Segmentation of glioblastomas in early post-operative multi-modal MRI with deep neural networks.

Extent of resection after surgery is one of the main prognostic factors for patients diagnosed with glioblastoma. To achieve this, accurate segmentation and classification of residual tumor from post-operative MR images is essential. The current standard method for estimating it is subject to high inter- and intra-rater variability, and an automated method for segmentation of residual tumor in early post-operative MRI could lead to a more accurate estimation of extent of resection. In this study, two state-of-the-art neural network architectures for pre-operative segmentation were trained for the task. The models were extensively validated on a multicenter dataset with nearly 1000 patients, from 12 hospitals in Europe and the United States. The best performance achieved was a 61% Dice score, and the best classification performance was about 80% balanced accuracy, with a demonstrated ability to generalize across hospitals. In addition, the segmentation performance of the best models was on par with human expert raters. The predicted segmentations can be used to accurately classify the patients into those with residual tumor, and those with gross total resection.

Fast intraoperative histology-based diagnosis of gliomas with third harmonic generation microscopy and deep learning.

Management of gliomas requires an invasive treatment strategy, including extensive surgical resection. The objective of the neurosurgeon is to maximize tumor removal while preserving healthy brain tissue. However, the lack of a clear tumor boundary hampers the neurosurgeon's ability to accurately detect and resect infiltrating tumor tissue. Nonlinear multiphoton microscopy, in particular higher harmonic generation, enables label-free imaging of excised brain tissue, revealing histological hallmarks within seconds. Here, we demonstrate a real-time deep learning-based pipeline for automated glioma image analysis, matching video-rate image acquisition. We used a custom noise detection scheme, and a fully-convolutional classification network, to achieve on average 79% binary accuracy, 0.77 AUC and 0.83 mean average precision compared to the consensus of three pathologists, on a preliminary dataset. We conclude that the combination of real-time imaging and image analysis shows great potential for intraoperative assessment of brain tissue during tumor surgery.

Predictors of Epileptic Seizures and Ability to Work in Supratentorial Cavernous Angioma Located Within Eloquent Brain Areas.

BACKGROUND: The postoperative outcomes and the predictors of seizure control are poorly studied for supratentorial cavernous angiomas (CA) within or close to the eloquent brain area. OBJECTIVE: To assess the predictors of preoperative seizure control, postoperative seizure control, and postoperative ability to work, and the safety of the surgery. METHODS: Multicenter international retrospective cohort analysis of adult patients benefitting from a functional-based surgical resection with intraoperative functional brain mapping for a supratentorial CA within or close to eloquent brain areas. RESULTS: A total of 109 patients (66.1% women; mean age 38.4 ± 12.5 yr), were studied. Age >38 yr (odds ratio [OR], 7.33; 95% confidence interval [CI], 1.53-35.19; P = .013) and time to surgery > 12 mo (OR, 18.21; 95% CI, 1.11-296.55; P = .042) are independent predictors of uncontrolled seizures at the time of surgery. Focal deficit (OR, 10.25; 95% CI, 3.16-33.28; P < .001) is an independent predictor of inability to work at the time of surgery. History of epileptic seizures at the time of surgery (OR, 7.61; 95% CI, 1.67-85.42; P = .003) and partial resection of the CA and/or of the hemosiderin rim (OR, 12.02; 95% CI, 3.01-48.13; P < .001) are independent predictors of uncontrolled seizures postoperatively. Inability to work at the time of surgery (OR, 19.54; 95% CI, 1.90-425.48; P = .050), Karnofsky Performance Status ≤ 70 (OR, 51.20; 95% CI, 1.20-2175.37; P = .039), uncontrolled seizures postoperatively (OR, 105.33; 95% CI, 4.32-2566.27; P = .004), and worsening of cognitive functions postoperatively (OR, 13.71; 95% CI, 1.06-176.66; P = .045) are independent predictors of inability to work postoperatively. CONCLUSION: The functional-based resection using intraoperative functional brain mapping allows safe resection of CA and the peripheral hemosiderin rim located within or close to eloquent brain areas.

Early postoperative MRI overestimates residual tumour after resection of gliomas with no or minimal enhancement.

BACKGROUND: Standards for residual tumour measurement after resection of gliomas with no or minimal enhancement have not yet been established. In this study residual volumes on early and late postoperative T2-/FLAIR-weighted MRI are compared. METHODS: A retrospective cohort included 58 consecutive glioma patients with no or minimal preoperative gadolinium enhancement. Inclusion criteria were first-time resection between 2007 and 2009 with a T2-/FLAIR-based target volume and availability of preoperative, early (<48 h) and late (1-7 months) postoperative MRI. The volumes of non-enhancing T2/FLAIR tissue and diffusion restriction areas were measured. RESULTS: Residual tumour volumes were 22% smaller on late postoperative compared with early postoperative T2-weighted MRI and 49% smaller for FLAIR-weighted imaging. Postoperative restricted diffusion volume correlated with the difference between early and late postoperative FLAIR volumes and with the difference between T2 and FLAIR volumes on early postoperative MRI. CONCLUSION: We observed a systematic and substantial overestimation of residual non-enhancing volume on MRI within 48 h of resection compared with months postoperatively, in particular for FLAIR imaging. Resection-induced ischaemia contributes to this overestimation, as may other operative effects. This indicates that early postoperative MRI is less reliable to determine the extent of non-enhancing residual glioma and restricted diffusion volumes are imperative.

[Course of disease and treatment policy in suspected low-grade glioma]. / Beloop en beleid bij vermoeden van een laaggradig glioom.

Watchful waiting has long been justified in the Netherlands for patients in whom a low-grade glioma is suspected. According to recent advances in knowledge it is clear that the course of a suspected low-grade glioma cannot be reliably determined by clinical characteristics, imaging or biopsy. Early resection of the tumour provides a histological diagnosis, the possibility of removing a source of epilepsy and postponement of tumour growth and progression. Alleviation of symptoms, sustained quality of life and cognition are at least as important an aim of treatment as survival and postponement of tumour progression. In our opinion, early resection should be strongly considered in every patient with a suspected low-grade glioma. However, radiotherapy or chemotherapy should only be considered early in the presence of unfavourable prognostic factors or persistent epilepsy. Each patient in whom a low-grade glioma is suspected should receive specific treatment advice from a neuro-oncological team.