Computational modelling of traumatic cerebral vasculature injury

Abstract

The variety and severity of the physical and emotional effects caused by Traumatic Brain Injury (TBI) has made it stand out among other prevalent diseases. Prognosis can vary substantially between individuals despite similarity in injury cause or demographics, which makes each case complex and unique. Microbleeds are a pathology seen from TBI and their occurrence and amount are indicative of poorer outcomes. It is thought that their occurrence is linked to biomechanical factors of the vessels; however, to-date there is no study which has investigated this as of yet. This work aims to understand whether extension and/or rate of extension of the veins in the brain is a key component of injury seen during a TBI. First, the computational method smoothed particle hydrodynamics was used to test whether greater accuracy in displacement was seen when modelling the fluid-like structure surrounding the brain, a medium important in transferring impulses from the rigid-esque skull to the soft brain matter. Good agreement to in-vivo brain displacements was found when using the method, which was to the same level as the traditional method. A greater accuracy in strain was also seen; however, the instabilities and resource requirements of the method were found to be impractical and was therefore unable to be taken forward for the following modelling studies. Secondly, to successfully model damage to the venous system a highly detailed model was created from three and seven tesla brain images to create representations of over 50 structures in the brain. The venous system was segmented to diameters as low as 0.33 mm from the same patient using the bioimaging software Materialise Mimics. Reconstructions of 46 TBI cadaver tests across five subjects from literature were used to compare actual to predict displacement of the brain. Good agreement of displacements was found in all tests showing the model is capable of accurately predicting displacements of injury events typically seen during a TBI. The efficacy of the model for predicting vasculature damage was shown through reconstruction of a case study where microbleeds were seen. The strain and strain rate data from the simulation were found to peak in a similar location to those of the microbleeds found in the subject showing a correlation between the vein biomechanics and pathology. Lastly, two more case studies with microbleeds present were simulated, in addition to and enhanced version of the case study used previously. The brain was divided into lobar or tract-based regions for statistical analysis. The regions which experience larger 95th percentile axial strain and strain rates were seen to correlate with the occurrence of venous damage. A large cohort (n = 74) of patient data was used to show that the regions of the brain which frequently saw microbleeds also saw above average 95th percentile axial strain and strain rates. This shows damage to the cerebral veins, which cause microbleeds, is likely linked to axial extension they undergo during a TBI event. Limitations exist with respect to the work presented, such as sample size, variety of case studies, and exclusion of the cerebral arteries; however, these are found to not limit the impact of the work and are natural topics for future study and investigation.