Biology subjects

Ogmen, H.

Publications and source records attributed to Ogmen, H..

3 recordsLinked to original sources

bioRxiv2024

Cortical Dynamics during Contour Integration

Integrating visual elements into contours is important for object recognition. Previous studies emphasized the role that the primary visual cortex (V1) plays in this process. However, recent evidence suggests that contour integration relies on the coordination of hierarchical substrates of cortical regions through recurrent connections. Many previous studies presented the contour at the same onset-time as the trial, which caused the subsequent neural imaging data to incorporate both visual evocation and contour integration activities, and thus confounding the two. In this study, we varied both the contour onset-time and contour fidelity and used EEG to examine the cortical activities under these conditions. Our results suggest that the temporal N300 represents the grouping and integration of visual elements into contours. Before this signature, we observed interhemispheric connections between lateral frontal and posterior parietal regions that were contingent on the contour location and peaked at around 150ms after contour appearance. Also, the magnitudes of connections between medial frontal and superior parietal regions were dependent on the timing of contour onset and peaked at around 250ms after contour onset. These activities appear to be related to the bottom-up and top-down attentional processing during contour integration, respectively, and shed light on how these processes cooperate dynamically during contour integration.

neuroscience↗

bioRxiv2023

Deep Learning and Transfer Learning for Brain Tumor Detection and Classification

Convolutional neural networks (CNNs) are powerful tools that can be trained on image classification tasks and share many structural and functional similarities with biological visual systems and mechanisms of learning. In addition to serving as a model of biological systems, CNNs possess the convenient feature of transfer learning where a network trained on one task may be repurposed for training on another, potentially unrelated, task. In this retrospective study of public domain MRI data, we investigate the ability of neural network models to be trained on brain cancer imaging data while introducing a unique camouflage animal detection transfer learning step as a means of enhancing the networks tumor detection ability. Training on glioma and normal brain MRI data, post-contrast T1-weighted and T2-weighted, we demonstrate the potential success of this training strategy for improving neural network classification accuracy. Qualitative metrics such as feature space and DeepDreamImage analysis of the internal states of trained models were also employed, which show improved generalization ability by the models following camouflage animal transfer learning. Image sensitivity functions further this investigation by allowing us to visualize the most salient image regions from a networks perspective while learning. Such methods demonstrate that the networks not only look at the tumor itself when deciding, but also at the impact on the surrounding tissue in terms of compressions and midline shifts. These results suggest an approach to brain tumor MRIs that is comparatively similar to that of trained radiologists while also exhibiting a high sensitivity to subtle structural changes resulting from the presence of a tumor. These findings present an opportunity for further research and potential use in a clinical setting.

bioinformatics↗

bioRxiv2022

Perception of Rigidity in Three- and Four-Dimensional Spaces

Our brain employs mechanisms to adapt to changing visual conditions. In addition to natural changes in our physiology and those in the environment, our brain is also capable of adapting to "unnatural" changes, such as inverted visual-inputs generated by inverting prisms. In this study, we examined the brains capability to adapt to hyperspaces. We generated four spatial-dimensional stimuli in virtual reality and tested the ability to distinguish between rigid and non-rigid motion. We found that observers are able to differentiate rigid and non-rigid motion of tesseracts (4D) with a performance comparable to that obtained using cubes (3D). Moreover, observers performance improved when they were provided with more immersive 3D experience but remained robust against increasing shape variations. At this juncture, we characterize our findings as "[Formula] perception" since, while we show the ability to extract and use 4D information, we do not have yet evidence of a complete phenomenal 4D experience.

neuroscience↗