Added User Interface-level functions:
a)Measure Surface Wave distance to Inside of the ROI
b)Measure Surface Wave distance to Outside of the ROI

SIEMENS De-identified DICOM load:  corrected the logic so the 4D dataset could be logically combined and loaded properly.  The problem was in De-dentifies replacing the TAG_PATIENT_ID differently for each timepoint.

Two timepoint T0(2009) and T1(2011) were
a) Whole brain segmented for masks M0 and M1.
b) Brain masks M0 and M1 were co-registered.


Result is below:
M0 - red mask (earlier)
M1 - blue mask (later)
Purple (red+blue) = overlap of M0 & M1

Prototype of Sulci-segmentation from T2-weighted image  MS31a_T2 using HullWave algorithm

I had to try BrainMask on T2, was not able to find optimal parameters to generate full BrainMask on T2.  But result seems to be good for the upper surface of the brain relevant for Sulci-segmentation.

HullWave works with the binary brain mask obtained from T1,T2 using any segmentation algorithm (FireVoxel\EdgeWave was used in this instance).

Map of detected Sulcal depth (in 0.01 mm)

Volume Arithmetic operations on very large volumes:  Optimized the memory workflow so the operations do not fail on volumes with upto 1.5 billion voxels /timepoint.

DICOM load of very large: optimized memory workflow so no memory failure happens on 1.5 billion voxel volume.

Volume histogram calculation: optimized the memory workflow so very large volumes could be processed.  This function is part of every volume loading operation.  Added optional progress indicator.

Volume Affine transform:  optimized memory usage  and workflow so it is possible to have a destination volume exceeding 1.5 billion voxels. This includes viewing of the resulting volume (rest of the image processing operations is not guaranteed due to 32-bit limits).  Added an optional progress indicator during the Affine reslicing.

Fixing defect:  ABT   Rotate+Scale subtest with the Scale=3 produces a crash.

Continued the work to correctly create DICOM header after the transformation {Affine}.
Added ABT test for the Rotation+LargeScale of the 3D brain volume, and checking the SaveDicom\LoadDicom combination.

ABT: adding a subtest for Rotate+Scale of large volumes with all possible interpolations.

Corrected the crash related to DICOM features:  when volume is transformed with the dimension change (or similar operation),  new internal DICOM headers have to be rebuilt.  Presently it produces a crash because the new number of slices can be bigger.

When volume is reformatted (f.e. changing dimensions), internal DICOM headers were not generated properly for the new volume, producing a crash.  Now fixed. 

Prototype of Sulci segmentation from  GM-CSF interface.
In the literature two possible approaches are described:
a) Having the Whole Brain (WM+GM) mask, segment Sulci using the GM-CSF interface.
b) Having the separate WM and GM mask, segment Sulci using the WM-GM interface.

MethodA is simpler to implement but there are much more difficult cases many sulci are too narrow to be reflected on the whole Brain Mask.

MethodB is considered more reliable, due to thicker layer of GM, but requires the WM-GM segmentation (not present in FireVoxel at the moment).

Here are the results of MethodA implemented in FireVoxel.  Sulci is represented as a color map where color corresponds to the length of the curved path from the "brain hull\envelop" to the particular point in sulcal space (measured in mm).

"Curved depth" map in 0.01 mm units.

Implementing, testing and debugging the Support Module for measuring the Wave Distances in various scenarios.

Sulci-ROI segmentation:  implementing function for defining the "Wave Distance" from the Brain mask ROI surface to its convex hull.

Experimenting with the "Sulci-ROI" detection algorithm using the brain mask (EdgeWave) as the input.

FireVoxel Build 148 is released.

1. Fixed the defect of crashing (greek)  GE DICOM.
2. Fixed the defect with initial document projection when loading the FVX document file.
3. Fixed the crash, when the RAW file picked from the "Recent File List" but actually missing on disk produces the crash.
4. Fixed the load RAW (.im) -> save FVX -> load FVX  failure.
5. Fixed the "Simple Model" not displaying the result of Parametric Map calculation.
6. ABT test:  added subtest to the "Signal measurements" (#0) model.
7. FireVoxel distribution package:  removed redundant libimp5md.dll ( Intel OpenMP)
8. Removed all the usage of Intel Performance Primitives and remove 4 corresponding DLLs from the distribution package.

9.  Fixed RoiStats3D dialog error in reporting the Median value of the ROI.

Fixed RoiStats3D dialog error in reporting the Median value of the ROI (reported by Tracy Butler).

