Focused stimulation of dorsal versus ventral subthalamic nucleus enhances action-outcome learning in patients with Parkinson's disease.

Deep brain stimulation of the subthalamic nucleus is an effective treatment for the clinical motor symptoms of Parkinson's disease, but may alter the ability to learn contingencies between stimuli, actions and outcomes. We investigated how stimulation of the functional subregions in the subthalamic nucleus (motor and cognitive regions) modulates stimulus-action-outcome learning in Parkinson's disease patients. Twelve Parkinson's disease patients with deep brain stimulation of the subthalamic nucleus completed a probabilistic stimulus-action-outcome task while undergoing ventral and dorsal subthalamic nucleus stimulation (within subjects, order counterbalanced). The task orthogonalized action choice and outcome valence, which created four action-outcome learning conditions: action-reward, inhibit-reward, action-punishment avoidance and inhibit-punishment avoidance. We compared the effects of deep brain stimulation on learning rates across these conditions as well as on computed Pavlovian learning biases. Dorsal stimulation was associated with higher overall learning proficiency relative to ventral subthalamic nucleus stimulation. Compared to ventral stimulation, stimulating the dorsal subthalamic nucleus led to a particular advantage in learning to inhibit action to produce desired outcomes (gain reward or avoid punishment) as well as better learning proficiency across all conditions providing reward opportunities. The Pavlovian reward bias was reduced with dorsal relative to ventral subthalamic nucleus stimulation, which was reflected by improved inhibit-reward learning. Our results show that focused stimulation in the dorsal compared to the ventral subthalamic nucleus is relatively more favourable for learning action-outcome contingencies and reduces the Pavlovian bias that could lead to reward-driven behaviour. Considering the effects of deep brain stimulation of the subthalamic nucleus on learning and behaviour could be important when optimizing stimulation parameters to avoid side effects like impulsive reward-driven behaviour.

Reward Circuit Local Field Potential Modulations Precede Risk Taking.

Risk taking behavior is a symptom of multiple neuropsychiatric disorders and often lacks effective treatments. Reward circuitry regions including the amygdala, orbitofrontal cortex, insula, and anterior cingulate have been implicated in risk-taking by neuroimaging studies. Electrophysiological activity associated with risk taking in these regions is not well understood in humans. Further characterizing the neural signalling that underlies risk-taking may provide therapeutic insight into disorders associated with risk-taking. Eleven patients with pharmacoresistant epilepsy who underwent stereotactic electroencephalography with electrodes in the amygdala, orbitofrontal cortex, insula, and/or anterior cingulate participated. Patients participated in a gambling task where they wagered on a visible playing card being higher than a hidden card, betting $5 or $20 on this outcome, while local field potentials were recorded from implanted electrodes. We used cluster-based permutation testing to identify reward prediction error signals by comparing oscillatory power following unexpected and expected rewards. We also used cluster-based permutation testing to compare power preceding high and low bets in high-risk (<50% chance of winning) trials and two-way ANOVA with bet and risk level to identify signals associated with risky, risk averse, and optimized decisions. We used linear mixed effects models to evaluate the relationship between reward prediction error and risky decision signals across trials, and a linear regression model for associations between risky decision signal power and Barratt Impulsiveness Scale scores for each patient. Reward prediction error signals were identified in the amygdala (p=0.0066), anterior cingulate (p=0.0092), and orbitofrontal cortex (p=6.0E-4, p=4.0E-4). Risky decisions were predicted by increased oscillatory power in high-gamma frequency range during card presentation in the orbitofrontal cortex (p=0.0022), and by increased power following bet cue presentation across the theta-to-beta range in the orbitofrontal cortex ( p =0.0022), high-gamma in the anterior cingulate ( p =0.0004), and high-gamma in the insula ( p =0.0014). Risk averse decisions were predicted by decreased orbitofrontal cortex gamma power ( p =2.0E-4). Optimized decisions that maximized earnings were preceded by decreases within the theta to beta range in orbitofrontal cortex ( p =2.0E-4), broad frequencies in amygdala ( p =2.0E-4), and theta to low-gamma in insula ( p =4.0E-4). Insula risky decision power was associated with orbitofrontal cortex high-gamma reward prediction error signal ( p =0.0048) and with patient impulsivity ( p =0.00478). Our findings identify and help characterize reward circuitry activity predictive of risk-taking in humans. These findings may serve as potential biomarkers to inform the development of novel treatment strategies such as closed loop neuromodulation for disorders of risk taking.

