Introduction to Deep Learning in Clinical Neuroscience.

The use of deep learning (DL) is rapidly increasing in clinical neuroscience. The term denotes models with multiple sequential layers of learning algorithms, architecturally similar to neural networks of the brain. We provide examples of DL in analyzing MRI data and discuss potential applications and methodological caveats.Important aspects are data pre-processing, volumetric segmentation, and specific task-performing DL methods, such as CNNs and AEs. Additionally, GAN-expansion and domain mapping are useful DL techniques for generating artificial data and combining several smaller datasets.We present results of DL-based segmentation and accuracy in predicting glioma subtypes based on MRI features. Dice scores range from 0.77 to 0.89. In mixed glioma cohorts, IDH mutation can be predicted with a sensitivity of 0.98 and specificity of 0.97. Results in test cohorts have shown improvements of 5-7% in accuracy, following GAN-expansion of data and domain mapping of smaller datasets.The provided DL examples are promising, although not yet in clinical practice. DL has demonstrated usefulness in data augmentation and for overcoming data variability. DL methods should be further studied, developed, and validated for broader clinical use. Ultimately, DL models can serve as effective decision support systems, and are especially well-suited for time-consuming, detail-focused, and data-ample tasks.

Variations in the management of diffuse low-grade gliomas-A Scandinavian multicenter study.

BACKGROUND: Early extensive surgery is a cornerstone in treatment of diffuse low-grade gliomas (DLGGs), and an additional survival benefit has been demonstrated from early radiochemotherapy in selected "high-risk" patients. Still, there are a number of controversies related to DLGG management. The objective of this multicenter population-based cohort study was to explore potential variations in diagnostic work-up and treatment between treating centers in 2 Scandinavian countries with similar public health care systems. METHODS: Patients screened for inclusion underwent primary surgery of a histopathologically verified diffuse WHO grade II glioma in the time period 2012 through 2017. Clinical and radiological data were collected from medical records and locally conducted research projects, whereupon differences between countries and inter-hospital variations were explored. RESULTS: A total of 642 patients were included (male:female ratio 1:4), and annual age-standardized incidence rates were 0.9 and 0.8 per 100 000 in Norway and Sweden, respectively. Considerable inter-hospital variations were observed in preoperative work-up, tumor diagnostics, surgical strategies, techniques for intraoperative guidance, as well as choice and timing of adjuvant therapy. CONCLUSIONS: Despite geographical population-based case selection, similar health care organizations, and existing guidelines, there were considerable variations in DLGG management. While some can be attributed to differences in clinical implementation of current scientific knowledge, some of the observed inter-hospital variations reflect controversies related to diagnostics and treatment. Quantification of these disparities renders possible identification of treatment patterns associated with better or worse outcomes and may thus represent a step toward more uniform evidence-based care.

Shared decision-making in neurosurgery: a scoping review.

BACKGROUND: In modern neurosurgery, there are often several treatment alternatives, with different risks and benefits. Shared decision-making (SDM) has gained interest during the last decade, although SDM in the neurosurgical field is not widely studied. Therefore, the aim of this scoping review was to present the current landscape of SDM in neurosurgery. METHODS: A literature review was carried out in PubMed and Scopus. We used a search strategy based on keywords used in existing literature on SDM in neurosurgery. Full-text, peer-reviewed articles published from 2000 up to the search date February 16, 2021, with patients 18 years and older were included if articles evaluated SDM in neurosurgery from the patient's perspective. RESULTS: We identified 22 articles whereof 7 covered vestibular schwannomas, 7 covered spinal surgery, and 4 covered gliomas. The other topics were brain metastases, benign brain lesions, Parkinson's disease and evaluation of neurosurgical care. Different methods were used, with majority using forms, questionnaires, or interviews. Effects of SDM interventions were studied in 6 articles; the remaining articles explored factors influencing patients' decisions or discussed SDM aids. CONCLUSION: SDM is a tool to involve patients in the decision-making process and considers patients' preferences and what the patients find important. This scoping review illustrates the relative lack of SDM in the neurosurgical literature. Even though results indicate potential benefit of SDM, the extent of influence on treatment, outcome, and patient's satisfaction is still unknown. Finally, the use of decision aids may be a meaningful contribution to the SDM process.

Domain Mapping and Deep Learning from Multiple MRI Clinical Datasets for Prediction of Molecular Subtypes in Low Grade Gliomas.

Brain tumors, such as low grade gliomas (LGG), are molecularly classified which require the surgical collection of tissue samples. The pre-surgical or non-operative identification of LGG molecular type could improve patient counseling and treatment decisions. However, radiographic approaches to LGG molecular classification are currently lacking, as clinicians are unable to reliably predict LGG molecular type using magnetic resonance imaging (MRI) studies. Machine learning approaches may improve the prediction of LGG molecular classification through MRI, however, the development of these techniques requires large annotated data sets. Merging clinical data from different hospitals to increase case numbers is needed, but the use of different scanners and settings can affect the results and simply combining them into a large dataset often have a significant negative impact on performance. This calls for efficient domain adaption methods. Despite some previous studies on domain adaptations, mapping MR images from different datasets to a common domain without affecting subtitle molecular-biomarker information has not been reported yet. In this paper, we propose an effective domain adaptation method based on Cycle Generative Adversarial Network (CycleGAN). The dataset is further enlarged by augmenting more MRIs using another GAN approach. Further, to tackle the issue of brain tumor segmentation that requires time and anatomical expertise to put exact boundary around the tumor, we have used a tight bounding box as a strategy. Finally, an efficient deep feature learning method, multi-stream convolutional autoencoder (CAE) and feature fusion, is proposed for the prediction of molecular subtypes (1p/19q-codeletion and IDH mutation). The experiments were conducted on a total of 161 patients consisting of FLAIR and T1 weighted with contrast enhanced (T1ce) MRIs from two different institutions in the USA and France. The proposed scheme is shown to achieve the test accuracy of 74 . 81 % on 1p/19q codeletion and 81 . 19 % on IDH mutation, with marked improvement over the results obtained without domain mapping. This approach is also shown to have comparable performance to several state-of-the-art methods.