Small Animal Imaging Resource Program @ Johns Hopkins


Truncating HRRT data

The Siemens HRRT PET camera is a high resolution dedicated to brain imaging. An HRRT data set is very large as a single frame has dimensions of 207 slices X 512 columns X 512 rows = 207x512x512 = over 50Mb! Thus a 20 frame study would have over 1Gb! For such a large data volume, at least 2Gb RAM is needed, and even then the data processing will be painfully slow. To alleviate the problem, it is often possible to load in a subset of the data since much of the image lies outside the brain. By reducing the data set, image computations become much more manageable.

Using Analyze
Step 1 : Select File->Load As then in the popup window select the file you wish to read in.
Step 2 : Initially read in only a single late image frame. E.g., if there are 22 frames you can read in frame 19 or 20. This gives a good view of the overall uptake and image boundaries. Use the SubRegion tab in the Load As window to select only a single frame 'V'. At this point you can also do a conservative truncation of X, Y, and Z, making sure that nothing in the subject is truncated.
Step 3 : Load the image into the ROI tool. It is easiest to first determine how much to truncate in the 'Z' direction. Simply scroll through the image (use Generate->Slice) until you determine the axial extent of the images. It is generally the case that the first 6 or 7 slices may be discarded, and in human studies the last 50 or so slices are beyond the top of the head. Thus of the 207 slices, we may truncate to slices 7-150 or so.
Step 4 : With the image highlighted in the main Analyze window select File->Save As. Use the SubRegion tab to experiment with truncating different X and Y ranges. Use the preview button to verify that important parts of the image are not being cut off. In most cases the X can be truncated at least 30 voxels on either side and the Y can be truncated by 20 on the bottom. The nose of the patient is usually near the top in the Y direction. It's not important for activity measures but it can help with coregistration so keeping the nose is recommended.
Step 5 : After X, Y, Z truncation is determined, the entire image volume including all frames can be loaded using Load->As with the appropriate truncation specified using the SubRegion tab.