Optimizing interface and performance of basic VolumeGetBox and VolumeSetBox procedures.  Attempting to explicitly move DIM3D structure into FRAME class.

In all 4D to 3D registrations, user is offered a choice of currently selected timepoint as a Base, or instead determine timepoint of Maximal Information.  In PET to Anatomy registration it is always
Maximal Information.  Maximal information timepoint is assumed to have the best contrast between tissues and thus provide the best registration quality.

In all registrations, 3D and 4D,  implemented mechanism of tolerating the memory allocation failure.  Since Finetune is highly parallelized at many leveel and pet-thread memory requirements are high, for large images and high core count a memory allocation error occurs.  Old versions of FireVoxel either failed or lately the Parallel mechanism was disabled.  Now fully parallel veersion is ran,  the threads which failed are recorded into "Failed" thread array and then ran in serial mode.

Sokoloff model verified working on Brookhaven data.

For ALL  FireVoxel operations using the Vector ROIs as seeds:  new better logic is implemented.
For when a single seed is required and several seeds are present, Fvx requires a single seed to be
selected and warns otherwise.

Ensure correct functioning of the Anatomy-constrained (tube) IDIF function on the Brookhaven PET data.
In this example, two Macroseeds left and right are placed by User on the branches of the carotid artery.
Algorithm ran twice with identical (default) parameters.

Positioning of both corresponding ROIs are shown on orthogonal projections.

Testing latest Anatomy IDIF function on the PET case from  Brookhaven.  This (along with blood data) will be used to validate the IDIF function.

Performed a better registration for SL patient (based on URAL registration measure, not MI), created a Hybrid document, placed Macroseed for the new IDIF function.  Uploading the document to the SkyDrive.

Finalizing the parameter Dialog Box for the latest   Anatomy Constrained Input Function segmenter.

Specifically for calculating distance of the fit of 2 curves that are sampled at highly uneven intervals:  Implemented the function for L2 integration based distance for two functions F(t) and G(t).  Perform exact integration of 2 piecewise linear functions.

Replace the "goodness of the fit" with the new integration function.  Old function was simply a sum of difference at sampled timepoints that was giving unjust weight to the beginning of the curve were sampling was much more dense.

Corrected defect:   after IDIF is segmented, the TAC dialog pops up and now correctly displays the size of the vessel in (mm3) and appends it to the name of the vessel correctly.

MG tumor patient:   created a well-registered  MR-PET overlay.  Included an easy and well working Macroseed.  Uploading this to SkyDrive along with FireVoxel Build 129A.

Substantially increased quality of the registration of PET to MRI in the new version comparing to the variant of registration used by Anne.  This is for the patient MG.  (old=left, right=Build 129A).
Also, with the respect to "reconstructing  PET with MRI priors", notice a very large angle between MRI and PET in the original data.

And new registration result in multiple projections:

Temporarily disabled the parallel version of Amoeba optimization in the Finetune stage of the registration.  This is until we moved to 64-bit code.  This is due to high memory requirement by
certain registration measures, such as URAL. At 64-bit memory above 3GB (under Win) will be accessible.

FireVoxel Build 129 is released.

Tested new "Anatomy constrained" IDIF function segmenter on the PET dataset provided by Henry.

Dialog ROI Stats 4D (TAC viewing dialog):  removed redundancies so the dialog "start-up" time is 2X times faster now. This might be very substantial for large images and large number of timepoints.

Anatomy constrained IDIF extraction function: Continuing testing\debugging.

Will output vessel ROI resembling a tube of the
specified diameter.  User has to specify "dominant direction" for the vessel.
Otherwise parameters are similar to the old IDIF variant.

Performed initial evaluation of BiCal algorithm on Greg Chang's 7T hip dataset.

Non-uniformity (measured by 48 manually placed seeds) is reduced from the original NU=0.397, down to NU=0.127.

original with seeds,  NU=0.397

BiCal-corrected, NU = 0.127

FireVoxel Build 128I is released.