swi - Susceptibility Weighted Image post-processing

blue bar

In this sense, susceptibility weighted imaging refers to the post-processing technique which highlights image contrast in T2* weighted images depending on relative phase shifts. In this implementation, both the modulus and phase T2* weighted images are required. The phase map undergoes high pass filtering to reduce slow-varying phase changes across the object, and is then used to produce a mask where positive phase yields a 1 and negative phase results in a value linearly falling from 1 to 0 as the phase reduces from 0 to - π. This mask is raised to the fourth power and multiplied by the original modulus image to produce the SWI.

swi is a command line program to peform this analysis. Options may be entered on the same command line; if the minium set of options are not present then the program will prompt for the relavant missing values. The phase maps should be scaled to radians, either as floating point pixels or by using appropriate scaling and offset values in the Analyze header. During execution, the program will also save the high-pass filtered ("_hpf") phase map and the phase mask ("_mask") calculated from it. For multi-volume files, each volume should be processed independently (untested). 

swi [-2d | -3d] [-zz] [-f <num>] [-m <num>] [-d]
       [input_modulus_filename] [input_phasemap_filename] [output_SWI_filename]
-2d
Assuming a 2D multislice T2* acquisition, processes each slice separately using a 2D Hann filter.
-3d
Assuming a 3D volume T2* acquisition, processes the volume using a 3D Hann filter. This is the default.
-zz
Depending on the MR scanner configuration, the phase map may have been masked already, with the background intensity values set to zero. If conversion to radians is to be obtained by using the scaling factor and offset values in the Analyze header then the background may be converted to -π. By specifying this option, zero intensities in the input phase map are preserved regardless of the offset value in the Analyze header. However, the same will occur for true zero phase values inside the object.
-f <num>
The effective high-pass filtering of the phase maps is performed by first producing a low-pass phase map from the input complex data and then dividing the original input complex data by this low-pass phase map. The low-pass Hann filter used is, by default, 0.125 times the width of the image matrix. This fraction may be specified as num explicitly by using this option.
-m <num>
The number of times to multiple the modulus image by the phase mask in order to obtain the SWI. The default is 4.
-d
Displays additional diagnostic output.
input_modulus_filename
The filename of the Analyze file containing the input T2* weighted images. Filenames may contain either the .img or .hdr extension, or neither. If the input_modulus_filename is not specified on the command line then the program will prompt for a filename.
input_phasemap_filename
The filename of the Analyze file containing the input phase map of the T2* weighted acquisition.
output_SWI_filename
Analyze file which will contain the processed SWI data.

References