Brain-wide human oscillatory local field potential activity during visual working memory.

Oscillatory activity in the local field potential (LFP) is thought to be a marker of cognitive processes. To understand how it differentiates tasks and brain areas in humans, we recorded LFPs in 15 adults with intracranial depth electrodes, as they performed visual-spatial and shape working memory tasks. Stimulus appearance produced widespread, broad-band activation, including in occipital, parietal, temporal, insular, and prefrontal cortex, and the amygdala and hippocampus. Occipital cortex was characterized by most elevated power in the high-gamma (100-150 Hz) range during the visual stimulus presentation. The most consistent feature of the delay period was a systematic pattern of modulation in the beta frequency (16-40 Hz), which included a decrease in power of variable timing across areas, and rebound during the delay period. These results reveal the widespread nature of oscillatory activity across a broad brain network and region-specific signatures of oscillatory processes associated with visual working memory.

Electrode position and cognitive outcome following deep brain stimulation surgery.

OBJECTIVE: Subthalamic nucleus (STN) and globus pallidus internus (GPI) deep brain stimulation (DBS) effectively treat motor symptoms in Parkinson's disease (PD) but may be associated with cognitive and psychiatric changes in some patients. Evaluation of changes in cognitive and psychiatric symptoms following DBS is complicated by changes in these symptoms that occur as part of the natural disease course. The aim of this study was to evaluate whether electrode position was associated with changes in neurocognitive symptoms in patients who underwent STN and GPI DBS. METHODS: A single-institution retrospective cohort study was conducted on patients with PD who underwent DBS from 2008 to 2019. Cognitive and psychiatric outcomes included Beck Depression Inventory II (BDI-II) score, presence of impulsive-compulsive behavior (ICB), Mini-Mental State Examination (MMSE) score, and overall cognitive status grade determined by comprehensive neuropsychology testing (normal, mild impairment, moderate impairment, and dementia). Pre- and postoperative comparisons were performed using a Wilcoxon signed-rank test or paired t-test. Patients with and without cognitive decline were compared using a Mann-Whitney U-test or unpaired t-test. A chi-square test was used for categorical comparisons. RESULTS: One hundred thirty patients were included (mean age 62.5 ± 7.9 years). At a mean postoperative follow-up from DBS of 13.0 ± 12.7 (range 6-66) months, there was an improvement in ICB (26.3% preoperatively vs 15.0% postoperatively, p = 0.017), but a decline in MMSE score (28.6 ± 1.6 vs 27.6 ± 2.0, p < 0.001) and overall cognitive status (normal: 66.2% vs 39.2%; mild: 12.3% vs 17.7%; moderate: 21.5% vs 33.1%; dementia: 0.0% vs 10.0%; p < 0.001). Patients undergoing STN DBS had a worse decline in overall cognitive status than patients who underwent GPI DBS (p = 0.006). Postoperative cognitive decline was associated with a more medial electrode position only for patients who underwent STN DBS. CONCLUSIONS: Cognitive change was observed in some patients with PD who underwent both GPI and STN DBS, likely due partly to underlying disease progression. Compared with GPI DBS, STN DBS was associated with a greater likelihood of cognitive decline. In STN but not GPI DBS, cognitive decline was associated with medialized electrode position, suggesting modulation of nonmotor STN divisions may contribute to cognitive changes following STN DBS.

Crowdsourcing Skin Demarcations of Chronic Graft-Versus-Host Disease in Patient Photographs: Training Versus Performance Study.

