Tonio Ball

The heart produces small magnetic signals, but detecting them without touching the body and without heavy shielding equipment remains a major challenge. In this study, we demonstrate a noninvasive method to measure these signals using compact quantum sensors made from diamond. These sensors operate at room temperature and can detect extremely weak magnetic fields produced by the heart, even outside highly shielded environments. By averaging signals over many heartbeats and using noise suppression techniques, we successfully recorded clear magnetocardiography traces. Although current measurements require averaging, our results show a clear path toward real-time detection. We also explored a gradiometry protocol for unshielded noisy environments, paving the way for contactless cardiac detection in challenging clinical conditions. This work points toward a path for using quantum sensors in practical medical diagnostics of the heart and brain.

State-of-the-art performance in electroencephalography (EEG) decoding tasks is currently often achieved with either Deep-Learning (DL) or Riemannian-Geometry-based decoders (RBDs). Recently, there is growing interest in Deep Riemannian Networks (DRNs) possibly combining the advantages of both previous classes of methods. However, there are still a range of topics where additional insight is needed to pave the way for a more widespread application of DRNs in EEG. These include architecture design questions such as network size and end-to-end ability.How these factors affect model performance has not been explored. Additionally, it is not clear how the data within these networks is transformed, and whether this would correlate with traditional EEG decoding. Our study aims to lay the groundwork in the area of these topics through the analysis of DRNs for EEG with a wide range of hyperparameters. Networks were tested on two public EEG datasets and compared with state-of-the-art ConvNets. Here we propose end-to-end EEG SPDNet (EE(G)-SPDNet), and we show that this wide, end-to-end DRN can outperform the ConvNets, and in doing so use physiologically plausible frequency regions. We also show that the end-to-end approach learns more complex filters than traditional band-pass filters targeting the classical alpha, beta, and gamma frequency bands of the EEG, and that performance can benefit from channel specific filtering approaches. Additionally, architectural analysis revealed areas for further improvement due to the possible loss of Riemannian specific information throughout the network. Our study thus shows how to design and train DRNs to infer task-related information from the raw EEG without the need of handcrafted filterbanks and highlights the potential of end-to-end DRNs such as EE(G)-SPDNet for high-performance EEG decoding.

Brain-Computer Interfaces (BCIs) may open up new possibilities for Virtual Reality (VR) applications: BCIs may be used for active brain control of VR avatars, or to make VR content passively-adaptive based on information decoded from ongoing brain activity. Application domains for such Brain-VR Interfaces (BVRI) include medical and healthcare, entertainment, and education. Conversely, VR technology also opens up new possibilities for BCI research and development: E.g., gamified immersive BCI paradigms may improve subject engagement and long-term motivation, helping to study learning and adaptivity in the BCI-control context. Previously, we have demonstrated a first adaptive, deep-learning-based online BCI for the control of robotic assistants. Here, we describe the extension of this setup to a modular, extensible, VR-compatible online BCI setup. We describe how we integrated a classical active BCI control paradigm using motor imagery into a gamified interactive VR scenario, designed to enhance the long-term motivation of subjects. We also present an initial quality assessment of electroencephalographic (EEG) signals acquired with a dry-electrode system. We anticipate that the presented modular adaptive Brain-VR Interface will help to understand and facilitate (co-)adaptivity during long-term BCI usage.

Objective.Functional specialization is fundamental to neural information processing. Here, we study whether and how functional specialization emerges in artificial deep convolutional neural networks (CNNs) during a brain–computer interfacing (BCI) task.Approach.We trained CNNs to predict hand movement speed from intracranial electroencephalography (iEEG) and delineated how units across the different CNN hidden layers learned to represent the iEEG signal.Main results.We show that distinct, functionally interpretable neural populations emerged as a result of the training process. While some units became sensitive to either iEEG amplitude or phase, others showed bimodal behavior with significant sensitivity to both features. Pruning of highly sensitive units resulted in a steep drop of decoding accuracy not observed for pruning of less sensitive units, highlighting the functional relevance of the amplitude- and phase-specialized populations.Significance.We anticipate that emergent functional specialization as uncovered here will become a key concept in research towards interpretable deep learning for neuroscience and BCI applications.

The linguistic complexity of words has largely been studied on the behavioral level and in experimental settings. Only little is known about the neural processes underlying it in uninstructed, spontaneous conversations. We built up a multimodal neurolinguistic corpus composed of synchronized audio, video, and electrocorticographic (ECoG) recordings from the fronto-temporo-parietal cortex to address this phenomenon based on uninstructed, spontaneous speech production. We performed extensive linguistic annotations of the language material and calculated word complexity using several numeric parameters. We orthogonalized the parameters with the help of a linear regression model. Then, we correlated the spectral components of neural activity with the individual linguistic parameters and with the residuals of the linear regression model, and compared the results. The proportional relation between the number of consonants and vowels, which was the most informative parameter with regard to the neural representation of word complexity, showed effects in two areas: the frontal one was at the junction of the premotor cortex, the prefrontal cortex, and Brodmann area 44. The postcentral one lay directly above the lateral sulcus and comprised the ventral central sulcus, the parietal operculum and the adjacent inferior parietal cortex. Beyond the physiological findings summarized here, our methods may be useful for those interested in ways of studying neural effects related to natural language production and in surmounting the intrinsic problem of collinearity between multiple features of spontaneously spoken material.

