical median time to progression of 4·27 months (95% CI 4·14 to 4·57) when using 10-times-undersampled dCNN-reconstructed or original MRI (log-rank p=0·80) and agreement in the time to progression in 374 (95·2%) of 393 patients with data. The dCNN generalised well to the fastMRI brain dataset, with significant improvements in the median SSIM when using multi-coil compared with single-coil k-space data (p<0·0001). INTERPRETATION: Deep-learning-based reconstruction of undersampled MRI allows for a substantial reduction of scan times, with a 10-times acceleration demonstrating excellent image quality while preserving the accuracy of derived imaging biomarkers for the assessment of oncological treatment response. Our developments are available as open source software and hold considerable promise for increasing the accessibility to MRI, pending further prospective validation. FUNDING: Deutsche Forschungsgemeinschaft (German Research Foundation) and an Else Kröner Clinician Scientist Endowed Professorship by the Else Kröner Fresenius Foundation.

Deep Learning-based Identification of Brain MRI Sequences Using a Model Trained on Large Multicentric Study Cohorts.

Purpose To develop a fully automated device- and sequence-independent convolutional neural network (CNN) for reliable and high-throughput labeling of heterogeneous, unstructured MRI data. Materials and Methods Retrospective, multicentric brain MRI data (2179 patients with glioblastoma, 8544 examinations, 63 327 sequences) from 249 hospitals and 29 scanner types were used to develop a network based on ResNet-18 architecture to differentiate nine MRI sequence types, including T1-weighted, postcontrast T1-weighted, T2-weighted, fluid-attenuated inversion recovery, susceptibility-weighted, apparent diffusion coefficient, diffusion-weighted (low and high b value), and gradient-recalled echo T2*-weighted and dynamic susceptibility contrast-related images. The two-dimensional-midsection images from each sequence were allocated to training or validation (approximately 80%) and testing (approximately 20%) using a stratified split to ensure balanced groups across institutions, patients, and MRI sequence types. The prediction accuracy was quantified for each sequence type, and subgroup comparison of model performance was performed using χ2 tests. Results On the test set, the overall accuracy of the CNN (ResNet-18) ensemble model among all sequence types was 97.9% (95% CI: 97.6, 98.1), ranging from 84.2% for susceptibility-weighted images (95% CI: 81.8, 86.6) to 99.8% for T2-weighted images (95% CI: 99.7, 99.9). The ResNet-18 model achieved significantly better accuracy compared with ResNet-50 despite its simpler architecture (97.9% vs 97.1%; P ≤ .001). The accuracy of the ResNet-18 model was not affected by the presence versus absence of tumor on the two-dimensional-midsection images for any sequence type (P > .05). Conclusion The developed CNN (www.github.com/neuroAI-HD/HD-SEQ-ID) reliably differentiates nine types of MRI sequences within multicenter and large-scale population neuroimaging data and may enhance the speed, accuracy, and efficiency of clinical and research neuroradiologic workflows. Keywords: MR-Imaging, Neural Networks, CNS, Brain/Brain Stem, Computer Applications-General (Informatics), Convolutional Neural Network (CNN), Deep Learning Algorithms, Machine Learning Algorithms Supplemental material is available for this article. © RSNA, 2023.

Cerebrospinal fluid cfDNA sequencing for classification of central nervous system glioma.

PURPOSE: Primary central nervous system (CNS) gliomas can be classified by characteristic genetic alterations. In addition to solid tissue obtained via surgery or biopsy, cell-free DNA (cfDNA) from cerebrospinal fluid (CSF) is an alternative source of material for genomic analyses. EXPERIMENTAL DESIGN: We performed targeted next-generation sequencing (NGS) of CSF cfDNA in a representative cohort of 85 patients presenting at two neurooncological centers with suspicion of primary or recurrent glioma. Copy-number variation (CNV) profiles, single nucleotide variants (SNVs), and small insertions/ deletions (indels) were combined into a molecular-guided tumor classification. Comparison with the solid tumor was performed for 38 cases with matching solid tissue available. RESULTS: Cases were stratified into four groups: glioblastoma (n = 32), other glioma (n = 19), non-malignant (n = 17) and non-diagnostic (n = 17). We introduced a molecular-guided tumor classification, which enabled identification of tumor entities and/ or cancer specific alterations in 75.0 % (n = 24) of glioblastoma and 52.6 % (n = 10) of other glioma cases. The overlap between CSF and matching solid tissue was highest for CNVs (26-48 %) and SNVs at pre-defined gene loci (44 %), followed by SNVs/ indels identified via uninformed variant calling (8-14 %). A molecular-guided tumor classification was possible for 23.5 % (n = 4) of non-diagnostic cases. CONCLUSIONS: We developed a targeted sequencing workflow for CSF cfDNA as well as a strategy for interpretation and reporting of sequencing results based on a molecular-guided tumor classification in glioma.