BACKGROUND: Chronic graft-versus-host disease (cGVHD) is a significant cause of long-term morbidity and mortality in patients after allogeneic hematopoietic cell transplantation. Skin is the most commonly affected organ, and visual assessment of cGVHD can have low reliability. Crowdsourcing data from nonexpert participants has been used for numerous medical applications, including image labeling and segmentation tasks. OBJECTIVE: This study aimed to assess the ability of crowds of nonexpert raters-individuals without any prior training for identifying or marking cGHVD-to demarcate photos of cGVHD-affected skin. We also studied the effect of training and feedback on crowd performance. METHODS: Using a Canfield Vectra H1 3D camera, 360 photographs of the skin of 36 patients with cGVHD were taken. Ground truth demarcations were provided in 3D by a trained expert and reviewed by a board-certified dermatologist. In total, 3000 2D images (projections from various angles) were created for crowd demarcation through the DiagnosUs mobile app. Raters were split into high and low feedback groups. The performances of 4 different crowds of nonexperts were analyzed, including 17 raters per image for the low and high feedback groups, 32-35 raters per image for the low feedback group, and the top 5 performers for each image from the low feedback group. RESULTS: Across 8 demarcation competitions, 130 raters were recruited to the high feedback group and 161 to the low feedback group. This resulted in a total of 54,887 individual demarcations from the high feedback group and 78,967 from the low feedback group. The nonexpert crowds achieved good overall performance for segmenting cGVHD-affected skin with minimal training, achieving a median surface area error of less than 12% of skin pixels for all crowds in both the high and low feedback groups. The low feedback crowds performed slightly poorer than the high feedback crowd, even when a larger crowd was used. Tracking the 5 most reliable raters from the low feedback group for each image recovered a performance similar to that of the high feedback crowd. Higher variability between raters for a given image was not found to correlate with lower performance of the crowd consensus demarcation and cannot therefore be used as a measure of reliability. No significant learning was observed during the task as more photos and feedback were seen. CONCLUSIONS: Crowds of nonexpert raters can demarcate cGVHD images with good overall performance. Tracking the top 5 most reliable raters provided optimal results, obtaining the best performance with the lowest number of expert demarcations required for adequate training. However, the agreement amongst individual nonexperts does not help predict whether the crowd has provided an accurate result. Future work should explore the performance of crowdsourcing in standard clinical photos and further methods to estimate the reliability of consensus demarcations.

Evaluation of synthetically generated computed tomography for use in transcranial focused ultrasound procedures.

Purpose: Transcranial focused ultrasound (tFUS) is a therapeutic ultrasound method that focuses sound through the skull to a small region noninvasively and often under magnetic resonance imaging (MRI) guidance. CT imaging is used to estimate the acoustic properties that vary between individual skulls to enable effective focusing during tFUS procedures, exposing patients to potentially harmful radiation. A method to estimate acoustic parameters in the skull without the need for CT is desirable. Approach: We synthesized CT images from routinely acquired T1-weighted MRI using a 3D patch-based conditional generative adversarial network and evaluated the performance of synthesized CT (sCT) images for treatment planning with tFUS. We compared the performance of sCT with real CT (rCT) images for tFUS planning using Kranion and simulations using the acoustic toolbox, k-Wave. Simulations were performed for 3 tFUS scenarios: (1) no aberration correction, (2) correction with phases calculated from Kranion, and (3) phase shifts calculated from time reversal. Results: From Kranion, the skull density ratio, skull thickness, and number of active elements between rCT and sCT had Pearson's correlation coefficients of 0.94, 0.92, and 0.98, respectively. Among 20 targets, differences in simulated peak pressure between rCT and sCT were largest without phase correction (12.4%±8.1%) and smallest with Kranion phases (7.3%±6.0%). The distance between peak focal locations between rCT and sCT was <1.3 mm for all simulation cases. Conclusions: Real and synthetically generated skulls had comparable image similarity, skull measurements, and acoustic simulation metrics. Our work demonstrated similar results for 10 testing cases comparing MR-sCTs and rCTs for tFUS planning. Source code and a docker image with the trained model are available at https://github.com/han-liu/SynCT_TcMRgFUS.

Brain-wide human oscillatory LFP activity during visual working memory.

