Browsing by Author "Dar, Salman UI Hassan"

Now showing 1 - 2 of 2

Open Access
Inter-regional connectivity in the human brain during visual search
(2016-08) Dar, Salman UI Hassan
Separate groups of regions in the human brain are thought to be functionally specialized for representing specific categories of visual objects, and for controlling and deployments of visual attention. It is commonly assumed that the information ow between these regions is altered during visual search. However, little is known about the magnitude and extent of these changes during natural visual search. Here, we assess the changes in functional connectivity between the attention-control network and the category-selective regions during category-based visual search in natural movies. Brain activity was recorded using functional magnetic resonance imaging (fMRI) while subjects viewed natural movies. To investigate the changes in connectivity strength between pairs of brain regions, we employed coherence analysis. Coherence is a non-directional measure of association, which identifies correlation in frequency domain. To infer the in uence of attention-control areas on category-selective areas, Granger causality analysis was carried out. Granger causality uses the idea of temporal precedence that cause precedes the effect. Furthermore, to examine whether attention changes inter-regional connectivity after accounting for stimulus-driven brain activity, two separate encoding models were used to capture brain responses elicited by low-level structural and high-level category features in natural movies via L2-regularized linear regression. Response predictions of the structural and category models were removed from the recorded blood-oxygen-level dependent (BOLD) responses to obtain the residual responses. The connectivity analyses were repeated on the residuals to determine if the attentional changes in connectivity persist even after projecting out the stimulus-driven brain activity. The results indicate that performing visual search for a specific object category enhances the in uence of high-level attention-control network on category-selective areas in ventral temporal cortex. Furthermore, these connectivity patterns persist even after projecting out the stimulus-driven brain activity from the recorded BOLD responses.
Open Access
One model to unite them all: Personalized federated learning of multi-contrast MRI synthesis
(ELSEVIER, 2024-05) Dalmaz, Onat; Mirza, Muhammad Usama; Elmas, Gökberk; Özbey, Muzaffer; Dar, Salman UI Hassan; Ceyani, Emir; Karlı Oğuz, Kader; Avestimehr, Salman; Çukur, Tolga
Curation of large, diverse MRI datasets via multi-institutional collaborations can help improve learningof generalizable synthesis models that reliably translate source- onto target-contrast images. To facilitatecollaborations, federated learning (FL) adopts decentralized model training while mitigating privacy concernsby avoiding sharing of imaging data. However, conventional FL methods can be impaired by the inherentheterogeneity in the data distribution, with domain shifts evident within and across imaging sites. Here weintroduce the first personalized FL method for MRI Synthesis (pFLSynth) that improves reliability against dataheterogeneity via model specialization to individual sites and synthesis tasks (i.e., source-target contrasts).To do this, pFLSynth leverages an adversarial model equipped with novel personalization blocks that controlthe statistics of generated feature maps across the spatial/channel dimensions, given latent variables specificto sites and tasks. To further promote communication efficiency and site specialization, partial networkaggregation is employed over later generator stages while earlier generator stages and the discriminatorare trained locally. As such, pFLSynth enables multi-task training of multi-site synthesis models with highgeneralization performance across sites and tasks. Comprehensive experiments demonstrate the superiorperformance and reliability of pFLSynth in MRI synthesis against prior federated methods