Whole brain tractography in Malaysian patients using DMRI

Abstract

Language processing involves multiple brain regions in a predominantly left lateralised cortical network. Language networks however are still very much elusive due to complexity of brain activity profiles that may vary considerably among different individuals particularly when there is a tumour present in the network. The ability to delineate fibres accurately is fundamental in the treatment planning of tumour patients. Tractography is useful in mapping white matter networks, however numerous methods like deterministic, probabilistic and global exist, each presenting with advantages and limitations. This study thus aims to investigate and map structural networks in patients with a left hemisphere primary brain tumour and healthy subjects using dMRI and tractography, with the intent of comparing these three methods. In this study, retrospective Hospital USM radiological 3T MRI treatment planning data between September 2012 and December 2017 of 6 left hemisphere tumour patients (3 gliomas, 3 meningiomas) and 3 radiologically normal brain patients is being used. Manual processing and packages DSI Studio, MRtrix3, ITK and FSL were used for pre-processing the data, setting up ROIs and running the different methods of tractography analysis. Fibre length and coefficient data were statistically compared using SPSS. The results indicated that probabilistic algorithms delineated the selected networks better than deterministic. Global tractography yielded slightly better results however in the case of tumour patients some false positives were indicated. Deterministic analysis terminated prematurely in line with previous research. Local probabilistic seems to yield more applicable as previous research has indicated in the case of other pathology.