Oscillatory activity is thought to be a marker of cognitive processes, although its role and distribution across the brain during working memory has been a matter of debate. To understand how oscillatory activity differentiates tasks and brain areas in humans, we recorded local field potentials (LFPs) in 12 adults as they performed visual-spatial and shape-matching memory tasks. Tasks were designed to engage working memory processes at a range of delay intervals between stimulus delivery and response initiation. LFPs were recorded using intracranial depth electrodes implanted to localize seizures for management of intractable epilepsy. Task-related LFP power analyses revealed an extensive network of cortical regions that were activated during the presentation of visual stimuli and during their maintenance in working memory, including occipital, parietal, temporal, insular, and prefrontal cortical areas, and subcortical structures including the amygdala and hippocampus. Across most brain areas, the appearance of a stimulus produced broadband power increase, while gamma power was evident during the delay interval of the working memory task. Notable differences between areas included that occipital cortex was characterized by elevated power in the high gamma (100-150 Hz) range during the 500 ms of visual stimulus presentation, which was less pronounced or absent in other areas. A decrease in power centered in beta frequency (16-40 Hz) was also observed after the stimulus presentation, whose magnitude differed across areas. These results reveal the interplay of oscillatory activity across a broad network, and region-specific signatures of oscillatory processes associated with visual working memory.

Intraoperative physiology augments atlas-based data in awake deep brain stimulation.

BACKGROUND: Deep brain stimulation (DBS) is commonly performed with patients awake to perform intraoperative microelectrode recordings and/or macrostimulation testing to guide final electrode placement. Supplemental information from atlas-based databases derived from prior patient data and visualised as efficacy heat maps transformed and overlaid onto preoperative MRIs can be used to guide preoperative target planning and intraoperative final positioning. Our quantitative analysis of intraoperative testing and corresponding changes made to final electrode positioning aims to highlight the value of intraoperative neurophysiological testing paired with image-based data to optimise final electrode positioning in a large patient cohort. METHODS: Data from 451 patients with movement disorders treated with 822 individual DBS leads at a single institution from 2011 to 2021 were included. Atlas-based data was used to guide surgical targeting. Intraoperative testing data and coordinate data were retrospectively obtained from a large patient database. Medical records were reviewed to obtain active contact usage and neurologist-defined outcomes at 1 year. RESULTS: Microelectrode recording firing profiles differ per track, per target and inform the locations where macrostimulation testing is performed. Macrostimulation performance correlates with the final electrode track chosen. Centroids of atlas-based efficacy heat maps per target were close in proximity to and may predict active contact usage at 1 year. Overall, patient outcomes at 1 year were improved for patients with better macrostimulation response. CONCLUSIONS: Atlas-based imaging data is beneficial for target planning and intraoperative guidance, and in conjunction with intraoperative neurophysiological testing during awake DBS can be used to individualize and optimise final electrode positioning, resulting in favourable outcomes.

Structural brain differences in essential tremor and Parkinson's disease deep brain stimulation patients.

BACKGROUND: Essential tremor (ET) and Parkinson's disease (PD) are the most common tremor disorders and are common indications for deep brain stimulation (DBS). In some patients, PD and ET symptoms overlap and diagnosis can be challenging based on clinical criteria alone. The objective of this study was to identify structural brain differences between PD and ET DBS patients to help differentiate these disorders and improve our understanding of the different brain regions involved in these pathologic processes. METHODS: We included ET and PD patients scheduled to undergo DBS surgery in this observational study. Patients underwent 3T brain MRI while under general anesthesia as part of their procedure. Cortical thicknesses and subcortical volumes were quantified from T1-weighted images using automated multi-atlas segmentation. We used logistic regression analysis to identify brain regions associated with diagnosis of ET or PD. RESULTS: 149 ET and 265 PD patients were included. Smaller volumes in the pallidum and thalamus and reduced thickness in the anterior orbital gyrus, lateral orbital gyrus, and medial precentral gyrus were associated with greater odds of ET diagnosis. Conversely, reduced volumes in the caudate, amygdala, putamen, and basal forebrain, and reduced thickness in the orbital part of the inferior frontal gyrus, supramarginal gyrus, and posterior cingulate were associated with greater odds of PD diagnosis. CONCLUSIONS: These findings identify structural brain differences between PD and ET patients. These results expand our understanding of the different brain regions involved in these disorders and suggest that structural MRI may help to differentiate patients with these two disorders.

