Ayan Sengupta renamed Methods.tex to Methods continued.tex  over 9 years ago

Commit id: bc1271c9b064792aa9328c8e0cc230d4ebb32d2d

deletions | additions      

         

\section{Image Processing Methods}  The Image processing methods consist of few pre-processing steps. Generally the volumetric magnitude image is used for masking the ROI in the phase volume. In order to do that a standard brain extraction(FSL BET) is performed and also a binary brain mask is generated. The extracted brain thus obtained is further co-registered with the MNI standard 1mm brain using FSL FLIRT and corresponding forward and inverse transformation matrix are generated for projecting analysed veinous masks to and fro the subject space and standard space.   \subsection{Method 1}  This image processing method considers only the magnitude volume of SW images. This method is not perfect, but this gives a 'quick and dirty' way of looking into the distribution of veins in a particular ROI. The extracted magnitude brain does show the veinous arrangements but it is dificult to segment it out, because the veins appear black in the magnitude image with not so good contrast with surrounding tissues.   \begin{itemize}  \item \textbf{Step 1: Inverting}  The magnitude image is inverted by multiplying each voxel value with '-1'.   \item \textbf{Step 2: Normalizing and Contrast Scaling}  The inverted image is normalized in the range '0-1' by subtracting each voxel value with the minimum value and dividing it with the range. Then It is scaled along any arbitrary intensity scale as required. In this step it might be considered to mask the image in order to restrict the analysis to a particular ROI.  \item \textbf{Step 2: Thresholding}  Upon looking at voxel intensities of the invered and scaled image in fslview, aparticular threshold limit has to be decided. All voxels having intensities avove the particular set threshhold are going to be considered as 'veins' voxel.  \end{itemize}