Beyond one-to-one mappings: Modelling distributed lesion-symptom relationships with multilayer networks
Abstract
Lesion-symptom mapping is widely used to identify causal relationships between brain structures and behaviour, and has played a central role in neuropsychologically informed network models of cognition. However, even recent approaches remain constrained by a one-to-one mapping framework, which oversimplifies the complex relationships between network-level damage and cognitive deficits. In addition, the non-orthogonality of cortical and white matter damage makes it difficult to disentangle their distinct contributions. Here, we used graph-based multilayer network analysis to address these limitations and evaluate clinical relevance. Using neuroanatomical and longitudinal neuropsychological data from 252 patients who underwent awake neurosurgery for low-grade glioma, we constructed interactive, three-layer networks for each hemisphere. Layer 1 comprised neuropsychological tasks (NT), layer 2 structural disconnections (SD), and layer 3 cortical damage (CD). Nodes represented tasks, white matter tracts, and cortical parcels, respectively, whereas within-layer edges captured correlations in performance or co-occurring damage patterns. Multilayer community detection identified domain- and hemisphere-specific brain-behaviour motifs linking executive, language, and spatial functions to distinct combinations of cortical and white matter disruption, a pattern confirmed by two spatial embedding approaches. Centrality analyses revealed a continuum of mapping relationships, ranging from one-to-one to one-to-many associations, indicating that tasks such as verbal fluency are better explained by multiple disconnection mechanisms. Additional analyses uncovered many-to-one and many-to-many relationships and highlighted tracts and cortical regions with domain-general relevance. Together, these findings support a neurobiologically grounded, network-oriented account of how structural brain damage gives rise to cognitive deficits, with implications for clinical care.
Related
- Multilayer MEG source modelling enables depth-resolved laminar inference in humans
- Multilayer interaction networks reveal multidimensional ecological strategies and context-dependent competition in soil actinobacteria
- Coupling with opinion dynamics promotes prosocial behavior in multilayer networks
- Distributed implantable microchips as wireless networks for neural sensing and stimulation
- A comparison of artificial neural networks and partial least squares modelling for the rapid detection of the microbial spoilage of beef fillets based on Fourier transform infrared spectral fingerprints
BibTeX
@article{nave2026beyond,
title = {{Beyond one-to-one mappings: Modelling distributed lesion-symptom relationships with multilayer networks}},
author = {Roxane Nave and Sam Ng and Sylvie Moritz-Gasser and Lorelei Berger and Emmanuel Mandonnet and Hugues Duffau and Guillaume Herbet},
journal = {Neuroimage-Clinical},
year = {2026},
month = sep,
doi = {10.1016/j.nicl.2026.104052},
url = {https://hal.science/hal-05736070v1},
note = {CC BY-NC-ND 4.0},
}
RIS
TY - JOUR
TI - Beyond one-to-one mappings: Modelling distributed lesion-symptom relationships with multilayer networks
AU - Roxane Nave
AU - Sam Ng
AU - Sylvie Moritz-Gasser
AU - Lorelei Berger
AU - Emmanuel Mandonnet
AU - Hugues Duffau
AU - Guillaume Herbet
PY - 2026
DA - 2026/09/02
JO - Neuroimage-Clinical
DO - 10.1016/j.nicl.2026.104052
UR - https://hal.science/hal-05736070v1
AB - Lesion-symptom mapping is widely used to identify causal relationships between brain structures and behaviour, and has played a central role in neuropsychologically informed network models of cognition. However, even recent approaches remain constrained by a one-to-one mapping framework, which oversimplifies the complex relationships between network-level damage and cognitive deficits. In addition, the non-orthogonality of cortical and white matter damage makes it difficult to disentangle their distinct contributions. Here, we used graph-based multilayer network analysis to address these limitations and evaluate clinical relevance. Using neuroanatomical and longitudinal neuropsychological data from 252 patients who underwent awake neurosurgery for low-grade glioma, we constructed interactive, three-layer networks for each hemisphere. Layer 1 comprised neuropsychological tasks (NT), layer 2 structural disconnections (SD), and layer 3 cortical damage (CD). Nodes represented tasks, white matter tracts, and cortical parcels, respectively, whereas within-layer edges captured correlations in performance or co-occurring damage patterns. Multilayer community detection identified domain- and hemisphere-specific brain-behaviour motifs linking executive, language, and spatial functions to distinct combinations of cortical and white matter disruption, a pattern confirmed by two spatial embedding approaches. Centrality analyses revealed a continuum of mapping relationships, ranging from one-to-one to one-to-many associations, indicating that tasks such as verbal fluency are better explained by multiple disconnection mechanisms. Additional analyses uncovered many-to-one and many-to-many relationships and highlighted tracts and cortical regions with domain-general relevance. Together, these findings support a neurobiologically grounded, network-oriented account of how structural brain damage gives rise to cognitive deficits, with implications for clinical care.
SN - 2213-1582
ER -