Machine Learning Approach for Screening Cochlear Implant Candidates: Comparing With the 60/60 Guideline.

OBJECTIVE: To develop a machine learning-based referral guideline for patients undergoing cochlear implant candidacy evaluation (CICE) and to compare with the widely used 60/60 guideline. STUDY DESIGN: Retrospective cohort. SETTING: Tertiary referral center. PATIENTS: 772 adults undergoing CICE from 2015 to 2020. INTERVENTIONS: Variables included demographics, unaided thresholds, and word recognition score. A random forest classification model was trained on patients undergoing CICE, and bootstrap cross-validation was used to assess the modeling approach's performance. MAIN OUTCOME MEASURES: The machine learning-based referral tool was evaluated against the 60/60 guideline based on ability to identify CI candidates under traditional and expanded criteria. RESULTS: Of 587 patients with complete data, 563 (96%) met candidacy at our center, and the 60/60 guideline identified 512 (87%) patients. In the random forest model, word recognition score; thresholds at 3000, 2000, and 125; and age at CICE had the largest impact on candidacy (mean decrease in Gini coefficient, 2.83, 1.60, 1.20, 1.17, and 1.16, respectively). The 60/60 guideline had a sensitivity of 0.91, a specificity of 0.42, and an accuracy of 0.89 (95% confidence interval, 0.86-0.91). The random forest model obtained higher sensitivity (0.96), specificity (1.00), and accuracy (0.96; 95% confidence interval, 0.95-0.98). Across 1,000 bootstrapped iterations, the model yielded a median sensitivity of 0.92 (interquartile range [IQR], 0.85-0.98), specificity of 1.00 (IQR, 0.88-1.00), accuracy of 0.93 (IQR, 0.85-0.97), and area under the curve of 0.96 (IQR, 0.93-0.98). CONCLUSIONS: A novel machine learning-based screening model is highly sensitive, specific, and accurate in predicting CI candidacy. Bootstrapping confirmed that this approach is potentially generalizable with consistent results.

Electrode array positioning after cochlear reimplantation from single manufacturer.

OBJECTIVE: To investigate whether revision surgery with the same device results in a change in three key indicators of electrode positioning: scalar location, mean modiolar distance (M¯), and angular insertion depth (AID). METHODS: Retrospective analysis of a cochlear implant database at a university-based tertiary medical center. Intra-operative CT scans were obtained after initial and revision implantation. Electrode array (EA) position was calculated using auto-segmentation techniques. Initial and revision scalar location, M¯, and AID were compared. RESULTS: Mean change in M¯ for all ears was -0.07 mm (SD 0.24 mm; P = 0.16). The mean change in AID for all ears was -5° (SD 67°; P = 0.72). Three initial implantations with pre-curved EAs resulted in a translocation from Scala Tympani (ST) to Scala Vestibuli (SV). Two remained translocated after revision, while one was corrected when revised with a straight EA. An additional five translocations occurred after revision. CONCLUSIONS: In this study examining revision cochlear implantation from a single manufacturer, we demonstrated no significant change in key indicators of EA positioning, even when revising with a different style of electrode. However, the revision EA is not necessarily confined by the initial trajectory and there may be an increased risk of translocation.

Quantifying emphysema in lung screening computed tomography with robust automated lobe segmentation.

