Using functional magnetic resonance imaging to plan surgical resections of brain tumours
View/ Open
Date
02/07/2013Author
Gorgolewski, Krzysztof Jacek
Metadata
Abstract
Brain tumours, even though rare, are one of the deadliest types of cancer. The five
year survival rate for the most malignant type of brain tumours is below 5%. Modern
medicine provides many options for treating brain cancer such as radiotherapy and
chemotherapy. However, one of the most effective ways of fighting the disease is
surgical resection. During such a procedure the tumour is partially or completely
removed.
Unfortunately, even after a complete resection some tumourous tissue is left behind
and can grow back or metastasise to a different location in the brain. It has been
shown, however, that more aggressive resections lead to longer life expectancy. This
does not come without risks. Depending on tumour location, extensive resections can
lead to transient or permanent post-operative neurological deficits. Therefore, when
planning a procedure, the neurosurgeon needs to find balance between extending
patients life and maintaining its quality.
Recent developments in Magnetic Resonance Imaging (MRI) fueled by the field of
human cognitive neuroscience have led to improved methods of non-invasive imaging
of the brain function. Such methods allow the creation of functional brain maps
of populations or individual subjects. Adapting this technique to the clinical environment
enables the assessment of the risks and to plan surgical procedures. The
following work aims at improving the use of functional MRI with a specific clinical
goal in mind.
The thesis begins with description of etiology, epidemiology and treatment options
for brain tumours. This is followed by a description of MRI and related data
processing methods, which leads to introduction of a new technique for thresholding
statistical maps which improves upon existing solutions by adapting to the nature of
the problem at hand. In contrast to methods used in cognitive neuroscience our approach
is optimized to work on single subjects and maintain a balance between false
positive and false negative errors. This balance is crucial for accurate assessment of
the risk of a surgical procedure.
Using this method a test-retest reliability study was performed to assess five different
behavioural paradigms and scanning parameters. This experiment was performed
on healthy controls and was aimed at selecting which paradigms produce
reliable results and therefore can be used for presurgical planning.
This allowed the creation of a battery of task that was applied to glioma patients.
Functional maps created before the surgeries were compared with electrocortical stimulation
performed during the surgeries.
The final contribution of this work focuses on technical aspects of performing neuroimaging
data analysis. A novel data processing framework which provides means
for rapid prototyping and easy translation and adaptation of already existing methods
taken from cognitive neuroscience field is introduced. The framework enables
fully automatic processing of patient data and therefore greatly reduced costs while
maintaining quality control. A discussion of future directions and challenges in using
functional MRI for presurgical planning concludes the thesis.
Collections
The following license files are associated with this item: