Béla Weiss

Dr. Béla Weiss


Research Focus & Metrics

Academic Career Length

6Years (2021-Present)

(Calculated once Scilit publication data is synced: current year − year of first publication + 1)

Representative Publications

View All Publications
23 May 2023 in Medical Image Analysis
Verified
Journal Article
Open access
PVPál Vakli
Pál Vakli
BWBéla Weiss
Béla Weiss
JSJános Szalma
János Szalma
PBPéter Barsi
Péter Barsi
IGIstván Gyuricza
István Gyuricza
ZVZoltán Vidnyánszky
Zoltán Vidnyánszky
Head motion artifacts in magnetic resonance imaging (MRI) are an important confounding factor concerning brain research as well as clinical practice. For this reason, several machine learning-based methods have been developed for the automatic quality control of structural MRI scans. Deep learning offers a promising solution to this problem, however, given its data-hungry nature and the scarcity of expert-annotated datasets, its advantage over traditional machine learning methods in identifying motion-corrupted brain scans is yet to be determined. In the present study, we investigated the relative advantage of the two methods in structural MRI quality control. To this end, we collected publicly available T1-weighted images and scanned subjects in our own lab under conventional and active head motion conditions. The quality of the images was rated by a team of radiologists from the point of view of clinical diagnostic use. We present a relatively simple, lightweight 3D convolutional neural network trained in an end-to-end manner that achieved a test set (N = 411) balanced accuracy of 94.41% in classifying brain scans into clinically usable or unusable categories. A support vector machine trained on image quality metrics achieved a balanced accuracy of 88.44% on the same test set. Statistical comparison of the two models yielded no significant difference in terms of confusion matrices, error rates, or receiver operating characteristic curves. Our results suggest that these machine learning methods are similarly effective in identifying severe motion artifacts in brain MRI scans, and underline the efficacy of end-to-end deep learning-based systems in brain MRI quality control, allowing the rapid evaluation of diagnostic utility without the need for elaborate image pre-processing.

15 June 2021 in Brain Topography
Verified
Journal Article
Open access
PDF file
JKJános Körmendi
János Körmendi
EFEszter Ferentzi
Eszter Ferentzi
BWBéla Weiss
Béla Weiss
ZNZoltán György Dr. Nagy
Zoltán György Dr. Nagy
The aim of this study was to analyse the high density EEG during movement execution guided by visual attention to reveal the detailed topographic distributions of delta and theta oscillations. Twenty right-handed young subjects performed a finger tapping task, paced by a continuously transited repeating visual stimuli. Baseline corrected power of scalp current density transformed EEG was statistically assessed with cluster-based permutation testing. Delta and theta activities revealed differences in their spatial properties at the time of finger tapping execution. Theta synchronization showed a contralateral double activation in the parietal and fronto-central regions, while delta activity appeared in the central contralateral channels. Differences in the spatiotemporal topography between delta and theta activity in the course of movement execution were identified on high density EEG.

14 June 2022 in NeuroImage
Verified
Journal Article
Open access
BWBéla Weiss
Béla Weiss
ÁNÁdám Nárai
Ádám Nárai
ZVZoltán Vidnyánszky
Zoltán Vidnyánszky
Skilled reading requires specialized visual cortical processing of orthographic information and its impairment has been proposed as a potential correlate of compromised reading in dyslexia. However, which stage of orthographic information processing during natural reading is disturbed in dyslexics remains unexplored. Here we addressed this question by simultaneously measuring the eye movements and EEG of dyslexic and control young adults during natural reading. Isolated meaningful sentences were presented at five inter-letter spacing levels spanning the range from minimal to extra-large spacing, and participants were instructed to read the text silently at their own pace. Control participants read faster, performed larger saccades and shorter fixations compared to dyslexics. While reading speed peaked around the default letter spacing, saccade amplitude increased and fixation duration decreased with the increase of letter spacing in both groups. Lateralization of occipito-temporal fixation-related EEG activity (FREA) was found in three consecutive time intervals corresponding to early orthographic processing in control readers. Importantly, the lateralization in the time range of the first negative left occipito-temporal FREA peak was specific for first fixations and exhibited an interaction effect between reading ability and letter spacing. The interaction originated in the significant decrease of FREA lateralization at extra-large compared to default letter spacing in control readers and the lack of lateralization in both letter spacing conditions in the case of dyslexics. These findings suggest that expertise-driven hemispheric functional specialization for early orthographic processing thought to be responsible for letter identity extraction during natural reading is compromised in dyslexia.

