<< Click to Display Table of Contents >> Navigation: Correction/Calibration of twilite Measurements > Correction and Calibration Procedure > Cutting and Resampling |
The Correction window offers two functions for editing of the Corrected & calibrated curve: removal of data segments, and data resampling.
Data Cutting
As soon as Cut between markers is enabled, a blue area is shown in the curve area, typically at the extreme right hand side. This cutting definition area can be placed on a curve section by dragging the edges or the center line.
When the Cut button is activated, the samples covered by the cutting area are replaced by samples linearly interpolated between the left and right area boundaries as illustrated below.
As the immediate boundary samples are influenced by noise, averaging of a number of samples close to the boundaries can be enabled by the he Average flag. Samples within that duration before and after the cutting edges will be averaged to provide start and end values for the interpolated portion.
Data Resampling
If the native 1 second sampling resolution of the twilite and PSAMPLE is used, the raw data may be too dense for post-processing on older computers, slowing the calculations down. Therefore, the data can be resampled (resulting in implicit smoothing/denoising) as follows: enable Resampling, and activate Edit Timing. A dialog window is shown which lists each sample with a START and an END time relative to the PET scan start.
Use the EDIT TIME functionality to replace the sampling interval for either the entire curve, or in sections. For instance, entering
and then activating Insert Sampling Segment will replace the 1 sec samples between 5 and 10 minutes by 5 sec samples, whereby 5 raw samples are averaged. Insert Sampling Segment can be applied several times to create a sampling scheme with variable intervals.
After confirming the timing definition with Ok, the Corrected & calibrated curve is shown with the new sampling and can be saved for post processing.
Given sufficient processing power, we recommend transfer of all data points to PKIN, where functions can be fitted to the whole blood data. This results in a noise-free input function to which whole-blood/plasma and authentic parent correction ratios can be applied.