Establishing recombinant production of pediocin PA-1 in Corynebacterium glutamicum.

Bacteriocins are antimicrobial peptides produced by bacteria to inhibit competitors in their natural environments. Some of these peptides have emerged as commercial food preservatives and, due to the rapid increase in antibiotic resistant bacteria, are also discussed as interesting alternatives to antibiotics for therapeutic purposes. Currently, commercial bacteriocins are produced exclusively with natural producer organisms on complex substrates and are sold as semi-purified preparations or crude fermentates. To allow clinical application, efficacy of production and purity of the product need to be improved. This can be achieved by shifting production to recombinant microorganisms. Here, we identify Corynebacterium glutamicum as a suitable production host for the bacteriocin pediocin PA-1. C. glutamicum CR099 shows resistance to high concentrations of pediocin PA-1 and the bacteriocin was not inactivated when spiked into growing cultures of this bacterium. Recombinant C. glutamicum expressing a synthetic pedACDCgl operon releases a compound that has potent antimicrobial activity against Listeria monocytogenes and Listeria innocua and matches size and mass:charge ratio of commercial pediocin PA-1. Fermentations in shake flasks and bioreactors suggest that low levels of dissolved oxygen are favorable for production of pediocin. Under these conditions, however, reduced activity of the TCA cycle resulted in decreased availability of the important pediocin precursor l-asparagine suggesting options for further improvement. Overall, we demonstrate that C. glutamicum is a suitable host for recombinant production of bacteriocins of the pediocin family.

Anatomo-Functional Mapping of the Primate Mesencephalic Locomotor Region Using Stereotactic Lesions.

BACKGROUND: Dysfunction of the mesencephalic locomotor region has been implicated in gait disorders. However, the role of its 2 components, the pedunculopontine and the cuneiform nuclei, in locomotion is poorly understood in primates. OBJECTIVES: To analyze the effect of cuneiform lesions on gait and balance in 2 monkeys and to compare them with those obtained after cholinergic pedunculopontine lesions in 4 monkeys and after lesions in both the cuneiform and pedunculopontine nuclei in 1 monkey. METHODS: After each stereotactic lesion, we performed a neurological examination and gait and balance assessments with kinematic measures during a locomotor task. The 3-dimensional location of each lesion was analyzed on a common brainstem space. RESULTS: After each cuneiform lesion, we observed a contralateral cervical dystonia including an increased tone in the proximal forelimb and an increase in knee angle, back curvature and walking speed. Conversely, cholinergic pedunculopontine lesions increased tail rigidity and back curvature and an imbalance of the muscle tone between the ipsi- and contralateral hindlimb with decreased knee angles. The walking speed was decreased. Moreover, pedunculopontine lesions often resulted in a longer time to waking postsurgery. CONCLUSIONS: The location of the lesions and their behavioral effects revealed a somatotopic organization of muscle tone control, with the neck and forelimb represented within the cuneiform nucleus and hindlimb and tail represented within the pedunculopontine nucleus. Cuneiform lesions increased speed, whereas pedunculopontine lesions decreased it. These findings confirm the complex and specific role of the cuneiform and pedunculopontine nuclei in locomotion and suggest the role of the pedunculopontine in sleep control. © 2020 International Parkinson and Movement Disorder Society.

Normal and pathological neuronal distribution of the human mesencephalic locomotor region.

BACKGROUND: Deep brain stimulation of the pedunculopontine nucleus has been performed to treat dopamine-resistant gait and balance disorders in patients with degenerative diseases. The outcomes, however, are variable, which may be the result of the lack of a well-defined anatomical target. OBJECTIVES: The objectives of this study were to identify the main neuronal populations of the pedunculopontine and the cuneiform nuclei that compose the human mesencephalic locomotor region and to compare their 3-dimensional distribution with those found in patients with Parkinson's disease and progressive supranuclear palsy. METHODS: We used high-field MRI, immunohistochemistry, and in situ hybridization to characterize the distribution of the different cell types, and we developed software to merge all data within a common 3-dimensional space. RESULTS: We found that cholinergic, GABAergic, and glutamatergic neurons comprised the main cell types of the mesencephalic locomotor region, with the peak densities of cholinergic and GABAergic neurons similarly located within the rostral pedunculopontine nucleus. Cholinergic and noncholinergic neuronal losses were homogeneous in the mesencephalic locomotor region of patients, with the peak density of remaining neurons at the same location as in controls. The degree of denervation of the pedunculopontine nucleus was highest in patients with progressive supranuclear palsy, followed by Parkinson's disease patients with falls. CONCLUSIONS: The peak density of cholinergic and GABAergic neurons was located similarly within the rostral pedunculopontine nucleus not only in controls but also in pathological cases. The neuronal loss was homogeneously distributed and highest in the pedunculopontine nucleus of patients with falls, which suggests a potential pathophysiological link. © 2018 International Parkinson and Movement Disorder Society.

