Search

In recent years, deep learning (DL) has become more widespread in the fields of cognitive and clinical neuroimaging. Using deep neural network models to process neuroimaging data is an efficient method to classify brain disorders and identify individuals who are at increased risk of age-related cognitive decline and neurodegenerative disease. Here we investigated, for the first time, whether structural brain imaging and DL can be used for predicting a physical trait that is of significant clinical relevance-the body mass index (BMI) of the individual. We show that individual BMI can be accurately predicted using a deep convolutional neural network (CNN) and a single structural magnetic resonance imaging (MRI) brain scan along with information about age and sex. Localization maps computed for the CNN highlighted several brain structures that strongly contributed to BMI prediction, including the caudate nucleus and the amygdala. Comparison to the results obtained via a standard automatic brain segmentation method revealed that the CNN-based visualization approach yielded complementary evidence regarding the relationship between brain structure and BMI. Taken together, our results imply that predicting BMI from structural brain scans using DL represents a promising approach to investigate the relationship between brain morphological variability and individual differences in body weight and provide a new scope for future investigations regarding the potential clinical utility of brain-predicted BMI., (Copyright © 2020 Vakli, Deák-Meszlényi, Auer and Vidnyánszky.)

Machine learning techniques have become increasingly popular in the field of resting state fMRI (functional magnetic resonance imaging) network based classification. However, the application of convolutional networks has been proposed only very recently and has remained largely unexplored. In this paper we describe a convolutional neural network architecture for functional connectome classification called connectome-convolutional neural network (CCNN). Our results on simulated datasets and a publicly available dataset for amnestic mild cognitive impairment classification demonstrate that our CCNN model can efficiently distinguish between subject groups. We also show that the connectome-convolutional network is capable to combine information from diverse functional connectivity metrics and that models using a combination of different connectivity descriptors are able to outperform classifiers using only one metric. From this flexibility follows that our proposed CCNN model can be easily adapted to a wide range of connectome based classification or regression tasks, by varying which connectivity descriptor combinations are used to train the network.

Traditional resting-state network concept is based on calculating linear dependence of spontaneous low frequency fluctuations of the BOLD signals of different brain areas, which assumes temporally stable zero-lag synchrony across regions. However, growing amount of experimental findings suggest that functional connectivity exhibits dynamic changes and a complex time-lag structure, which cannot be captured by the static zero-lag correlation analysis. Here we propose a new approach applying Dynamic Time Warping (DTW) distance to evaluate functional connectivity strength that accounts for non-stationarity and phase-lags between the observed signals. Using simulated fMRI data we found that DTW captures dynamic interactions and it is less sensitive to linearly combined global noise in the data as compared to traditional correlation analysis. We tested our method using resting-state fMRI data from repeated measurements of an individual subject and showed that DTW analysis results in more stable connectivity patterns by reducing the within-subject variability and increasing robustness for preprocessing strategies. Classification results on a public dataset revealed a superior sensitivity of the DTW analysis to group differences by showing that DTW based classifiers outperform the zero-lag correlation and maximal lag cross-correlation based classifiers significantly. Our findings suggest that analysing resting-state functional connectivity using DTW provides an efficient new way for characterizing functional networks.

Multi-echo fMRI data acquisition has been widely investigated and suggested to optimize sensitivity for detecting the BOLD signal. Several methods have also been proposed for the combination of data with different echo times. The aim of the present study was to investigate whether these advanced echo combination methods provide advantages over the simple averaging of echoes when state-of-the-art group-level random-effect analyses are performed. Both resting-state and task-based dual-echo fMRI data were collected from 27 healthy adult individuals (14 male, mean age = 25.75 years) using standard echo-planar acquisition methods at 3T. Both resting-state and task-based data were subjected to a standard image pre-processing pipeline. Subsequently the two echoes were combined as a weighted average, using four different strategies for calculating the weights: (1) simple arithmetic averaging, (2) BOLD sensitivity weighting, (3) temporal-signal-to-noise ratio weighting and (4) temporal BOLD sensitivity weighting. Our results clearly show that the simple averaging of data with the different echoes is sufficient. Advanced echo combination methods may provide advantages on a single-subject level but when considering random-effects group level statistics they provide no benefit regarding sensitivity (i.e., group-level t -values) compared to the simple echo-averaging approach. One possible reason for the lack of clear advantages may be that apart from increasing the average BOLD sensitivity at the single-subject level, the advanced weighted averaging methods also inflate the inter-subject variance. As the echo combination methods provide very similar results, the recommendation is to choose between them depending on the availability of time for collecting additional resting-state data or whether subject-level or group-level analyses are planned.

Visual cortical alpha oscillations are involved in attentional gating of incoming visual information. It has been shown that spatial and feature-based attentional selection result in increased alpha oscillations over the cortical regions representing sensory input originating from the unattended visual field and task-irrelevant visual features, respectively. However, whether attentional gating in the case of object based selection is also associated with alpha oscillations has not been investigated before. Here we measured anticipatory electroencephalography (EEG) alpha oscillations while participants were cued to attend to foveal face or word stimuli, the processing of which is known to have right and left hemispheric lateralization, respectively. The results revealed that in the case of simultaneously displayed, overlapping face and word stimuli, attending to the words led to increased power of parieto-occipital alpha oscillations over the right hemisphere as compared to when faces were attended. This object category-specific modulation of the hemispheric lateralization of anticipatory alpha oscillations was maintained during sustained attentional selection of sequentially presented face and word stimuli. These results imply that in the case of object-based attentional selection-similarly to spatial and feature-based attention-gating of visual information processing might involve visual cortical alpha oscillations.

Training on a visual task leads to increased perceptual and neural responses to visual features that were attended during training as well as decreased responses to neglected distractor features. However, the time course of these attention-based modulations of neural sensitivity for visual features has not been investigated before. Here we measured event related potentials (ERP) in response to motion stimuli with different coherence levels before and after training on a speed discrimination task requiring object-based attentional selection of one of the two competing motion stimuli. We found that two peaks on the ERP waveform were modulated by the strength of the coherent motion signal; the response amplitude associated with motion directions that were neglected during training was smaller than the response amplitude associated with motion directions that were attended during training. The first peak of motion coherence-dependent modulation of the ERP responses was at 300 ms after stimulus onset and it was most pronounced over the occipitotemporal cortex. The second peak was around 500 ms and was focused over the parietal cortex. A control experiment suggests that the earlier motion coherence-related response modulation reflects the extraction of the coherent motion signal whereas the later peak might index accumulation and readout of motion signals by parietal decision mechanisms. These findings suggest that attention-based learning affects neural responses both at the sensory and decision processing stages.