How do neural network image classifiers respond to simpler and simpler inputs? And what do such responses reveal about the learning process? To answer these questions, we need a clear measure of input simplicity (or inversely, complexity), an optimization objective that correlates with simplification, and a framework to incorporate such objective into training and inference. Lastly we need a variety of testbeds to experiment and evaluate the impact of such simplification on learning. In this work, we measure simplicity with the encoding bit size given by a pretrained generative model, and minimize the bit size to simplify inputs in training and inference. We investigate the effect of such simplification in several scenarios: conventional training, dataset condensation and post-hoc explanations. In all settings, inputs are simplified along with the original classification task, and we investigate the trade-off between input simplicity and task performance. For images with injected distractors, such simplification naturally removes superfluous information. For dataset condensation, we find that inputs can be simplified with almost no accuracy degradation. When used in post-hoc explanation, our learning-based simplification approach offers a valuable new tool to explore the basis of network decisions.

Population-level modeling can define quantitative measures of individual aging by applying machine learning to large volumes of brain images. These measures of brain age, obtained from the general population, helped characterize disease severity in neurological populations, improving estimates of diagnosis or prognosis. Magnetoencephalography (MEG) and Electroencephalography (EEG) have the potential to further generalize this approach towards prevention and public health by enabling assessments of brain health at large scales in socioeconomically diverse environments. However, more research is needed to define methods that can handle the complexity and diversity of M/EEG signals across diverse real-world contexts. To catalyse this effort, here we propose reusable benchmarks of competing machine learning approaches for brain age modeling. We benchmarked popular classical machine learning pipelines and deep learning architectures previously used for pathology decoding or brain age estimation in 4 international M/EEG cohorts from diverse countries and cultural contexts, including recordings from more than 2500 participants. Our benchmarks were built on top of the M/EEG adaptations of the BIDS standard, providing tools that can be applied with minimal modification on any M/EEG dataset provided in the BIDS format. Our results suggest that, regardless of whether classical machine learning or deep learning was used, the highest performance was reached by pipelines and architectures involving spatially aware representations of the M/EEG signals, leading to R^2 scores between 0.60-0.71. Hand-crafted features paired with random forest regression provided robust benchmarks even in situations in which other approaches failed. Taken together, this set of benchmarks, accompanied by open-source software and high-level Python scripts, can serve as a starting point and quantitative reference for future efforts at developing M/EEG-based measures of brain aging. The generality of the approach renders this benchmark reusable for other related objectives such as modeling specific cognitive variables or clinical endpoints. Highlights: - We provide systematic reusable benchmarks for brain age from M/EEG signals - The benchmarks were carried out on M/EEG from four countries > 2500 recordings - We compared machine learning pipelines capable of handling the non-linear regression task of relating biomedical outcomes to M/EEG dynamics, based on classical machine learning and deep learning - Next to data-driven methods we benchmarked template-based source localization as a practical tool for generating features less affected by electromagnetic field spread - The benchmarks are built on top of the MNE ecosystem and the braindecode package and can be applied on any M/EEG dataset presented in the BIDS format

Joint applications of virtual reality (VR) systems and electroencephalography (EEG) offer numerous new possibilities ranging from behavioral science to therapy. VR systems allow for highly controlled experimental environments, while EEG offers a non-invasive window to brain activity with a millisecond-ranged temporal resolution. However, EEG measurements are highly susceptible to electromagnetic (EM) noise and the influence of EM noise of head-mounted-displays (HMDs) on EEG signal quality has not been conclusively investigated. In this paper, we propose a structured approach to test HMDs for EM noise potentially harmful to EEG measures. The approach verifies the impact of HMDs on the frequency- and time-domain of the EEG signal recorded in healthy subjects. The verification task includes a comparison of conditions with and without an HMD during (i) an eyes-open vs. eyes-closed task, and (ii) with respect to the sensory- evoked brain activity. The approach is developed and tested to derive potential effects of two commercial HMDs, the Oculus Rift and the HTC Vive Pro, on the quality of 64-channel EEG measurements. The results show that the HMDs consistently introduce artifacts, especially at the line hum of 50 Hz and the HMD refresh rate of 90 Hz, respectively, and their harmonics. The frequency range that is typically most important in non-invasive EEG research and applications (<50 Hz) however, remained largely unaffected. Hence, our findings demonstrate that high-quality EEG recordings, at least in the frequency range up to 50 Hz, can be obtained with the two tested HMDs. However, the number of commercially available HMDs is constantly rising. We strongly suggest to thoroughly test such devices upfront since each HMD will most likely have its own EM footprint and this article provides a structured approach to implement such tests with arbitrary devices.

Deep generative networks trained via maximum likelihood on a natural image dataset like CIFAR10 often assign high likelihoods to images from datasets with different objects (e.g., SVHN). We refine previous investigations of this failure at anomaly detection for invertible generative networks and provide a clear explanation of it as a combination of model bias and domain prior: Convolutional networks learn similar low-level feature distributions when trained on any natural image dataset and these low-level features dominate the likelihood. Hence, when the discriminative features between inliers and outliers are on a high-level, e.g., object shapes, anomaly detection becomes particularly challenging. To remove the negative impact of model bias and domain prior on detecting high-level differences, we propose two methods, first, using the log likelihood ratios of two identical models, one trained on the in-distribution data (e.g., CIFAR10) and the other one on a more general distribution of images (e.g., 80 Million Tiny Images). We also derive a novel outlier loss for the in-distribution network on samples from the more general distribution to further improve the performance. Secondly, using a multi-scale model like Glow, we show that low-level features are mainly captured at early scales. Therefore, using only the likelihood contribution of the final scale performs remarkably well for detecting high-level feature differences of the out-of-distribution and the in-distribution. This method is especially useful if one does not have access to a suitable general distribution. Overall, our methods achieve strong anomaly detection performance in the unsupervised setting, and only slightly underperform state-of-the-art classifier-based methods in the supervised setting. Code can be found at https://github.com/boschresearch/hierarchical_anomaly_detection.