Multicenter validation of automated trajectories for selective laser amygdalohippocampectomy.

OBJECTIVE: Laser interstitial thermal therapy (LITT) is a novel minimally invasive alternative to open mesial temporal resection in drug-resistant mesial temporal lobe epilepsy (MTLE). The safety and efficacy of the procedure are dependent on the preplanned trajectory and the extent of the planned ablation achieved. Ablation of the mesial hippocampal head has been suggested to be an independent predictor of seizure freedom, whereas sparing of collateral structures is thought to result in improved neuropsychological outcomes. We aim to validate an automated trajectory planning platform against manually planned trajectories to objectively standardize the process. METHODS: Using the EpiNav platform, we compare automated trajectory planning parameters derived from expert opinion and machine learning to undertake a multicenter validation against manually planned and implemented trajectories in 95 patients with MTLE. We estimate ablation volumes of regions of interest and quantify the size of the avascular corridor through the use of a risk score as a marker of safety. We also undertake blinded external expert feasibility and preference ratings. RESULTS: Automated trajectory planning employs complex algorithms to maximize ablation of the mesial hippocampal head and amygdala, while sparing the parahippocampal gyrus. Automated trajectories resulted in significantly lower calculated risk scores and greater amygdala ablation percentage, whereas overall hippocampal ablation percentage did not differ significantly. In addition, estimated damage to collateral structures was reduced. Blinded external expert raters were significantly more likely to prefer automated to manually planned trajectories. SIGNIFICANCE: Retrospective studies of automated trajectory planning show much promise in improving safety parameters and ablation volumes during LITT for MTLE. Multicenter validation provides evidence that the algorithm is robust, and blinded external expert ratings indicate that the trajectories are clinically feasible. Prospective validation studies are now required to determine if automated trajectories translate into improved seizure freedom rates and reduced neuropsychological deficits.

Biotechnological production of mono- and diamines using bacteria: recent progress, applications, and perspectives.

Common plastics such as polyamides are derived typically from petroleum or natural gas. Fossil-based polyamide production often involves toxic precursors or intermediates. By contrast, bio-based polyamides offer a realistic alternative. Bio-based routes to monomeric precursors of polyamides such as diamines, dicarboxylic acids, and omega-amino acids have been developed. Recent advances in the metabolic engineering of the biotechnologically relevant Escherichia coli and Corynebacterium glutamicum for the production of monoalkylamines such as omega-amino acids as well as diamines are presented.

Improved fermentative production of gamma-aminobutyric acid via the putrescine route: Systems metabolic engineering for production from glucose, amino sugars, and xylose.

Gamma-aminobutyric acid (GABA) is a non-protein amino acid widespread in Nature. Among the various uses of GABA, its lactam form 2-pyrrolidone can be chemically converted to the biodegradable plastic polyamide-4. In metabolism, GABA can be synthesized either by decarboxylation of l-glutamate or by a pathway that starts with the transamination of putrescine. Fermentative production of GABA from glucose by recombinant Corynebacterium glutamicum has been described via both routes. Putrescine-based GABA production was characterized by accumulation of by-products such as N-acetyl-putrescine. Their formation was abolished by deletion of the spermi(di)ne N-acetyl-transferase gene snaA. To improve provision of l-glutamate as precursor 2-oxoglutarate dehydrogenase activity was reduced by changing the translational start codon of the chromosomal gene for 2-oxoglutarate dehydrogenase subunit E1o to the less preferred TTG and by maintaining the inhibitory protein OdhI in its inhibitory form by changing amino acid residue 15 from threonine to alanine. Putrescine-based GABA production by the strains described here led to GABA titers up to 63.2 g L-1 in fed-batch cultivation at maximum volumetric productivities up to 1.34 g L-1 h-1 , the highest volumetric productivity for fermentative GABA production reported to date. Moreover, GABA production from the carbon sources xylose, glucosamine, and N-acetyl-glucosamine that do not have competing uses in the food or feed industries was established. Biotechnol. Bioeng. 2017;114: 862-873. © 2016 Wiley Periodicals, Inc.

Engineering Corynebacterium glutamicum for fast production of L-lysine and L-pipecolic acid.

The Gram-positive Corynebacterium glutamicum is widely used for fermentative production of amino acids. The world production of L-lysine has surpassed 2 million tons per year. Glucose uptake and phosphorylation by C. glutamicum mainly occur by the phosphotransferase system (PTS) and to lesser extent by inositol permeases and glucokinases. Heterologous expression of the genes for the high-affinity glucose permease from Streptomyces coelicolor and Bacillus subtilis glucokinase fully compensated for the absence of the PTS in Δhpr strains. Growth of PTS-positive strains with glucose was accelerated when the endogenous inositol permease IolT2 and glucokinase from B. subtilis were overproduced with balanced translation initiation rates using plasmid pEKEx3-IolTBest. When the genome-reduced C. glutamicum strain GRLys1 carrying additional in-frame deletions of sugR and ldhA to derepress glycolytic and PTS genes and to circumvent formation of L-lactate as by-product was transformed with this plasmid or with pVWEx1-IolTBest, 18 to 20 % higher volumetric productivities and 70 to 72 % higher specific productivities as compared to the parental strain resulted. The non-proteinogenic amino acid L-pipecolic acid (L-PA), a precursor of immunosuppressants, peptide antibiotics, or piperidine alkaloids, can be derived from L-lysine. To enable production of L-PA by the constructed L-lysine-producing strain, the L-lysine 6-dehydrogenase gene lysDH from Silicibacter pomeroyi and the endogenous pyrroline 5-carboxylate reductase gene proC were overexpressed as synthetic operon. This enabled C. glutamicum to produce L-PA with a yield of 0.09 ± 0.01 g g(-1) and a volumetric productivity of 0.04 ± 0.01 g L(-1) h(-1).To the best of our knowledge, this is the first fermentative process for the production of L-PA from glucose.

Roles of export genes cgmA and lysE for the production of L-arginine and L-citrulline by Corynebacterium glutamicum.

L-arginine is a semi-essential amino acid with application in cosmetic, pharmaceutical, and food industries. Metabolic engineering strategies have been applied for overproduction of L-arginine by Corynebacterium glutamicum. LysE was the only known L-arginine exporter of this bacterium. However, an L-arginine-producing strain carrying a deletion of lysE still accumulated about 10 mM L-arginine in the growth medium. Overexpression of the putative putrescine and cadaverine export permease gene cgmA was shown to compensate for the lack of lysE with regard to L-arginine export. Moreover, plasmid-borne overexpression of cgmA rescued the toxic effect caused by feeding of the dipeptide Arg-Ala to lysE-deficient C. glutamicum and argO-deficient Escherichia coli strains. Deletion of the repressor gene cgmR improved L-arginine titers by 5 %. Production of L-lysine and L-citrulline was not affected by cgmA overexpression. Taken together, CgmA may function as an export system not only for the diamine putrescine and cadaverine but also for L-arginine. The major export system for L-lysine and L-arginine LysE may also play a role in L-citrulline export since production of L-citrulline was reduced when lysE was deleted and improved by 45 % when lysE was overproduced.

Updates on industrial production of amino acids using Corynebacterium glutamicum.

L-Amino acids find various applications in biotechnology. L-Glutamic acid and its salts are used as flavor enhancers. Other L-amino acids are used as food or feed additives, in parenteral nutrition or as building blocks for the chemical and pharmaceutical industries. L-amino acids are synthesized from precursors of central carbon metabolism. Based on the knowledge of the biochemical pathways microbial fermentation processes of food, feed and pharma amino acids have been developed. Production strains of Corynebacterium glutamicum, which has been used safely for more than 50 years in food biotechnology, and Escherichia coli are constantly improved using metabolic engineering approaches. Research towards new processes is ongoing. Fermentative production of L-amino acids in the million-ton-scale has shaped modern biotechnology and its markets continue to grow steadily. This review focusses on recent achievements in strain development for amino acid production including the use of CRISPRi/dCas9, genome-reduced strains, biosensors and synthetic pathways to enable utilization of alternative carbon sources.

Riboflavin overproduction from diverse feedstocks with engineeredCorynebacterium glutamicum.

Riboflavin overproduction byCorynebacterium glutamicumwas achieved by screening synthetic operons, enabling fine-tuned expression of the riboflavin biosynthetic genesribGCAH.The synthetic operons were designed by means of predicted translational initiation rates of each open reading frame, with the best-performing selection enabling riboflavin overproduction without negatively affecting cell growth. Overexpression of the fructose-1,6-bisphosphatase (fbp) and 5-phosphoribosyl 1-pyrophosphate aminotransferase (purF) encoding genes was then done to redirect the metabolic flux towards the riboflavin precursors. The resulting strain produced 8.3 g l-1of riboflavin in glucose-based fed-batch fermentations, which is the highest reported riboflavin titer withC. glutamicum. Further genetic engineering enabled both xylose and mannitol utilization byC. glutamicum, and we demonstrated riboflavin overproduction with the xylose-rich feedstocks rice husk hydrolysate and spent sulfite liquor, and the mannitol-rich feedstock brown seaweed hydrolysate. Remarkably, rice husk hydrolysate provided 30% higher riboflavin yields compared to glucose in the bioreactors.

MONAI Label: A framework for AI-assisted interactive labeling of 3D medical images.

The lack of annotated datasets is a major bottleneck for training new task-specific supervised machine learning models, considering that manual annotation is extremely expensive and time-consuming. To address this problem, we present MONAI Label, a free and open-source framework that facilitates the development of applications based on artificial intelligence (AI) models that aim at reducing the time required to annotate radiology datasets. Through MONAI Label, researchers can develop AI annotation applications focusing on their domain of expertise. It allows researchers to readily deploy their apps as services, which can be made available to clinicians via their preferred user interface. Currently, MONAI Label readily supports locally installed (3D Slicer) and web-based (OHIF) frontends and offers two active learning strategies to facilitate and speed up the training of segmentation algorithms. MONAI Label allows researchers to make incremental improvements to their AI-based annotation application by making them available to other researchers and clinicians alike. Additionally, MONAI Label provides sample AI-based interactive and non-interactive labeling applications, that can be used directly off the shelf, as plug-and-play to any given dataset. Significant reduced annotation times using the interactive model can be observed on two public datasets.

Metabolic engineering of thermophilic Bacillus methanolicus for riboflavin overproduction from methanol.

The growing need of next generation feedstocks for biotechnology spurs an intensification of research on the utilization of methanol as carbon and energy source for biotechnological processes. In this paper, we introduced the methanol-based overproduction of riboflavin into metabolically engineered Bacillus methanolicus MGA3. First, we showed that B. methanolicus naturally produces small amounts of riboflavin. Then, we created B. methanolicus strains overexpressing either homologous or heterologous gene clusters encoding the riboflavin biosynthesis pathway, resulting in riboflavin overproduction. Our results revealed that the supplementation of growth media with sublethal levels of chloramphenicol contributes to a higher plasmid-based riboflavin production titre, presumably due to an increase in plasmid copy number and thus biosynthetic gene dosage. Based on this, we proved that riboflavin production can be increased by exchanging a low copy number plasmid with a high copy number plasmid leading to a final riboflavin titre of about 523 mg L-1 in methanol fed-batch fermentation. The findings of this study showcase the potential of B. methanolicus as a promising host for methanol-based overproduction of extracellular riboflavin and serve as basis for metabolic engineering of next generations of riboflavin overproducing strains.

From Brown Seaweed to a Sustainable Microbial Feedstock for the Production of Riboflavin.

The increasing global demand for food and energy production encourages the development of new production strategies focused on sustainability. Often, microbial bioprocesses rely on food or feed competitive feedstocks; hence, there is a trending need for green substrates. Here, we have proven the potential of brown seaweed biomass as microbial feedstock on account of its content of mannitol and the glucose polymer laminarin. Our host, Corynebacterium glutamicum, was engineered to enable access to mannitol as a carbon source through the heterologous expression of the mannitol-specific phosphotransferase system and the mannitol-1-phosphate-5-dehydrogenase from Bacillus subtilis. Overproduction of riboflavin was coupled with mannitol and glucose consumption via constitutive overexpression of the biosynthetic riboflavin operon ribGCAH from C. glutamicum. Brown seaweed extract and brown seaweed hydrolysate from Laminaria hyperborea, containing mannitol and glucose, were used as a carbon source for flask and bioreactor fermentations. In a seaweed-based fed-batch fermentation, the riboflavin final titer, yield, and volumetric productivity values of 1,291.2 mg L-1, 66.1 mg g-1, and 17.2 mg L-1 h-1, respectively, were achieved.

