Back

Beyond the Anterior Temporal Lobe: Domain-Related Degeneration of Cortical Language Network Dynamics in Semantic Dementia

Li, J.; Huang, L.; Hu, Z.; Wang, Y.; Guo, X.; Yang, J.; Liu, J.; Yao, R.; Yuan, B.; Guo, Q.

2025-08-31 neuroscience
10.1101/2025.08.29.673019 bioRxiv
Show abstract

Semantic dementia (SD) is a neurodegenerative disorder marked by a progressive decline in semantic memory, mainly due to focal atrophy in the anterior temporal lobe (ATL). As the disease advances, atrophy spreads to the perisylvian regions, accompanied by non-semantic language impairments. Despite these clinical findings, the network mechanisms behind cross-domain linguistic deficits remain poorly understood. In this study, we used our recently developed meta-networking framework of cortical language network dynamics to systematically examine domain-specific network disruptions in SD. Using resting-state functional MRI (fMRI) and comprehensive neuropsychological tests across several language domains, we analyzed data from SD patients at two timepoints: baseline (n = 42) and a 2-year follow-up (n = 24). Our findings showed that the framework successfully identified domain-specific language network degeneration. Beyond the ATL, progressive atrophy disrupted the dynamic separation of language networks involved in semantic processing, phonological processing, and speech production. These disruptions were characterized by state-specific hypo- and hyper-connectivity patterns that related to distinct language impairments. At follow-up, atrophy extended to posterior temporal and prefrontal regions, worsening network function. Importantly, the patterns of language network disruption predicted individual language deficits, providing a mechanistic link between structural degeneration, functional network changes, and clinical symptoms.

Matching journals

The top 7 journals account for 50% of the predicted probability mass.

50% of probability mass above

"Similar papers" are the closest papers from that journal in the model's embedding space. They show what the match is built on, but the ranking comes mostly from a classifier over the whole training set, not from these examples alone.