Deep-learning based detection of vessel occlusions on CT-angiography in patients with suspected acute ischemic stroke.

Swift diagnosis and treatment play a decisive role in the clinical outcome of patients with acute ischemic stroke (AIS), and computer-aided diagnosis (CAD) systems can accelerate the underlying diagnostic processes. Here, we developed an artificial neural network (ANN) which allows automated detection of abnormal vessel findings without any a-priori restrictions and in <2 minutes. Pseudo-prospective external validation was performed in consecutive patients with suspected AIS from 4 different hospitals during a 6-month timeframe and demonstrated high sensitivity (≥87%) and negative predictive value (≥93%). Benchmarking against two CE- and FDA-approved software solutions showed significantly higher performance for our ANN with improvements of 25-45% for sensitivity and 4-11% for NPV (p ≤ 0.003 each). We provide an imaging platform ( https://stroke.neuroAI-HD.org ) for online processing of medical imaging data with the developed ANN, including provisions for data crowdsourcing, which will allow continuous refinements and serve as a blueprint to build robust and generalizable AI algorithms.

Dynamics of cerebral perfusion and oxygenation parameters following endovascular treatment of acute ischemic stroke.

BACKGROUND: We studied the effects of endovascular treatment (EVT) and the impact of the extent of recanalization on cerebral perfusion and oxygenation parameters in patients with acute ischemic stroke (AIS) and large vessel occlusion (LVO). METHODS: Forty-seven patients with anterior LVO underwent computed tomography perfusion (CTP) before and immediately after EVT. The entire ischemic region (Tmax >6 s) was segmented before intervention, and tissue perfusion (time-to-maximum (Tmax), time-to-peak (TTP), mean transit time (MTT), cerebral blood volume (CBV), cerebral blood flow (CBF)) and oxygenation (coefficient of variation (COV), capillary transit time heterogeneity (CTH), metabolic rate of oxygen (CMRO2), oxygen extraction fraction (OEF)) parameters were quantified from the segmented area at baseline and the corresponding area immediately after intervention, as well as within the ischemic core and penumbra. The impact of the extent of recanalization (modified Treatment in Cerebral Infarction (mTICI)) on CTP parameters was assessed with the Wilcoxon test and Pearson's correlation coefficients. RESULTS: The Tmax, MTT, OEF and CTH values immediately after EVT were lower in patients with complete (as compared with incomplete) recanalization, whereas CBF and COV values were higher (P<0.05) and no differences were found in other parameters. The ischemic penumbra immediately after EVT was lower in patients with complete recanalization as compared with those with incomplete recanalization (P=0.002), whereas no difference was found for the ischemic core (P=0.12). Specifically, higher mTICI scores were associated with a greater reduction of ischemic penumbra volumes (R²=-0.48 (95% CI -0.67 to -0.22), P=0.001) but not of ischemic core volumes (P=0.098). CONCLUSIONS: Our study demonstrates that the ischemic penumbra is the key target of successful EVT in patients with AIS and largely determines its efficacy on a tissue level. Furthermore, we confirm the validity of the mTICI score as a surrogate parameter of interventional success on a tissue perfusion level.

Posterior insular activity contributes to the late laser-evoked potential component in EEG recordings.

OBJECTIVE: Nociceptive activity in some brain areas has concordantly been reported in EEG source models, such as the anterior/mid-cingulate cortex and the parasylvian area. Whereas the posterior insula has been constantly reported to be active in intracortical and fMRI studies, non-invasive EEG and MEG recordings mostly failed to detect activity in this region. This study aimed to determine an appropriate inverse modeling approach in EEG recordings to model posterior insular activity, assuming the late LEP (laser evoked potential) time window to yield a better separation from other ongoing cortical activity. METHODS: In 12 healthy volunteers, nociceptive stimuli of three intensities were applied. LEP were recorded using 32-channel EEG recordings. Source analysis was performed in specific time windows defined in the grand-average dataset. Two distinct dipole-pairs located close to the operculo-insular area were compared. RESULTS: Our results show that posterior insular activity yields a substantial contribution to the latest part (positive component) of the LEP. CONCLUSIONS: Even though the initial insular activity onset is in the early LEP time window,modelingthe insular activity in the late LEP time window might result in better separation from other ongoing cortical activity. SIGNIFICANCE: Modeling the late LEP activity might enable to distinguish posterior insular activity.