Altered eye movements during reading under degraded viewing conditions: Background luminance, text blur, and text contrast.

Degraded viewing conditions caused by either natural environments or visual disorders lead to slow reading. Here, we systematically investigated how eye movement patterns during reading are affected by degraded viewing conditions in terms of spatial resolution, contrast, and background luminance. Using a high-speed eye tracker, binocular eye movements were obtained from 14 young normally sighted adults. Images of text passages were manipulated with varying degrees of background luminance (1.3-265 cd/m2), text blur (severe blur to no blur), or text contrast (2.6%-100%). We analyzed changes in key eye movement features, such as saccades, microsaccades, regressive saccades, fixations, and return-sweeps across different viewing conditions. No significant changes were observed for the range of tested background luminance values. However, with increasing text blur and decreasing text contrast, we observed a significant decrease in saccade amplitude and velocity, as well as a significant increase in fixation duration, number of fixations, proportion of regressive saccades, microsaccade rate, and duration of return-sweeps. Among all, saccade amplitude, fixation duration, and proportion of regressive saccades turned out to be the most significant contributors to reading speed, together accounting for 90% of variance in reading speed. Our results together showed that, when presented with degraded viewing conditions, the patterns of eye movements during reading were altered accordingly. These findings may suggest that the seemingly deviated eye movements observed in individuals with visual impairments may be in part resulting from active and optimal information acquisition strategies operated when visual sensory input becomes substantially deprived.

A Deep Learning Framework Based on Dynamic Channel Selection for Early Classification of Left and Right Hand Motor Imagery Tasks.

Ideal brain-computer interfaces (BCIs) need to be efficient and accurate, demanding for classifiers that can work across subjects while providing high classification accu-racy results from recordings with short duration. To address this problem, we present a new deep learning framework for discriminating motor imagery (MI) tasks from electroen-cephalography (EEG) signals. The framework consists of a 1D convolutional neural network-long short-term memory (CNN-LSTM), combined with a dynamic channel selection approach based on Davies-Bouldin index (DBI). Using data from BCI competition IV-IIa data, the proposed framework reports an average classification accuracy of 70.17% and 76.18% when using only 800 ms and 1500 ms of the EEG data after the task onset, respectively. The proposed framework dynamically balances individual differences, achieves comparable or better performance compared to existing work, while using short duration of EEG.

Foveal crowding appears to be robust to normal aging and glaucoma unlike parafoveal and peripheral crowding.

Visual crowding is the inability to recognize a target object in clutter. Previous studies have shown an increase in crowding in both parafoveal and peripheral vision in normal aging and glaucoma. Here, we ask whether there is any increase in foveal crowding in both normal aging and glaucomatous vision. Twenty-four patients with glaucoma and 24 age-matched normally sighted controls (mean age = 65 ± 7 vs. 60 ± 8 years old) participated in this study. For each subject, we measured the extent of foveal crowding using Pelli's foveal crowding paradigm (2016). We found that the average crowding zone was 0.061 degrees for glaucoma and 0.056 degrees for age-matched normal vision, respectively. These values fall into the range of foveal crowding zones (0.0125 degrees to 0.1 degrees) observed in young normal vision. We, however, did not find any evidence supporting increased foveal crowding in glaucoma (p = 0.375), at least in the early to moderate stages of glaucoma. In the light of previous studies on foveal crowding in normal young vision, we did not find any evidence supporting age-related changes in foveal crowding. Even if there is any, the effect appears to be rather inconsequential. Taken together, our findings suggest unlike parafoveal or peripheral crowding (2 degrees, 4 degrees, 8 degrees, and 10 degrees eccentricities), foveal crowding (<0.25 degrees eccentricity) appears to be less vulnerable to normal aging or moderate glaucomatous damage.

Identifying the Retinal Layers Linked to Human Contrast Sensitivity Via Deep Learning.

