Note: You must be registered in order to post a reply. To register, click here. Registration is FREE!
T O P I C R E V I E W
petty
Posted - Feb 22 2010 : 3:13:25 PM We are implementing a change in the SENSE reconstruction's scaling factor to prevent data overflow. This will not affect activation signal, functional SNR, or derived statistics.
This change will fit imaging results into current standards that are compatible with FSL/etc. Previously the data was scaled by 1000 (raw value * 1000) and in a int16 datatype. This occasionally caused an issue with data overflow in a very small population of scans. The problematic data showed wrap-around errors, where a problematic voxel with be very negative (ie: -32000). To avoid that issue, the data-type was changed to uint16. This avoided the wrap-around and overflow issues, by providing a larger range of values.
Its recently come to our attention that the uint16 data, is not natively compatible with FSL. To resolve this issue, we are reducing the scaling factor to 500 and shifting back to the int16 data-type. By reducing the scaling factor we are much less likely to hit the maximum value allowed by the data type.
What this will mean for you is that signal intensity values will appear lower, which is strictly due to the scaling factor. So if you are comparing types of raw values, previous data will need to be scaled accordingly (by half). If anything is dependent on a raw value (ie: masking thresholds, etc), those should be adjusted also. Percent signal change and derived stats will not be affected.