Removed all the usage of Intel Performance Primitives and remove 4 corresponding DLLs (ipp*.dll) from the distribution package.

FireVoxel distribution package:  removed redundant libimp5md.dll ( Intel OpenMP)

ABT test:  added subtest to the "Signal measurements" (#0) model.

Fixed the "Simple Model" not displaying the result of Parametric Map calculation.

Fixed the load RAW (.im) -> save FVX -> load FVX  failure.

Fixed the crash, when the RAW file picked from the "Recent File List" but actually missing on disk produces the crash.

Fixed the defect with initial document projection when loading the FVX document file.

Fixed the defect of crashing (greek)  GE DICOM.

FireVoxel Build 147 is released.

1. Dialog RoiStats 4D:  corrected the synchronization effect between displayed TAC and the underlying volume window.

2. Blood Vessel map segmentation: Implemented additional controls to adjust the how aggressive the segmentation is:

a) Introduced "vessel time delay" parameter due to dispersion

b) L2-Normalization of TAC comparison, since the numerical values of TACs and IFs are quite different due to dispersion and better shape comparison is expected that way.

Parallelized segmentation algorithm.

3. DCMTK integration:  implemented the DICOM Save function.

4. DCMTK integration:  Implemented the DICOM Crop & Save subtest

5. DCMTK library is fully integrated.

6. Fixed the RAW file load defect (initial slice\Film view projection).

RAW file load defect (initial slice\Film view projection):  fixed.
This was due to recently introduced feature of remembering the initial View in the FVX file format, an unintended consequence.

DCMTK library is now 100% integrated.

DCMTK integration:  implementing 2 core support function for reformatting the DCMTK headers after crop and other attribute change operations.  This is used in every single FVX operation that generates a new volume after some processing.

DCMTK library integration:  adding the Automatic Build Test subtest when
a. DICOM document loaded -> VA
b) Result is cropped and saved to DICOM
c) Result is loaded from created DICOM -> VB
d) VA and VB are compared.

DCMTK integration:  debugging the   Load RAW->Save DICOM-> Load DICOM->Compare to RAW operation.

DCMTK:  implemented the creation of the Basic  Dicom header when loading a non-DICOM file (f.e RAW).  This is required for the Dicom-Save functionality.

DCMTK integration:  implemented the DICOM Save function.

Blood Vessel map segmentation: Implemented additional controls to adjust the how aggressive the segmentation is:
a) Introduced "vessel time delay" parameter due to dispersion
b) L2-Normalization of TAC comparison, since the numerical values of TACs and IFs are quite different due to dispersion and better shape comparison is expected that way.

Parallelized segmentation algorithm.

If we make parameters of Blood Vessel segmentation to remove more susceptible blood vessels we can bring the fa (Arterial fraction) all the way down to minimal 0.05.  At that level the modelling result seem to stabilize, i.e. removing even  more vessels does not make difference.

Dialog RoiStats 4D:  corrected the synchronization effect between displayed TAC and the underlying volume window.

Liver DCDI:  Performed alternative Blood Vessel segmentation, using the 4D algorithm that takes into account all the timepoints. Result is quite different, with fa=0.24 (comparing to fa=0.32 for the 3D vessel segmentation)

Liver DCDI modelling:  comparison of Whole Liver ROI   vs.  Whole Liver without Blood vessels.

FireVoxel Build 146A is released.

1. FVX file format, version4:  implemented format extension to be able to save current dynamic model and the Input functions.  This would be a great productivity multiplier both for end-User and developer.
2. General  4D "Blood vessel segmentation" algorithm and UI with User-provided input function(s).
3. Implemented and added an UI item  "MainMenu>Segment>Segment dark ridges"  which complements the existing "Segment bright ridges" functionality.
4. Implemented Liver DCDI  Blood Vessel Segmentation 3D:  this is an alternative algorithm that uses a single pre-contrast timepoint  assuming the vessels are darker than the liver tissue.  Existing "Dark Ridge detection"  algorithm is used.  
5. Fixed a defect in the IDIF function.

6. Added several "Keep Highest voxels for each individual timepoint" options as specified by Jean Logan.

Added several "Keep Highest voxels for each individual timepoint" options as specified by Jean Logan.

Fixed a defect in the IDIF function.

Implementing the new IDIF features requested by Jean Logan. Select top voxels {%,count,mm3}
at each individual timepoint to keep as a part of the IDIF mask.

