| Περίληψη: | Computed Tomography (CT) is an imaging modality with many advantages but is lacking of the ability to image the tissues in the parenchyma of the brain and brain tumours in early stages. Aim of the presented research is to investigate whether CT could be used to image the soft tissue of the brain since this would provide an extremely powerful tool in the hand of the clinicians.
CT due to its low cost, high speed and ability to image simultaneously hard and soft tissue is a very often used imaging modality especially in emergency cases. Enabling CT to differentiate brain parenchyma, and thus also brain tumours in early stage, would greatly increase the chances of detection of many pathologies that are now imaged only with MRI such as demyelinating diseases, dementia, cerebrovascular disease, infectious diseases epilepsy, and brain tumours in early stage. There are many cases reported that brain pathologies (e.g. tumours) have been diagnosed from CT scans performed for completely different reasons. An advanced Brain CT Imaging technique could lead in much better monitoring of Central Nervous System (CNS) pathologies.
Main aim is to differentiate gray from white matter with X-ray imaging, since it will allow many pathologies to be detected, close the gap with MRI and is a condition that will lead to the visualization of tumours in early stages.
An investigation was carried out using simulated and real data in order to test the feasibility of using Dual Energy CT for imaging brain soft tissue. An in-house developed X-ray Imaging Simulator was used for the production of the simulated x-ray images.
In order to perform the investigation a brain model had to be used and altered in order to be applicable in x-ray simulations. As a second step small brain tumours were inserted in the created model in order to test the imaging of early stage brain tumours. Many different DE techniques were used and it was decided to use a non-linear decomposition dual energy technique for blending the images from different energies. The DE algorithm truly managed to produce contrast between brain parenchyma tissues in all a cases used even in real data.
Due to the great number of images created for all different settings and cases tests, the need of a good and reliable Figure Of Merit (FOM) came up, thus an in home FOM ware developed based on line profiles which had very good results and great ability of detecting the best images.
Additionally an optimization of the DE decomposition algorithm is proposed, based on solving inversely the problem of contrast between pixels and trying to find the local minima of the functions that relates them.
|
|---|