Software tool for visualization of a probabilistic map of the epileptogenic zone from seizure semiologies.

Around one third of epilepsies are drug-resistant. For these patients, seizures may be reduced or cured by surgically removing the epileptogenic zone (EZ), which is the portion of the brain giving rise to seizures. If noninvasive data are not sufficiently lateralizing or localizing, the EZ may need to be localized by precise implantation of intracranial electroencephalography (iEEG) electrodes. The choice of iEEG targets is influenced by clinicians' experience and personal knowledge of the literature, which leads to substantial variations in implantation strategies across different epilepsy centers. The clinical diagnostic pathway for surgical planning could be supported and standardized by an objective tool to suggest EZ locations, based on the outcomes of retrospective clinical cases reported in the literature. We present an open-source software tool that presents clinicians with an intuitive and data-driven visualization to infer the location of the symptomatogenic zone, that may overlap with the EZ. The likely EZ is represented as a probabilistic map overlaid on the patient's images, given a list of seizure semiologies observed in that specific patient. We demonstrate a case study on retrospective data from a patient treated in our unit, who underwent resective epilepsy surgery and achieved 1-year seizure freedom after surgery. The resected brain structures identified as EZ location overlapped with the regions highlighted by our tool, demonstrating its potential utility.

Pedunculopontine and Cuneiform Nuclei Deep Brain Stimulation for Severe Gait and Balance Disorders in Parkinson's Disease: Interim Results from a Randomized Double-Blind Clinical Trial.

BACKGROUND: Dopa-resistant freezing of gait (FOG) and falls represent the dominant motor disabilities in advanced Parkinson's disease (PD). OBJECTIVE: We investigate the effects of deep brain stimulation (DBS) of the mesencephalic locomotor region (MLR), comprised of the pedunculopontine (PPN) and cuneiform (CuN) nuclei, for treating gait and balance disorders, in a randomized double-blind cross-over trial. METHODS: Six PD patients with dopa-resistant FOG and/or falls were operated for MLR-DBS. Patients received three DBS conditions, PPN, CuN, or Sham, in a randomized order for 2-months each, followed by an open-label phase. The primary outcome was the change in anteroposterior anticipatory-postural-adjustments (APAs) during gait initiation on a force platformResults:The anteroposterior APAs were not significantly different between the DBS conditions (median displacement [1st-3rd quartile] of 3.07 [3.12-4.62] cm with sham-DBS, 1.95 [2.29-3.85] cm with PPN-DBS and 2.78 [1.66-4.04] cm with CuN-DBS; p = 0.25). Step length and velocity were significantly higher with CuN-DBS vs. both sham-DBS and PPN-DBS. Conversely, step length and velocity were lower with PPN-DBS vs. sham-DBS, with greater double stance and gait initiation durations. One year after surgery, step length was significantly lower with PPN-DBS vs. inclusion. We did not find any significant change in clinical scales between DBS conditions or one year after surgery. CONCLUSION: Two months of PPN-DBS or CuN-DBS does not effectively improve clinically dopa-resistant gait and balance disorders in PD patients.

Probabilistic landscape of seizure semiology localizing values.