Implemented Liver DCDI  Blood Vessel Segmentation 3D:  this is an alternative algorithm that uses a single pre-contrast timepoint  assuming the vessels are darker than the liver tissue.  Existing "Dark Ridge detection"  algorithm is used.  Existing Ridge dialog box is called.
Two ROIs are returned one for organ and the other for blood vessel.  There is no differentiation between Arterial and Venous vessels as in Segment4D variant.

Implemented and added an UI item  "MainMenu>Segment>Segment dark ridges"  which complements the existing "Segment bright ridges" functionality.

Implementing a 3-D base variant of the Blood Vessel Segmentation map.  This is in addition to the existing 4-D segmenter so the best option could be chosen.

Blood vessel segmentation  (using high "Organ Sensitivity Threshold"=20 to detect smaller vessels).

Blood vessel segmentation map in the multi-slice Liver ROI case:

Initial result for Liver Blood vessel map segmentation.  Procedure is general and is not Liver-specific.  User provides 1 or more Input Function by standard loading into the Dynamic Module. Whole organ ROI (liver in this case) is provided.  Result is the set of new ROIs for (NumVessels+1).   In this test  Arterial ROI was added for correctness verification.

Blood Vessel\Organ segmentation map: debugging algorithm and adjusting user interface.

FVX file format, version4:  implemented format extension to be able to save current dynamic model and the Input functions.  This would be a great productivity multiplier both for end-User and developer.

Blood vessel segmentation map:  continue the work on the novel segmentation algorithm.

FireVoxel Build 146 is released.

Tested the DTI problems reported by Andrew Rosenktantz but did not reproduce any defects.  Plan of action is proposed with the new Build to pinpoint the problem.

FireVoxel file format: correcting format so the complete "VOLUME_PROJECTION" structure is saved.  Presently save SliceIndex\TimeIndex result in frequent Blue Screen error, when opening the FireVoxel document.  New FVX format is backward compatible, i.e. old files will be loaded.

Fixed defects of Automatic WM brain seed detection.  Procedure was producing wildly incorrect results in case if some VOIDVOX values were present in the volume.

Some suspicious voxels present in brain image

Implementing a new UI-level function:
User Loads 4D volume + organ ROI.
One or more input functions corresponding to separate blood vessels
are loaded.
Function executes and the result is the Map of the blood vessels inside the
organ.

ABT  EdgeWave subtest: corrected defect: the difference in result between Release and Debug compiled versions.

Working on the final replacement of the DICOM legacy engine, with the DCMTK.

Fixed error with the floating Layer Control dialog box.  When 2 documents are present in the mainframe, and there is Layer Control dialog, switching the Active document updates the Layer Control correctly, but when the dialog values are modified the displayed images stop updating.

Corrected a defect in  RoiStats3D dialog when measuring the values over ROI and there are several 3D and 4D layers present (calling 3D layer with T>0)

LIVER DCDI case {LI}:  comparison of parametric maps {Total Flow,fa,EMTT,Ki}  REGISTERED vs. UNREGISTERED

Total flow,  UNREG (68.4 +/-41.4) - REG (133 +/- 55)

fa-Arterial fraction,  UNREG (0.64 +/- 0.35) - REG (0.37 +/- 0.34)

EMTT,  UNREG (26.1 +/- 12.1) - REG (17.7 +/-10)

Ki,  UNREG (4.67 +/- 1.37) - UNREG (5.28 +/- 1.64)

LIVER DCDI case {LI}:  comparison of parametric maps {Total Flow,fa,EMTT,Ki}  before and after whole liver 4D registration.

Total Flow, Unreg - Reg

fa-Arterial fraction, Unreg-Reg

EMTT, Unreg - Reg

Ki, Unreg - Reg

FireVoxel Build 145 is released.

Automatic Build Test: added subtest for the  PET "tau-ridges" segmentation.

Dialog Parametric Map calculation:  made the linear (T1W) concentration conversion consistent with the other parts of the FireVoxel.

IDIF functionality:  made a more User-convenient approach when unambiguous 4D volume is selected for the IDIF extraction.  Previously such layer had to be explicitly selected.

Restored the RoiStats4D dialog after the IDIF function is automatically calculated.

"Liver DCDI workflow" - added vessel names for AIF and VIF extraction functionality to clarify the meaning of the layers in the Layer Control Dialog and in the Dynamic Parametric Map calculation.

IDIF function:  when Auto-Direction is specified (f.e. Portal Vein),  the optimal direction is chosen by the intensity integral at the Peak timepoint, instead of simple count of the voxels in the vessel mask.

IDIF function full code review for a new release.

