Click here to

Session: Image Registration [Return to Session]

A Practical Guide to Radiotherapy Image Registration Software Commissioning with TG-132 Common Dataset

Y Natsuaki, M Fan, T McDaniels, R Shaw, P Sansourekidou*, University of New Mexico, Albuquerque, NM


MO-H345-IePD-F1-5 (Monday, 7/11/2022) 3:45 PM - 4:15 PM [Eastern Time (GMT-4)]

Exhibit Hall | Forum 1

Purpose: The AAPM TG-132 report is an intuitive reference to the image registration (Rigid (RIR) and Deformative (DIR)) in radiotherapy and it provides recommendations for these clinical processes. TG-132 also provides the common data set for the commissioning of the different registration software. The current report provides a practical guide to the radiotherapy registration software commissioning using TG-132 common dataset.

Methods: The common data set consists of (i)geometric digital phantom, (ii)anatomical pelvis digital phantom, and (iii)two-phases (inhale-exhale) lung CT DICOM-RT sets. The digital phantoms are for the RIR analysis (reference and target image sets of various imaging modalities, deviations from known processed trans&rot values vs software registration), and the lung data set is for the DIR analysis (deformative metrics: Point-Matching (TPE), Surface-Matching (MDA, Housdorff), and Volume Similarity (DSC)). A commonly utilized RT registration software (Velocity AI, Varian) was commissioned by establishing baseline to the aforementioned metrics and by assessing whether these baseline meets the tolerance level set by TG-132.

Results: For the RIR, the common data set was able to establish the baseline. However, the baseline failed to meet the TG-132 tolerance in multi-modality registration (PET to CT) as well as the rotations consistency (reference-target sets reversed to see if this provides the same result). For the DIR, the common dataset failed due to incomplete (missing RT structures) and incompatible (non DICOM-RT standard) dataset provided. Instead, a different public data set (e.g. POPI lung DICOM-RT) was utilized to establish the baseline. The baseline failed to meet the TG-132 tolerance level for the deformative metrics.

Conclusion: The TG-132 common dataset requires immediate revision to meet the DICOM-RT standard and have complete structure set required for the DIR metrics. Furthermore, vendor-independent metrics analysis tool should be provided with the dataset to unify the metrics measures and re-establish the attainable tolerance level.


Registration, Deformation


IM/TH- Image Registration: Multi-modality registration

Contact Email