A5029468195- Blind Source Separation Techniques
- Advanced Neuroimaging Techniques and Applications
- Speech and Audio Processing
- Hearing Loss and Rehabilitation
- Functional Brain Connectivity Studies
- Tensor decomposition and applications
- EEG and Brain-Computer Interfaces
- Advanced MRI Techniques and Applications
- Neonatal and fetal brain pathology
- Systemic Sclerosis and Related Diseases
- Advanced Adaptive Filtering Techniques
- Algorithms and Data Compression
- Cardiovascular Function and Risk Factors
- Multiple Sclerosis Research Studies
- Neural dynamics and brain function
- Dermatologic Treatments and Research
- COVID-19 diagnosis using AI
- Atomic and Subatomic Physics Research
- Cerebrospinal fluid and hydrocephalus
- Cerebral Palsy and Movement Disorders
- Dementia and Cognitive Impairment Research
- Interstitial Lung Diseases and Idiopathic Pulmonary Fibrosis
- Image Processing Techniques and Applications
- Ultrasound Imaging and Elastography
- Neural Networks and Applications
Icometrix (Belgium)
2021-2025
KU Leuven
2016-2022
IMEC
2017-2019
Dynamic Systems (United States)
2016
iMinds
2016
Objective: We aim to extract and denoise the attended speaker in a noisy two-speaker acoustic scenario, relying on microphone array recordings from binaural hearing aid, which are complemented with electroencephalography (EEG) infer of interest. Methods: In this study, we propose modular processing flow that first extracts two speech envelopes recordings, then selects envelope based EEG, finally uses inform multichannel separation denoising algorithm. Results: Strong suppression interfering...
Importance Amyloid-related imaging abnormalities (ARIA) are brain magnetic resonance (MRI) findings associated with the use of amyloid-β–directed monoclonal antibody therapies in Alzheimer disease (AD). ARIA monitoring is important to inform treatment dosing decisions and might be improved through assistive software. Objective To assess clinical performance an artificial intelligence (AI)–based software tool for assisting radiological interpretation MRI scans patients monitored ARIA. Design,...
Multiple sclerosis (MS) is a chronic autoimmune, inflammatory neurological disease of the central nervous system. Its diagnosis nowadays commonly includes performing an MRI scan, as it most sensitive imaging test for MS. MS plaques are identified from fluid-attenuated inversion recovery (FLAIR) images hyperintense regions that highly varying in terms their shapes, sizes and locations, routinely classified accordance to McDonald criteria. Recent years have seen increase works aimed at...
Abstract Automated quantification of brain tissues on MR images has greatly contributed to the diagnosis and follow-up neurological pathologies across various life stages. However, existing solutions are specifically designed for certain age ranges, limiting their applicability in monitoring development from infancy late adulthood. This retrospective study aims develop validate a segmentation model pediatric adult populations. First, we trained deep learning segment structures using...
Systemic sclerosis (SSc) is a rare connective tissue disease associated with rapidly evolving interstitial lung (ILD), driving its mortality. Specific imaging-based biomarkers the evolution of are needed to help predict and quantify ILD. We evaluated potential an automated ILD quantification system (icolung®) from chest CT scans, in prediction progression SSc-ILD. used retrospective cohort 75 SSc-ILD patients evaluate AI-based tool correlate image-based pulmonary function tests their over...
Assessing brain damage in children with spastic unilateral cerebral palsy (uCP) is challenging, particularly clinical settings. In this study, we developed and validated a deep learning-based pipeline to automatically quantify lesion-free volumes. Using T1-weighted FLAIR MRI data from 35 patients (aged 5–15 years), trained models segment structures lesions, utilizing an automatic label generation workflow. Validation was performed on 54 CP 7–16 years) using quantitative qualitative metrics,...
Hearing prostheses have built-in algorithms to perform acoustic noise reduction and improve speech intelligibility. However, in a multi-speaker scenario the algorithm has determine which speaker listener is focusing on, order enhance it while suppressing other interfering sources. Recently, been demonstrated that possible detect auditory attention using electroencephalography (EEG). In this paper, we use multi-channel Wiener filters (MWFs), filter out each stream from mixtures micro-phones...
Simultaneous recording of electroencephalographic (EEG) signals and functional magnetic resonance images (fMRI) has gained wide interest in brain research, thanks to the highly complementary spatiotemporal nature both modalities. We propose a novel technique extract sources neural activity from multimodal measurements, which relies on structured form coupled matrix-tensor factorization (CMTF). In data-symmetric fashion, we characterize these underlying spatial, temporal spectral domain,...