Purpose: Luminance contrast is the fundamental building block of human spatial vision. Therefore contrast sensitivity, the reciprocal of contrast threshold required for target detection, has been a barometer of human visual function. Although retinal ganglion cells (RGCs) are known to be involved in contrast coding, it still remains unknown whether the retinal layers containing RGCs are linked to a person's contrast sensitivity (e.g., Pelli-Robson contrast sensitivity) and, if so, to what extent the retinal layers are related to behavioral contrast sensitivity. Thus the current study aims to identify the retinal layers and features critical for predicting a person's contrast sensitivity via deep learning. Methods: Data were collected from 225 subjects including individuals with either glaucoma, age-related macular degeneration, or normal vision. A deep convolutional neural network trained to predict a person's Pelli-Robson contrast sensitivity from structural retinal images measured with optical coherence tomography was used. Then, activation maps that represent the critical features learned by the network for the output prediction were computed. Results: The thickness of both ganglion cell and inner plexiform layers, reflecting RGC counts, were found to be significantly correlated with contrast sensitivity (r = 0.26 ∼ 0.58, Ps < 0.001 for different eccentricities). Importantly, the results showed that retinal layers containing RGCs were the critical features the network uses to predict a person's contrast sensitivity (an average R2 = 0.36 ± 0.10). Conclusions: The findings confirmed the structure and function relationship for contrast sensitivity while highlighting the role of RGC density for human contrast sensitivity.

Binocularly Asymmetric Crowding in Glaucoma and a Lack of Binocular Summation in Crowding.

Purpose: Glaucoma is associated with progressive loss of retinal ganglion cells. Here we investigated the impact of glaucomatous damage on monocular and binocular crowding in parafoveal vision. We also examined the binocular summation of crowding to see if crowding is alleviated under binocular viewing. Methods: The study design included 40 individuals with glaucoma and 24 age-similar normal cohorts. For each subject, the magnitude of crowding was determined by the extent of crowding zone. Crowding zone measurements were made binocularly in parafoveal vision (i.e., at 2° and 4° retinal eccentricities) visual field. For a subgroup of glaucoma subjects (n = 17), crowding zone was also measured monocularly for each eye. Results: Our results showed that, compared with normal cohorts, individuals with glaucoma exhibited significantly larger crowding-enlargement of crowding zone (an increase by 21%; P < 0.01). Moreover, we also observed a lack of binocular summation (i.e., a binocular ratio of 1): binocular crowding was determined by the better eye. Hence, our results did not provide evidence supporting binocular summation of crowding in glaucomatous vision. Conclusions: Our findings show that crowding is exacerbated in parafoveal vision in glaucoma and binocularly asymmetric glaucoma seems to induce binocularly asymmetric crowding. Furthermore, the lack of binocular summation for crowding observed in glaucomatous vision combined with the lack of binocular summation reported in a previous study on normal healthy vision support the view that crowding may start in the early stages of visual processing, at least before the process of binocular integration takes place.

Functional Field of View Determined by Crowding, Aging, or Glaucoma Under Divided Attention.

Purpose: Parafoveal or peripheral vision is important for various everyday activities. This is particularly relevant to those who suffer from visual field defects. Here we quantified the effect of visual crowding, normal aging, and glaucoma on the spatial extent of the functional field of view (FFV) under divided attention. Methods: Unlike visual acuity measured by single-letter recognition or visual perimetry measured by light spot detection, we measured the FFV using a target letter presented either alone or in letter triplets appearing across the visual field. A subject's task was to report whether the target letter was the same as the letter displayed concurrently at the central fixation region (i.e., divided attention task). Over the trials, a plot of the proportion correct for letter recognition versus target location was constructed, resulting in a visual field map. Results: The results obtained from three subject groups-normal young adults, normal older adults, and patients with glaucoma-showed that on average the central 20° visual field was relatively robust to uncrowded target recognition under divided attention. However, the FFV shrunk down to the central 10° visual field when the target appeared in clutter, suggesting a strong crowding effect on FFV. An additional shrinkage of the FFV occurred in the presence of aging and glaucoma. Conclusions: Using a quantitative method, we demonstrate that crowding, aging, and glaucoma independently decrease the spatial extent of FFV under divided attention and that crowding seems to be the major contributor limiting FFV. Translational Relevance: Our FFV test may complement standard clinical measurements by providing functionally relevant visual field information.

Early classification of motor tasks using dynamic functional connectivity graphs from EEG.