Semiology describes the evolution of symptoms and signs during epileptic seizures and contributes to the evaluation of individuals with focal drug-resistant epilepsy for curative resection. Semiology varies in complexity from elementary sensorimotor seizures arising from primary cortex to complex behaviours and automatisms emerging from distributed cerebral networks. Detailed semiology interpreted by expert epileptologists may point towards the likely site of seizure onset, but this process is subjective. No study has captured the variances in semiological localizing values in a data-driven manner to allow objective and probabilistic determinations of implicated networks and nodes. We curated an open data set from the epilepsy literature, in accordance with Preferred Reporting Items for Systematic Reviews and Meta-Analyses guidelines, linking semiology to hierarchical brain localizations. A total of 11 230 data points were collected from 4643 patients across 309 articles, labelled using ground truths (postoperative seizure-freedom, concordance of imaging and neurophysiology, and/or invasive EEG) and a designation method that distinguished between semiologies arising from a predefined cortical region and descriptions of neuroanatomical localizations responsible for generating a particular semiology. This allowed us to mitigate temporal lobe publication bias by filtering studies that preselected patients based on prior knowledge of their seizure foci. Using this data set, we describe the probabilistic landscape of semiological localizing values as forest plots at the resolution of seven major brain regions: temporal, frontal, cingulate, parietal, occipital, insula, and hypothalamus, and five temporal subregions. We evaluated the intrinsic value of any one semiology over all other ictal manifestations. For example, epigastric auras implicated the temporal lobe with 83% probability when not accounting for the publication bias that favoured temporal lobe epilepsies. Unbiased results for a prior distribution of cortical localizations revised the prevalence of temporal lobe epilepsies from 66% to 44%. Therefore, knowledge about the presence of epigastric auras updates localization to the temporal lobe with an odds ratio (OR) of 2.4 [CI95% (1.9, 2.9); and specifically, mesial temporal structures OR: 2.8 (2.3, 2.9)], attesting the value of epigastric auras. As a further example, although head version is thought to implicate the frontal lobes, it did not add localizing value compared with the prior distribution of cortical localizations [OR: 0.9 (0.7, 1.2)]. Objectification of the localizing values of the 12 most common semiologies provides a complementary view of brain dysfunction to that of lesion-deficit mappings, as instead of linking brain regions to phenotypic-deficits, semiological phenotypes are linked back to brain sources. This work enables coupling of seizure propagation with ictal manifestations, and clinical support algorithms for localizing seizure phenotypes.

TorchIO: A Python library for efficient loading, preprocessing, augmentation and patch-based sampling of medical images in deep learning.

BACKGROUND AND OBJECTIVE: Processing of medical images such as MRI or CT presents different challenges compared to RGB images typically used in computer vision. These include a lack of labels for large datasets, high computational costs, and the need of metadata to describe the physical properties of voxels. Data augmentation is used to artificially increase the size of the training datasets. Training with image subvolumes or patches decreases the need for computational power. Spatial metadata needs to be carefully taken into account in order to ensure a correct alignment and orientation of volumes. METHODS: We present TorchIO, an open-source Python library to enable efficient loading, preprocessing, augmentation and patch-based sampling of medical images for deep learning. TorchIO follows the style of PyTorch and integrates standard medical image processing libraries to efficiently process images during training of neural networks. TorchIO transforms can be easily composed, reproduced, traced and extended. Most transforms can be inverted, making the library suitable for test-time augmentation and estimation of aleatoric uncertainty in the context of segmentation. We provide multiple generic preprocessing and augmentation operations as well as simulation of MRI-specific artifacts. RESULTS: Source code, comprehensive tutorials and extensive documentation for TorchIO can be found at http://torchio.rtfd.io/. The package can be installed from the Python Package Index (PyPI) running pip install torchio. It includes a command-line interface which allows users to apply transforms to image files without using Python. Additionally, we provide a graphical user interface within a TorchIO extension in 3D Slicer to visualize the effects of transforms. CONCLUSION: TorchIO was developed to help researchers standardize medical image processing pipelines and allow them to focus on the deep learning experiments. It encourages good open-science practices, as it supports experiment reproducibility and is version-controlled so that the software can be cited precisely. Due to its modularity, the library is compatible with other frameworks for deep learning with medical images.

Machine Learning for Localizing Epileptogenic-Zone in the Temporal Lobe: Quantifying the Value of Multimodal Clinical-Semiology and Imaging Concordance.