Purpose: Anatomy-based quantification of emphysema in a lung screening cohort has the potential to improve lung cancer risk stratification and risk communication. Segmenting lung lobes is an essential step in this analysis, but leading lobe segmentation algorithms have not been validated for lung screening computed tomography (CT). Approach: In this work, we develop an automated approach to lobar emphysema quantification and study its association with lung cancer incidence. We combine self-supervised training with level set regularization and finetuning with radiologist annotations on three datasets to develop a lobe segmentation algorithm that is robust for lung screening CT. Using this algorithm, we extract quantitative CT measures for a cohort (n=1189) from the National Lung Screening Trial and analyze the multivariate association with lung cancer incidence. Results: Our lobe segmentation approach achieved an external validation Dice of 0.93, significantly outperforming a leading algorithm at 0.90 (p<0.01). The percentage of low attenuation volume in the right upper lobe was associated with increased lung cancer incidence (odds ratio: 1.97; 95% CI: [1.06, 3.66]) independent of PLCOm2012 risk factors and diagnosis of whole lung emphysema. Quantitative lobar emphysema improved the goodness-of-fit to lung cancer incidence (χ2=7.48, p=0.02). Conclusions: We are the first to develop and validate an automated lobe segmentation algorithm that is robust to smoking-related pathology. We discover a quantitative risk factor, lending further evidence that regional emphysema is independently associated with increased lung cancer incidence. The algorithm is provided at https://github.com/MASILab/EmphysemaSeg.

Automated Whole Cochlear T2 Signal Demonstrates Weak Correlation with Hearing Loss in Observed Vestibular Schwannoma.

INTRODUCTION: We sought to evaluate the correlation between whole cochlear T2 signal changes obtained with a novel automated segmentation method and hearing levels, both at diagnosis and over time, in patients with observed vestibular schwannoma. METHODS: This retrospective correlation study within an academic medical center neurotology practice evaluated 127 patients with vestibular schwannoma observed over time, each with ≥2 MRI scans (367 total) and ≥2 audiograms (472 total). 86 patients had T2-weighted sequences with sufficient resolution for cochlear signal analysis, yielding 348 unique timepoint intervals. The main outcome measure was correlation of the ipsilateral-to-contralateral ratio of whole cochlear T2 signal with hearing outcomes as measured by pure tone average (PTA) and word recognition score (WRS). RESULTS: Whole cochlear T2 signal ratios did not show a correlation with hearing levels at diagnosis. Change in signal ratio over time showed weak correlation with changes in PTA, but not WRS, over time. Cochlear signal ratio did not precede changes in hearing but did follow changes in both PTA and WRS. CONCLUSION: Whole cochlear T2 signal ratios were weakly correlated with changes in hearing in patients with observed vestibular schwannoma. The technology of automated segmentation and signal processing holds promise for future evaluation of clinical entities causing cochlear signal changes.

Effects of the Number of Channels and Channel Stimulation Rate on Speech Recognition and Sound Quality Using Precurved Electrode Arrays.

PURPOSE: This study investigated the relationship between the number of active electrodes, channel stimulation rate, and their interaction on speech recognition and sound quality measures while controlling for electrode placement. Cochlear implant (CI) recipients with precurved electrode arrays placed entirely within scala tympani and closer to the modiolus were hypothesized to be able to utilize more channels and possibly higher stimulation rates to achieve better speech recognition performance and sound quality ratings than recipients in previous studies. METHOD: Participants included seven postlingually deafened adult CI recipients with Advanced Bionics Mid-Scala electrode arrays confirmed to be entirely within scala tympani using postoperative computerized tomography. Twelve conditions were tested using four, eight, 12, and 16 electrodes and channel stimulation rates of 600 pulse per second (pps), 1,200 pps, and each participant's maximum allowable rate (1,245-4,800 pps). Measures of speech recognition and sound quality were acutely assessed. RESULTS: For the effect of channels, results showed no significant improvements beyond eight channels for all measures. For the effect of channel stimulation rate, results showed no significant improvements with higher rates, suggesting that 600 pps was sufficient for maximum speech recognition performance and sound quality ratings. However, across all conditions, there was a significant relationship between mean electrode-to-modiolus distance and all measures, suggesting that a lower mean electrode-to-modiolus distance was correlated with higher speech recognition scores and sound quality ratings. CONCLUSION: These findings suggest that even well-placed precurved electrode array recipients may not be able to take advantage of more than eight channels or higher channel stimulation rates (> 600 pps), but that closer electrode array placement to the modiolus correlates with better outcomes for these recipients.

Deep whole brain segmentation of 7T structural MRI.

