Dynamic temporal patterns of DMN connectivity in epilepsy using hidden (semi-) Markov models.

Area of Science:

• Neuroscience
• Computational Neuroscience
• Medical Imaging

Background:

• Epilepsy is a neurological disorder defined by recurrent seizures.
• Altered default mode network (DMN) connectivity is implicated in epilepsy pathophysiology and seizure spread.
• Understanding DMN temporal dynamics is crucial for epilepsy research.

Purpose of the Study:

• To investigate temporal patterns of DMN functional connectivity in epilepsy patients versus healthy controls.
• To evaluate the efficacy of Hidden Semi-Markov Models (HSMMs) in characterizing dynamic functional connectivity (dFC) alterations in epilepsy.
• To identify specific dFC metrics indicative of disrupted DMN temporal organization in epilepsy.

Main Methods:

• Employed data-driven models, including Hidden Markov Models (HMM) and HSMMs with Gamma and Poisson sojourn distributions, to analyze dFC.
• Derived dynamic metrics: fractional occupancy, switching rate, and mean lifetime of brain states.
• Compared temporal properties of DMN connectivity states between epilepsy patients and healthy controls.

Main Results:

• Epilepsy patients exhibited prolonged dwell times in low-connectivity DMN states and reduced flexibility in state transitions.
• HSMMs, particularly the Gamma variant, showed higher sensitivity in detecting these DMN connectivity alterations compared to standard HMM.
• Group-specific transition patterns indicated disrupted temporal progression of DMN states in epilepsy.

Conclusions:

• HSMMs are effective tools for capturing alterations in functional brain states and DMN dynamics in epilepsy.
• Findings provide novel insights into the dynamic reorganization of the DMN in epilepsy.
• This study underscores the importance of flexible sojourn modeling in dFC analysis for neurological disorders.