Implemented the IDIF function to perform internal conversion to Isotropic resolution so there is no preffered direction during the segmentation.

Automatic IDIF function:  simplified logic and made more intuitive interface - search of the MicroSeed is confined to the User-specified "vesel seed box".

Liver DCDI modelling, case SK:  processed the whole 71-timepoint dataset (100 min).  Vertical motion in the coronal plane is almost entirely eliminated.

Preliminary evaluation of the PV-IF show a substantial sharpening of this function.  Exactly the same manually constructed ROI is used in this diagram both for registered and unregistered case.

RED -  IF obtained from unregistered data,  BLUE- from registered.

MIDAS\.time  Loading defect correction:  internal times were converted to relative with the Base 0, when the .TIME file is loaded.  Corrected to include absolute times.  This was causing a numerous defects in multi-layer entity processing (f.e. two 4D layers after registration).

Increased responsiveness of the application when is inside the long process.  This is done by additionally using DoEvents() function when the ProgressMessage function is called and the DoProgress=false

Moved development to the new 6-core workstation. Rebuild application and ensuring that the Automatic Build Test checks out.

Debugging and verifying a highly-efficient function to determine variable precision during the Amoeba non-linear optimization, specific to image registration.  This will work across all registration applications 3D\4D.

Most recent version of MERGE workflow.  Note that only DWI4D functionality is not implemented.  Result of the workflow is the 5-layer volume residing within the same spatial coordinates and sharing the timeline.  This volume could be viewed simultaneously so all the required ROIs can be constructed by viewing all overlaid layers.

Registration iterative non-linear optimization: investigating a feature matching function to speed up the "Determine Variable Precision" stage of the optimization.

Implementing function to Append\Truncate 4D Volume along dynamic axis.  This will substantially improve  the performance and open additional options for the "Register 4D to 3D" scenarios.

Employing Affine transform further improves the quality of registration.

SK:  t=19, before registration

SK: t=19, after registration

Implemented functionality and User's Interface for Smoothing the binary Volumes (ROIs)

Liver ROIs, semi-manually constructed by drawing the contour on every 2nd slice required further smoothing.

Preliminary result: Substantial registration improvement is achieved on Liver-GRASP-DCDI exam

 First, User contructs a fairly precise "Whole Liver ROI" on a single timepoint (SK,t=25).
This is done by outlining the liver on every 2nd slice and applying Fill&Morph operation (total time=5 min).  This Whole Liver ROI will be used in registration and further liver function evaluation.

ROI contours are manually drawn on Axial projection and simultaneously displayed on Coronal.

Before registration: SK-case, timepoint t=20, and the Target Liver ROI (t=25)

After registration: Anatomy and the Target ROI are aligned.

Parameter set that has been  used.

GRASP-4D:  Liver is clearly separated with the URAL-gradient at rad=4.1.   As usual with fuzzy boundaries the radius has to be chosen large enough.

FireVoxel Build 144 is released.

When saving FVX file  format, all the volumetric layer are checked to contain PHI information.  Is yes, the dialog box pops up that recommends to exit and run the Deidentifier.

Implemented a "De-identification" functionality for the FVX file format.  This is to further improve the protection of PHI.  All the tags minimally related to PHI are completely cleared.
 A single function, without parameters is placed under  MainMenu>File>Deidentify.

Provided initial description of the PET "detect ridges" algorithm.

Restoring Automatic Build Test that was unsynched during the FDG\DEC MERGE workflow development.

Corrected the display of VoidVoxel values in the parametric maps

Corrected "Volume Histogram" and related function to properly process volumes containing the
VOIDVOXEL values.

Good registration result in  ASL to MR-GRASp registration
This requires some preprocessing of GRASP to obtain an accurate BrainMask.

PET workflow: implemented User Interface-level function:  Crop Empty Areas.

Implemented UserInterface-level   "Dynamic Integral" function:

Brain FDG-DCE  merge workflow.  This workflow can be made 100% automatic. 
INPUT:      folder containing DICOM file
OUTPUT:  Single merged FVX file, containing both MR4D and PET4D.

Implemented "Convert Mosaic to Volume" function and tested it on MR-ASL data.

Tested the "Arterial delay" parameter in the GRASP-DCE abdominal.  Setting "Arterial delay" up to 5 sec improves the fit substantially.

Corrected problem, during the Save\Load of the 13-layer registered document to FVX format.
For unknown reasons, an UNICODE  "_fgetts" function was failing (part of VC++).  Was replaced by a FireVoxel_fgetts.