7T magnetic resonance imaging (MRI) has the potential to drive our understanding of human brain function through new contrast and enhanced resolution. Whole brain segmentation is a key neuroimaging technique that allows for region-by-region analysis of the brain. Segmentation is also an important preliminary step that provides spatial and volumetric information for running other neuroimaging pipelines. Spatially localized atlas network tiles (SLANT) is a popular 3D convolutional neural network (CNN) tool that breaks the whole brain segmentation task into localized sub-tasks. Each sub-task involves a specific spatial location handled by an independent 3D convolutional network to provide high resolution whole brain segmentation results. SLANT has been widely used to generate whole brain segmentations from structural scans acquired on 3T MRI. However, the use of SLANT for whole brain segmentation from structural 7T MRI scans has not been successful due to the inhomogeneous image contrast usually seen across the brain in 7T MRI. For instance, we demonstrate the mean percent difference of SLANT label volumes between a 3T scan-rescan is approximately 1.73%, whereas its 3T-7T scan-rescan counterpart has higher differences around 15.13%. Our approach to address this problem is to register the whole brain segmentation performed on 3T MRI to 7T MRI and use this information to finetune SLANT for structural 7T MRI. With the finetuned SLANT pipeline, we observe a lower mean relative difference in the label volumes of ~8.43% acquired from structural 7T MRI data. Dice similarity coefficient between SLANT segmentation on the 3T MRI scan and the after finetuning SLANT segmentation on the 7T MRI increased from 0.79 to 0.83 with p<0.01. These results suggest finetuning of SLANT is a viable solution for improving whole brain segmentation on high resolution 7T structural imaging.

The Interictal Suppression Hypothesis in focal epilepsy: network-level supporting evidence.

Why are people with focal epilepsy not continuously having seizures? Previous neuronal signalling work has implicated gamma-aminobutyric acid balance as integral to seizure generation and termination, but is a high-level distributed brain network involved in suppressing seizures? Recent intracranial electrographic evidence has suggested that seizure-onset zones have increased inward connectivity that could be associated with interictal suppression of seizure activity. Accordingly, we hypothesize that seizure-onset zones are actively suppressed by the rest of the brain network during interictal states. Full testing of this hypothesis would require collaboration across multiple domains of neuroscience. We focused on partially testing this hypothesis at the electrographic network level within 81 individuals with drug-resistant focal epilepsy undergoing presurgical evaluation. We used intracranial electrographic resting-state and neurostimulation recordings to evaluate the network connectivity of seizure onset, early propagation and non-involved zones. We then used diffusion imaging to acquire estimates of white-matter connectivity to evaluate structure-function coupling effects on connectivity findings. Finally, we generated a resting-state classification model to assist clinicians in detecting seizure-onset and propagation zones without the need for multiple ictal recordings. Our findings indicate that seizure onset and early propagation zones demonstrate markedly increased inwards connectivity and decreased outwards connectivity using both resting-state (one-way ANOVA, P-value = 3.13 × 10-13) and neurostimulation analyses to evaluate evoked responses (one-way ANOVA, P-value = 2.5 × 10-3). When controlling for the distance between regions, the difference between inwards and outwards connectivity remained stable up to 80 mm between brain connections (two-way repeated measures ANOVA, group effect P-value of 2.6 × 10-12). Structure-function coupling analyses revealed that seizure-onset zones exhibit abnormally enhanced coupling (hypercoupling) of surrounding regions compared to presumably healthy tissue (two-way repeated measures ANOVA, interaction effect P-value of 9.76 × 10-21). Using these observations, our support vector classification models achieved a maximum held-out testing set accuracy of 92.0 ± 2.2% to classify early propagation and seizure-onset zones. These results suggest that seizure-onset zones are actively segregated and suppressed by a widespread brain network. Furthermore, this electrographically observed functional suppression is disproportionate to any observed structural connectivity alterations of the seizure-onset zones. These findings have implications for the identification of seizure-onset zones using only brief electrographic recordings to reduce patient morbidity and augment the presurgical evaluation of drug-resistant epilepsy. Further testing of the interictal suppression hypothesis can provide insight into potential new resective, ablative and neuromodulation approaches to improve surgical success rates in those suffering from drug-resistant focal epilepsy.

CrossMoDA 2021 challenge: Benchmark of cross-modality domain adaptation techniques for vestibular schwannoma and cochlea segmentation.