Background: Epilepsy affects 50 million people worldwide and a third are refractory to medication. If a discrete cerebral focus or network can be identified, neurosurgical resection can be curative. Most excisions are in the temporal-lobe, and are more likely to result in seizure-freedom than extra-temporal resections. However, less than half of patients undergoing surgery become entirely seizure-free. Localizing the epileptogenic-zone and individualized outcome predictions are difficult, requiring detailed evaluations at specialist centers. Methods: We used bespoke natural language processing to text-mine 3,800 electronic health records, from 309 epilepsy surgery patients, evaluated over a decade, of whom 126 remained entirely seizure-free. We investigated the diagnostic performances of machine learning models using set-of-semiology (SoS) with and without hippocampal sclerosis (HS) on MRI as features, using STARD criteria. Findings: Support Vector Classifiers (SVC) and Gradient Boosted (GB) decision trees were the best performing algorithms for temporal-lobe epileptogenic zone localization (cross-validated Matthews correlation coefficient (MCC) SVC 0.73 ± 0.25, balanced accuracy 0.81 ± 0.14, AUC 0.95 ± 0.05). Models that only used seizure semiology were not always better than internal benchmarks. The combination of multimodal features, however, enhanced performance metrics including MCC and normalized mutual information (NMI) compared to either alone (p < 0.0001). This combination of semiology and HS on MRI increased both cross-validated MCC and NMI by over 25% (NMI, SVC SoS: 0.35 ± 0.28 vs. SVC SoS+HS: 0.61 ± 0.27). Interpretation: Machine learning models using only the set of seizure semiology (SoS) cannot unequivocally perform better than benchmarks in temporal epileptogenic-zone localization. However, the combination of SoS with an imaging feature (HS) enhance epileptogenic lobe localization. We quantified this added NMI value to be 25% in absolute terms. Despite good performance in localization, no model was able to predict seizure-freedom better than benchmarks. The methods used are widely applicable, and the performance enhancements by combining other clinical, imaging and neurophysiological features could be similarly quantified. Multicenter studies are required to confirm generalizability. Funding: Wellcome/EPSRC Center for Interventional and Surgical Sciences (WEISS) (203145Z/16/Z).

A self-supervised learning strategy for postoperative brain cavity segmentation simulating resections.

PURPOSE: Accurate segmentation of brain resection cavities (RCs) aids in postoperative analysis and determining follow-up treatment. Convolutional neural networks (CNNs) are the state-of-the-art image segmentation technique, but require large annotated datasets for training. Annotation of 3D medical images is time-consuming, requires highly trained raters and may suffer from high inter-rater variability. Self-supervised learning strategies can leverage unlabeled data for training. METHODS: We developed an algorithm to simulate resections from preoperative magnetic resonance images (MRIs). We performed self-supervised training of a 3D CNN for RC segmentation using our simulation method. We curated EPISURG, a dataset comprising 430 postoperative and 268 preoperative MRIs from 430 refractory epilepsy patients who underwent resective neurosurgery. We fine-tuned our model on three small annotated datasets from different institutions and on the annotated images in EPISURG, comprising 20, 33, 19 and 133 subjects. RESULTS: The model trained on data with simulated resections obtained median (interquartile range) Dice score coefficients (DSCs) of 81.7 (16.4), 82.4 (36.4), 74.9 (24.2) and 80.5 (18.7) for each of the four datasets. After fine-tuning, DSCs were 89.2 (13.3), 84.1 (19.8), 80.2 (20.1) and 85.2 (10.8). For comparison, inter-rater agreement between human annotators from our previous study was 84.0 (9.9). CONCLUSION: We present a self-supervised learning strategy for 3D CNNs using simulated RCs to accurately segment real RCs on postoperative MRI. Our method generalizes well to data from different institutions, pathologies and modalities. Source code, segmentation models and the EPISURG dataset are available at https://github.com/fepegar/resseg-ijcars .

A generative model of hyperelastic strain energy density functions for multiple tissue brain deformation.

PURPOSE: Estimation of brain deformation is crucial during neurosurgery. Whilst mechanical characterisation captures stress-strain relationships of tissue, biomechanical models are limited by experimental conditions. This results in variability reported in the literature. The aim of this work was to demonstrate a generative model of strain energy density functions can estimate the elastic properties of tissue using observed brain deformation. METHODS: For the generative model a Gaussian Process regression learns elastic potentials from 73 manuscripts. We evaluate the use of neo-Hookean, Mooney-Rivlin and 1-term Ogden meta-models to guarantee stability. Single and multiple tissue experiments validate the ability of our generative model to estimate tissue properties on a synthetic brain model and in eight temporal lobe resection cases where deformation is observed between pre- and post-operative images. RESULTS: Estimated parameters on a synthetic model are close to the known reference with a root-mean-square error (RMSE) of 0.1 mm and 0.2 mm between surface nodes for single and multiple tissue experiments. In clinical cases, we were able to recover brain deformation from pre- to post-operative images reducing RMSE of differences from 1.37 to 1.08 mm on the ventricle surface and from 5.89 to 4.84 mm on the resection cavity surface. CONCLUSION: Our generative model can capture uncertainties related to mechanical characterisation of tissue. When fitting samples from elastography and linear studies, all meta-models performed similarly. The Ogden meta-model performed the best on hyperelastic studies. We were able to predict elastic parameters in a reference model on a synthetic phantom. However, deformation observed in clinical cases is only partly explained using our generative model.