Source code for the FORCASTER and FORCASTEST algorithm.
Data is available at: https://doi.org/10.11588/data/GIWRQA
Also includes as a reference algorithm a CMA-ES minimizer using the NGI objective function.
The config file contains the function get_proj_paths which generates a list with tuples that contain a name, the path to the dicom files that should be calibrated, the path to dicom files with the prior image, and then a list with the calibration algorithms. Calibration algorithms have a numerical identifier, here are some:
FORCASTER: 33 FORCASTEST: 62 Rough estimator: 60 Refined Estimate: 61 Refined Estimate and CMA-ES with NGI: 70 Only translational calibration: 4 FORCASTER with NGI objective: 42 FORCASTEST with NGI objective: 63 CMA-ES with NGI: -70 CMA-ES with Feature Point Objective: -72 BFGS with NGI: -58 BFGS with FP Objective: -44
Calibration algorithms with a number >= 0 reside in cal.py, those with negative numbers are in cal_bfgs_both.py
For dicom files that are not acquired with an Artis Zeego System the function read_dicoms in load_data.py might need to be adapted. If the dicom files do not contain a coordinate system for each image the creation of the astra geometry in main.py:358 has to be adapted or uncommented to use the astra geometry created with the angles from the standard dicom fields.