Domain Adaptation (DA) has recently been of strong interest in the medical imaging community. While a large variety of DA techniques have been proposed for image segmentation, most of these techniques have been validated either on private datasets or on small publicly available datasets. Moreover, these datasets mostly addressed single-class problems. To tackle these limitations, the Cross-Modality Domain Adaptation (crossMoDA) challenge was organised in conjunction with the 24th International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI 2021). CrossMoDA is the first large and multi-class benchmark for unsupervised cross-modality Domain Adaptation. The goal of the challenge is to segment two key brain structures involved in the follow-up and treatment planning of vestibular schwannoma (VS): the VS and the cochleas. Currently, the diagnosis and surveillance in patients with VS are commonly performed using contrast-enhanced T1 (ceT1) MR imaging. However, there is growing interest in using non-contrast imaging sequences such as high-resolution T2 (hrT2) imaging. For this reason, we established an unsupervised cross-modality segmentation benchmark. The training dataset provides annotated ceT1 scans (N=105) and unpaired non-annotated hrT2 scans (N=105). The aim was to automatically perform unilateral VS and bilateral cochlea segmentation on hrT2 scans as provided in the testing set (N=137). This problem is particularly challenging given the large intensity distribution gap across the modalities and the small volume of the structures. A total of 55 teams from 16 countries submitted predictions to the validation leaderboard. Among them, 16 teams from 9 different countries submitted their algorithm for the evaluation phase. The level of performance reached by the top-performing teams is strikingly high (best median Dice score - VS: 88.4%; Cochleas: 85.7%) and close to full supervision (median Dice score - VS: 92.5%; Cochleas: 87.7%). All top-performing methods made use of an image-to-image translation approach to transform the source-domain images into pseudo-target-domain images. A segmentation network was then trained using these generated images and the manual annotations provided for the source image.

Non-Expert Markings of Active Chronic Graft-Versus-Host Disease Photographs: Optimal Metrics of Training Effects.

Lack of reliable measures of cutaneous chronic graft-versus-host disease (cGVHD) remains a significant challenge. Non-expert assistance in marking photographs of active disease could aid the development of automated segmentation algorithms, but validated metrics to evaluate training effects are lacking. We studied absolute and relative error of marked body surface area (BSA), redness, and the Dice index as potential metrics of non-expert improvement. Three non-experts underwent an extensive training program led by a board-certified dermatologist to mark cGVHD in photographs. At the end of the 4-month training, the dermatologist confirmed that each trainee had learned to accurately mark cGVHD. The trainees' inter- and intra-rater intraclass correlation coefficient estimates were "substantial" to "almost perfect" for both BSA and total redness. For fifteen 3D photos of patients with cGVHD, the trainees' median absolute (relative) BSA error compared to expert marking dropped from 20 cm2 (29%) pre-training to 14 cm2 (24%) post-training. Total redness error decreased from 122 a*·cm2 (26%) to 95 a*·cm2 (21%). By contrast, median Dice index did not reflect improvement (0.76 to 0.75). Both absolute and relative BSA and redness errors similarly and stably reflected improvements from this training program, which the Dice index failed to capture.

Cochlear Implant Fold Detection in Intra-operative CT Using Weakly Supervised Multi-task Deep Learning.

In cochlear implant (CI) procedures, an electrode array is surgically inserted into the cochlea. The electrodes are used to stimulate the auditory nerve and restore hearing sensation for the recipient. If the array folds inside the cochlea during the insertion procedure, it can lead to trauma, damage to the residual hearing, and poor hearing restoration. Intraoperative detection of such a case can allow a surgeon to perform reimplantation. However, this intraoperative detection requires experience and electrophysiological tests sometimes fail to detect an array folding. Due to the low incidence of array folding, we generated a dataset of CT images with folded synthetic electrode arrays with realistic metal artifact. The dataset was used to train a multitask custom 3D-UNet model for array fold detection. We tested the trained model on real post-operative CTs (7 with folded arrays and 200 without). Our model could correctly classify all the fold-over cases while misclassifying only 3 non fold-over cases. Therefore, the model is a promising option for array fold detection.