MSc projects 1516 001

Software tools for translational clinical to pre-clinical magnetic resonance image analysis.

Abstract:
  • Background: This project was about the design and software implementation of image analysis methods for brain magnetic resonance images (MRI) of animals, based on existing structural human brain image processing methods.
  • Methods: The algorithms developed were registration, brain extraction, white and gray matter detection, lesion detection and brain structure segmentation. All methods, implemented in MATLAB Graphic Unit Interfaces, are either fully or semi-automated and were tested using T1-weighted, T2 weighted and fluid attenuation inversion recovery (FLAIR) MRI of rodents and a dog.
  • Results: Statistical analysis of the results obtained from extensive numerical experiments illustrate that registration, brain extraction and brain structure segmentation perform well. Lesion detection methods, despite performing well in identifying lesions, have a high rate of false positives detected. To improve the process of lesion identification, prior knowledge of the lesion confounds needs to be incorporated to the algorithms.
  • Conclusion: Finally, further work needs to be done to make the process tolerant to inherent noise and inhomogeneity artifcats, with the use of machine learning techniques and robust bias field correction algorithms.
Project type:
  • Analysis of existing data
Imaging keywords:
Application / disease keywords:
Supervisor(s):
Programme:
Year:
  • 15-16