Repository logo

Bioengineering Analysis of Traumatic Brain Injury



Change log


Pan, Shijia 


Traumatic brain injury (TBI) is a serious health concern affecting over a million people in the UK. Brain shift and herniation, which are closely related to severe disability or death, are important signs of abnormally elevated intracranial pressure (ICP) or space-occupying intracranial mass after trauma.

This research aims to use medical image computing and biomechanical modelling techniques to characterise the specific deformation field of brain tissues under various TBI scenarios and strengthen the biomechanical understanding across the full spectrum of TBI.

Medical image computing provides the research with a solid clinical grounding. To better interpret the neuro-images, three computational tools have been developed, including a CT preprocessing pipeline, an automatic mid-sagittal plane detector and an automatic brain extractor. Using these tools, a novel concept of midplane shift (MPS) is developed to quantitatively evaluate the brain herniation condition across the mid-sagittal plane. In the meantime, a lesion heatmap is generated to quantify the asymmetric haematoma volumes across the mid-sagittal plane. The MPS heatmaps generated for 33 TBI patients with heterogeneous brain pathologies demonstrate highly similar shift patterns. Together with the lesion heatmap, a brain deformation mechanism has been presented: the brain will not deform randomly in response to trauma, instead, it will only deform in a regulated mechanism so that the deformation is directed and restricted to the soft ventricular region, thanks to the anatomic structures of the head such as the falx. The MPS heatmap, the lesion heatmap, together with the novel CT parameters derived from them, provide a rich abundance of information on intracranial brain herniation, for a more complete overview of TBI from medical images.

Biomechanical modelling, being one of the most important tools in trauma biomechanics, has been used to quantitatively simulate the brain shift and herniation condition caused by various intracranial lesions and increasing ICP. Preliminary finite element models reconstructed from the Virtual Human Project have demonstrated some limitations. To resolve the observed deficiencies, an advanced high-fidelity patient-specific FE brain model is constructed and explicitly assessed to optimise its injury simulation performance with the help of the developed medical image computing tools. During simulation, the patient-specific traumatic injuries have been reconstructed by imposing both the primary lesion and the secondary injury. The primary lesion simulation is achieved mechanically by ``indenting" a rigid lesion surface simulating the shape of the haematoma to the brain model. While the secondary swelling is modelled with a thermal-expansion-based method to simulate the bulging brain. Using this approach, the observed brain herniation can be decomposed into a deformation due to pure mass effect of space-occupying primary lesion and a shift as a result of secondary swelling. The head injuries of six different TBI patients have been reconstructed and simulated using the prescribed method. The realistic case study suggested that the subdural haematoma patients, as compared to the epidural haematoma patients, were exposed to more significant secondary swelling, which agrees well with the historical clinical findings. In addition to the realistic TBI case studies, an idealised traumatic lesion simulation is performed to investigate the role of lesion morphology and the lesion locations of onsets, in brain herniations during TBI. It is suggested by the idealised TBI cases that the brain is more sensitive to lesion that is more concentrated spatially, if lesion volumes and lesion locations were exactly the same. Moreover, in terms of lesion locations, lesions that strikes on the temporal region and the anterior region are more likely to lead to greater brain deformation, if other lesion morphologies were equal and no secondary swelling considered.

Ultimately, the developed tools are expected to help clinicians better understand and predict the brain behaviour after the onset of TBI and during subsequent injury evolution.





Sutcliffe, Michael


Traumatic brain injury


Doctor of Philosophy (PhD)

Awarding Institution

University of Cambridge
WD Armstrong Trust