Objective. Classification of electroencephalography (EEG) signals with high accuracy using short recording intervals has been a challenging problem in developing brain computer interfaces (BCIs). This paper presents a novel feature extraction method for EEG recordings to tackle this problem.Approach. The proposed approach is based on the concept that the brain functions in a dynamic manner, and utilizes dynamic functional connectivity graphs. The EEG data is first segmented into intervals during which functional networks sustain their connectivity. Functional connectivity networks for each identified segment are then localized, and graphs are constructed, which will be used as features. To take advantage of the dynamic nature of the generated graphs, a long short term memory classifier is employed for classification.Main results. Features extracted from various durations of post-stimulus EEG data associated with motor execution and imagery tasks are used to test the performance of the classifier. Results show an average accuracy of 85.32% about only 500 ms after stimulus presentation.Significance. Our results demonstrate, for the first time, that using the proposed feature extraction method, it is possible to classify motor tasks from EEG recordings using a short interval of the data in the order of hundreds of milliseconds (e.g. 500 ms). This duration is considerably shorter than what has been reported before. These results will have significant implications for improving the effectiveness and the speed of BCIs, particularly for those used in assistive technologies.

Recognizing Pain in Motor Imagery EEG Recordings Using Dynamic Functional Connectivity Graphs.

The goal of this paper is to investigate whether motor imagery tasks, performed under pain-free versus pain conditions, can be discriminated from electroencephalography (EEG) recordings. Four motor imagery classes of right hand, left hand, foot, and tongue are considered. A functional connectivity-based feature extraction approach along with a long short-term memory (LSTM) classifier are employed for classifying pain-free versus under-pain classes. Moreover, classification is performed in different frequency bands to study the significance of each band in differentiating motor imagery data associated with pain-free and under-pain states. When considering all frequency bands, the average classification accuracy is in the range of 77:86-80:04%. Our frequency-specific analysis shows that the gamma band results in a notably higher accuracy than other bands, indicating the importance of this band in discriminating pain/no-pain conditions during the execution of motor imagery tasks. In contrast, functional connectivity graphs extracted from delta and theta bands do not seem to provide discriminatory information between pain-free and under-pain conditions. This is the first study demonstrating that motor imagery tasks executed under pain and without pain conditions can be discriminated from EEG recordings. Our findings can provide new insights for developing effective brain computer interface-based assistive technologies for patients who are in real need of them.

Detecting mTBI by Learning Spatio-temporal Characteristics of Widefield Calcium Imaging Data Using Deep Learning.

Early diagnosis of mild traumatic brain injury (mTBI) is of great interest to the neuroscience and medical communities. Widefield optical imaging of neuronal populations over the cerebral cortex in animals provides a unique opportunity to study injury-induced alternations in brain function. Using this technique, along with deep learning, the goal of this paper is to develop a framework for the detection of mTBI. Cortical activities in transgenic calcium reporter mice expressing GCaMP6s are obtained using widefield imaging from 8 mice before and after inducing an injury. Two deep learning models for differentiating mTBI from normal conditions are proposed. One model is based on the convolutional neural network-long short term memory (CNN-LSTM), and the second model is based on a 3D-convolutional neural network (3D-CNN). These two models offer the ability to capture features both in the spatial and temporal domains. Results for the average classification accuracy for the CNN-LSTM and the 3D-CNN are 97.24% and 91.34%, respectively. These results are notably higher than the case of using the classical machine learning methods, demonstrating the importance of utilizing both the spatial and temporal information for early detection of mTBI.

On the Spatiotemporal Characteristics of Class-Discriminating Functional Networks.

We target the problem of identifying brain's functional networks that are discriminatory across classes of tasks, using data obtained through electroencephalography (EEG). A three-step framework is presented. First, the EEG data is segmented to identify the intervals during which cortical functional networks remain quasi-stationary. Second, these functional networks are spatially localized in the cortex. Finally, by employing the proposed discriminative Boolean matrix factorization (DBMF) algorithm, functional networks that are most recurrent in one class of tasks, but are least recurrent in the other are identified. The DBMF algorithm is capable of providing the spatial maps of the discriminative functional networks as well as information about their dynamic occurrence over time. The framework is applied to experimental EEG data, recorded during a motor task. The results show that the proposed framework identifies several parietal/motor functional networks as being the most discriminatory for motor execution trials from non-execution trials.