- Fixed bug in ROI itkRegion definition, missing division by spacing
- Consistent results using Small (1mm) or Large (2mm) FoV
- Some initial work on LUT, to polish...
- not usuable util the opti and metric types are better defined using enums
- ROI are passed to both metrics
- Full functionalty only for Powell(NCC or MI), mainly due to GetFinalR23Parameters which get those from Powell....
- itkDRT has method to set whether rotations should be used for import offset and hidden transform. Results checked against external code for final PAT conversion.
- helpers for query of station name and Fov moved to helpers functions file
- preparation of station name check in scout processor load of 2D and 3D images
- offset calc is verified against outside script. To be careful now in case we want to exclude the rotations from import offset. For PAT calc we need the complete matrix, therefore we may need to filter the rotations somewhere else.
- initial work on signal of current transform. was based on readout of spin box. i would use internal storage and check it against spin.
- i leave it broken so that we know where to continue
- automatic registration actually makes use of ROI. Before were ignored.
- cleanup of user and hidden transform and pass to UI after registration.
- change of internal image interpolator used in autoREG. Now autoREG performs well.
- general number of working units variable.
- GetFinalRotations, GetFinalTranslations and GetTransformParameters become obsolete.
- GetFinalR23Parameters and GetCompleteIsocentricTransform and GetUserIsocentricTransform are provided.
- only GetCompleteIsocentricTransform is missing implementation.
As a result, the feedback to regDialog is now already correct, no need for flips or sign change.
CalculateExternalTransform becomes obsolete... and has been commented out.
Methods for gui to get latest paremeters have been rewritten.
- introduction of CalculateInternalTransformV2
- Each projection maps the isocentric transform (RTIso COR) to its own projection center, including calibration offset.
- Direct comparison with previous implementation of CalculateInternalTransform not done... yet.
- Results on validation dataset are good.
- This change was triggered by rather large errors (ca. 0.8mm) on validation dataset.
- Both fakeIso and DCM RTPlan iso have been updated to V2.
- panel offset moved to gSiddon, output image w/ offset are available to reg23
- calibration in settings.ini is a mix of fitting and manual review in SRT
- calibration images reviewed (all: short long AP PA)
- Test with validation dataset is good. However only for IECS2CT2 w/ translations only. Import offset and hidden transform confuse the situation. Unclear if should be used or not when w/o resildual wrt LT is belo 0.3
- registration dialog is modal.'
This include as most important changes:
- itkImageProcessor UpdateProjectionGeometryMeta - SetOriginLPS now includes that offset
- ui_qvtk_handler: change in overlapping of two images. This now use the vtk provided image center which also consider image orientation and allows us to work under no assumption of center at same position between DRT and TOPO
This release was tested on Imaging calibration data. Consistency was good. On Phantom data also good consistency on Correction Vector between SHORT and LONG
- Interpolating all available space in DRT geometry.
- Isocenter visualisation based on same filter as contours
- Wrting SOP Class and SOP Instance UID in SRO