EEG-correlated fMRI analysis is widely used to detect regional BOLD fluctuations that are synchronized interictal epileptic discharges, which can provide evidence for localizing the ictal onset zone. However, typical, asymmetrical and mass-univariate approach cannot capture inherent, higher order structure in EEG data, nor multivariate relations it nontrivial accurately handle varying neurovascular coupling over patients brain regions. We aim overcome these drawbacks a data-driven manner by...
State-of-the-art hearing prostheses are equipped with acoustic noise reduction algorithms to improve speech intelligibility. Currently, one of the major challenges is perform in so-called cocktail party scenarios multiple speakers, particular because it difficult-if not impossible-for algorithm determine which target speaker(s) that should be enhanced, and treated as interfering sources. Recently, has been shown electroencephalography (EEG) can used auditory attention detection, i.e., detect...
We propose and evaluate a method to estimate respiratory signal from ungated cardiac magnetic resonance (CMR) images. Ungated CMR images were acquired in five subjects who performed exercise at different intensity levels under physiological conditions while breathing freely. The motion was estimated by applying principal components analysis (PCA). A sign correction procedure developed correctly define inspiration expiration, based on either tracking of the diaphragmatic or estimation lung...
The classification of brain states using neural recordings such as electroencephalography (EEG) finds applications in both medical and non-medical contexts, detecting epileptic seizures or discriminating mental brain-computer interfaces, respectively. Although this endeavor is well-established, existing solutions are typically restricted to lab hospital conditions because they operate on from a set EEG electrodes that covers the whole head. By contrast, true breakthrough for these would be...
Many interesting matrix decompositions/factorizations, and especially many tensor decompositions, have to be solved by non-convex optimization-based algorithms, that may converge local optima. Hence, when interpretability of the components is a requirement, practitioners compute decomposition (e.g. CPD) times, with different initializations, verify whether are reproducible over repetitions optimization. However, it non-trivial assess such reliability or stability multiple optima encountered....
Early diagnosis of COVID-19 is required to provide the best treatment our patients, prevent epidemic from spreading in community, and reduce costs associated with aggravation disease. We developed a decision tree model evaluate impact using an artificial intelligence-based chest computed tomography (CT) analysis software (icolung, icometrix) analyze CT scans for detection prognosis cases. The compared routine practice where patients receiving scan were not screened COVID-19, scenario icolung...
Objective: To improve the accuracy of detecting ictal onset zone, we propose to enhance epilepsy-related activity present in EEG signals, before mapping their BOLD correlates through EEG-correlated fMRI analysis. Methods: Based solely on a segmentation interictal epileptic discharges (IEDs) EEG, train multi-channel Wiener filters (MWF) which IED-like waveforms, and suppress background noisy influences. Subsequently, use find brain regions signal fluctuation corresponds filtered signals'...
Evaluate the performance of an automated detection and quantification tool for identification severity assessment amyloid-related imaging abnormalities (ARIA).
Functional ultrasound (fUS) is an exciting new neuroimaging technique that able to record brain activity similar functional magnetic resonance imaging, yet with higher spatiotemporal resolution and at lower cost. We consider the problem of jointly estimating underlying neural sources hemodynamic response function (HRF) from fUS recordings. propose model measured voxel time-series as a convolutive mixture multiple source signals solve blind deconvolution via block-term decomposition. This...
State-of-the-art hearing prostheses are equipped with acoustic noise reduction algorithms to improve speech intelligibility. However, cocktail party scenarios multiple speakers pose a major challenge since it is difficult for the algorithm determine which speaker should enhance. To address this problem, electroencephalography (EEG) signals can be used perform auditory attention detection (AAD), i.e., detect listener attending to. Taking step further towards realization of neuro-steered...
In a multi-speaker scenario, major challenge for noise suppression systems in hearing instruments is to determine which sound source the listener attending to. It has been shown that linear decoder can extract neural signal from EEG recordings better correlated with envelope of attended speech than envelopes other signals. This be exploited perform auditory attention detection (AAD), then steer algorithm. The passed through model periphery before extracting its envelope. We compared 7...
<b>Background:</b> Systemic sclerosis (SSc) is a rare connective tissue disease associated with rapid evolving interstitial lung (SSc-ILD). Imaging-based biomarkers the evolution of are highly needed to help in prediction outcome. <b>Methods:</b> We evaluated potential AI-based quantification tool on chest CT imaging <i>icolung</i> (icometrix, Belgium). To do so, we retrospectively correlated outcomes pulmonary function tests and over time. <b>Results:</b> group 75 patients suffering from...