1 January 2025 in Radiology: Artificial Intelligence
Verified
Journal Article
Closed access
DADamiano Archetti
Damiano Archetti
VVVikram Venkatraghavan
Vikram Venkatraghavan
BWBéla Weiss
Béla Weiss
PBPierrick Bourgeat
Pierrick Bourgeat
TATibor Auer
Tibor Auer
FCfor the E-DADS (Early Detection of Alzheimer's Disease Subtypes) Consortium
for the E-DADS (Early Detection of Alzheimer's Disease Subtypes) Consortium
A multiclass Neuroharmony model was developed and evaluated against other approaches for harmonizing volumetric data in a clinical setting using a large, multicenter brain MRI dataset of individuals undergoing neuroradiologic assessment of Alzheimer disease.

20 June 2022 in PsyArXiv
Verified
Preprint
Open access
BWBéla Weiss
Béla Weiss
JSJános Szalma
János Szalma
ZVZoltán Vidnyánszky
Zoltán Vidnyánszky
Detection of developmental dyslexia is paramount to mitigate the adversary effects of reading difficulties. Accordingly, efforts have been made on developing methods for data-driven detection of dyslexia based on behavioral and eye-tracking data recorded predominantly in children. However, it remains to be clarified how the classification performance of behavioral and eye-movement features compare to each other and how the findings obtained for children translate to adulthood.To address this shortcoming, here we collected behavioral and eye-movement data from dyslexic and control young adults. Using a machine-learning framework with different feature selection methods, cross-validation approaches and hyperparameter optimization, we developed classifiers for detection of dyslexic participants based on behavioral and eye-movement features.The final behavioral model achieved a classification performance of 92.91% significantly outperforming the classifier based on eye-movement features that reached 76.25% classification accuracy. Feature importance results were in agreement with earlier studies showing deterioration of short-term memory, non-word reading fluency, phonological awareness and saccade amplitude in dyslexics, and also supported the importance of glissades in characterization of eye movements.Our findings indicate the superiority of behavioral measures over eye-movement features in detecting reading difficulties and suggest that earlier results obtained for children may generalize to adulthood.

29 September 2022 in Cortex
Verified
Journal Article
Open access
ÁNÁdám Nárai
Ádám Nárai
ZNZsuzsanna Nemecz
Zsuzsanna Nemecz
ZVZoltán Vidnyánszky
Zoltán Vidnyánszky
BWBéla Weiss
Béla Weiss
Lateralized processing of orthographic information is a hallmark of proficient reading. However, how this finding obtained for fixed-gaze processing of orthographic stimuli translates to ecologically valid reading conditions remained to be clarified. To address this shortcoming, here we assessed the lateralization of early orthographic processing in fixed-gaze and natural reading conditions using concurrent eye-tracking and EEG data recorded from young adults without reading difficulties. Sensor-space analyses confirmed the well-known left-lateralized negative-going deflection of fixed-gaze EEG activity throughout the period of early orthographic processing. At the same time, fixation-related EEG activity exhibited left-lateralized followed by right-lateralized processing of text stimuli during natural reading. A strong positive relationship was found between the early leftward lateralization in fixed-gaze and natural reading conditions. Using source-space analyses, early left-lateralized brain activity was obtained in lateraloccipital and posterior ventral occipito-temporal cortices reflecting letter-level processing in both conditions. In addition, in the same time interval, left-lateralized source activity was found also in premotor and parietal brain regions during natural reading. While brain activity remained left-lateralized in later stages representing word-level processing in posterior and middle ventral temporal regions in the fixed-gaze condition, fixation-related source activity became stronger in the right hemisphere in medial and more anterior ventral temporal brain regions indicating higher-level processing of orthographic information. Although our results show a strong positive relationship between the lateralization of letter-level processing in the two reading modes and suggest lateralized brain activity as a general marker for processing of orthographic information, they also clearly indicate the need for reading research in ecologically valid conditions to identify the neural basis of visuospatial attentional, oculomotor and higher-level processes specific to natural reading.