A comparison of diffusion tractography techniques in simulating the generalized Ising model to predict the intrinsic activity of the brain.

Diffusion tractography is a non-invasive technique that is being used to estimate the location and direction of white matter tracts in the brain. Identifying the characteristics of white matter plays an important role in research as well as in clinical practice that relies on finding the relationship between the structure and function of the brain. An Ising model implemented on a structural connectivity (SC) has proven to explain the spontaneous fluctuations in the brain at criticality using brain's structure depicted by white matter tracts. Since the SC is the only input of the model, identifying the tractography technique which provides a SC that delivers the highest prediction of the brain's intrinsic activity via the generalized Ising model (GIM) is essential. Hence an Ising model is simulated on SCs generated using two different acquisition schemes (single and multi-shell) and two different tractography approaches (deterministic and probabilistic) and analyzed at criticality across 69 healthy subjects. Results showed that by introducing the GIM, predictability of the empirical correlation matrix increases on average from 0.2 to 0.6 compared to the predictability using the empirical connectivity matrix directly. It is also observed that the SC generated using deterministic tractography without fractional anisotropy resulted in the highest correlation coefficient value of 0.65 between the simulated and empirical correlation matrices. Additionally, calculated dimensionalities per simulation illustrated that the dimensionality depends upon the method of tractography that has been used to extract the SC.

Consciousness and the Dimensionality of DOC Patients via the Generalized Ising Model.

The data from patients with severe brain injuries show complex brain functions. Due to the difficulties associated with these complex data, computational modeling is an especially useful tool to examine the structure-function relationship in these populations. By using computational modeling for patients with a disorder of consciousness (DoC), not only we can understand the changes of information transfer, but we also can test changes to different states of consciousness by hypothetically changing the anatomical structure. The generalized Ising model (GIM), which specializes in using structural connectivity to simulate functional connectivity, has been proven to effectively capture the relationship between anatomical structures and the spontaneous fluctuations of healthy controls (HCs). In the present study we implemented the GIM in 25 HCs as well as in 13 DoC patients diagnosed at three different states of consciousness. Simulated data were analyzed and the criticality and dimensionality were calculated for both groups; together, those values capture the level of information transfer in the brain. Ratifying previous studies, criticality was observed in simulations of HCs. We were also able to observe criticality for DoC patients, concluding that the GIM is generalizable for DoC patients. Furthermore, dimensionality increased for the DoC group as compared to healthy controls, and could distinguish different diagnostic groups of DoC patients.