crcrewso / dicomstrictcompare Goto Github PK

The project name and namespace no longer make sense
Repository should be changed to DCM Batch Compare
Namespace should be DcmBatComp

The version of GPGPU calculation used with ALEA does not improve performance. All references remaining should be removed, however the code should remain flexible enough to reintroduce gpgpu features if a suitable library is discovered.

Target files that are not compared to source should be listed in a trivially obvious fashion

Add proper gamma algorithm in contrast to dta currently used

When producing PDD Plots, the surface to 80% and surface to 50% depths should be measured and reported on the PDD plots.

Identified by @AndrewWAlexander

Extends Issue #4

Create profile plots at specific depths specified by user, in addition to PDD's. These should be static based on the widest field size, or dynamic, depending on radio button.

Add the ability to create PDD's with percent then dta analysis. User configurable tolerance.

Give users the choice of matching by UUID, plan name, field parameters, or some combination.

Change matching algorithm to look at the jaw parameters of the dose files, not just the names.

Create a settings class to pass as an object so the number of function arguments stop growing with every feature

Like #20

Create Results class that replaces the use of Messages as the default mechanism to store results for each pair This will allow for more sane result reporting

Error in source geometries outside the control of this software:

Discrepancies have identified the potential for an off by half width registration between sources being compared. Add a DTA option, defaulting to half (or 1 quarter) the voxel size, to ensure grid to grid comparisons are not failing due to sub-native resolution geometric shifts.

Identified by @AndrewWAlexander

Export the gamma value for each comparison to an excel sheet. Include the name of the plan

Display of excel csv file is scattered

• Nicer way of displaying algorithm, plan and field name vs current plan name
• move percent failed to the left
• display MU's of each field /dose file
• add column flagging if dimensions of dose grid differed

For QA/Sanity purposes, report Resolutions along with MU's...

Tolerance and epsilon should be relative % in the view and relative float in the controller.

I.e. user enters 1. all functions consistently expects 0.01

SNC Patient refers to this delineation as Van Dyk local vs global.

Implement a comparison extension so matches are sort-able by time to calculate

mismatched > dimensions match
then number of voxels
then alphabetical by planname^fieldname

when the Produce PDD checkbox is not checked an unhandled exception is produced

possible solution - create default phrase in object constructor that is overwritten when PDD results are returned

Replace MS plotting (deprecated) with something
Possibilities include PLplot and plplotnet

Add user side features to control the number of surface pixels deep to ignore. Useful when two algorithms have different edge of volume calculations.

Add the ability to dynamically calculate the overlap of two different resolution dose files and compare them

The program should fill empty dta information with defaults or properly handle exception, not crash rudely

allow/require the presence of RP files, reading them in to get the plan names appropriately.

Add relative comparison
Relative means tol % of max dose
absolute means voxel to voxel difference is < tol %

For compatibility with other frameworks convert, wherever possible, all lists of doubles to fixed size arrays