Practical quantification of image registration accuracy following the AAPM TG‐132 report framework

Abstract

The AAPM TG 132 Report enumerates important steps for validation of the medical image registration process. While the Report outlines the general goals and criteria for the tests, specific implementation may be obscure to the wider clinical audience. We endeavored to provide a detailed step‐by‐step description of the quantitative tests’ execution, applied as an example to a commercial software package (Mirada Medical, Oxford, UK), while striving for simplicity and utilization of readily available software. We demonstrated how the rigid registration data could be easily extracted from the DICOM registration object and used, following some simple matrix math, to quantify accuracy of rigid translations and rotations. The options for validating deformable image registration (DIR) were enumerated, and it was shown that the most practically viable ones are comparison of propagated internal landmark points on the published datasets, or of segmented contours that can be generated locally. The multimodal rigid registration in our example did not always result in the desired registration error below ½ voxel size, but was considered acceptable with the maximum errors under 1.3 mm and 1°. The DIR target registration errors in the thorax based on internal landmarks were far in excess of the Report recommendations of 2 mm average and 5 mm maximum. On the other hand, evaluation of the DIR major organs’ contours propagation demonstrated good agreement for lung and abdomen (Dice Similarity Coefficients, DSC, averaged over all cases and structures of 0.92 ± 0.05 and 0.91 ± 0.06, respectively), and fair agreement for Head and Neck (average DSC = 0.73 ± 0.14). The average for head and neck is reduced by small volume structures such as pharyngeal constrictor muscles. Even these relatively simple tests show that commercial registration algorithms cannot be automatically assumed sufficiently accurate for all applications. Formalized task‐specific accuracy quantification should be expected from the vendors.

Evaluation of a deformable image registration quality assurance tool for head and neck cancer patients

“…It has been highlighted that issues with outliers do present when evaluating HD; thus, perhaps HD should be eliminated for future DIR analysis. 12 For DSC, the good values obtained were comparable to those in a study by Latifi et al 4 where the mandible obtained the highest DSC (0.88) followed by the right parotid (0.82) and left parotid (0.80) respectively. This is however different to findings by Varadhan et al 12 as the mandible obtained the lowest DSC (0.63) out of the three structures in their study.…”

“…Deformable image registration (DIR) is increasingly being implemented to correct for these differences, by transforming individual voxels by differing magnitudes from one image to align with those in another. 4 DIR is of particular value in the head and neck region as differences between the diagnostic CT and planning CT often result from changes in neck flexion, shoulder position, inclusion of immobilisation devices, changes in patient size and tumour growth. 5,6 Although DIR can correct for these differences, this method is more difficult to visually assess for accuracy after registration than RIR, creating challenges around communicating registration accuracy efficiently.…”

“…Until recently, rigid image registration (RIR) has been the most widely implemented technique in the clinical setting, which has limitations when registering images with positional and anatomical differences. Deformable image registration (DIR) is increasingly being implemented to correct for these differences, by transforming individual voxels by differing magnitudes from one image to align with those in another 4 . DIR is of particular value in the head and neck region as differences between the diagnostic CT and planning CT often result from changes in neck flexion, shoulder position, inclusion of immobilisation devices, changes in patient size and tumour growth 5,6 .…”

An international survey on the clinical use of rigid and deformable image registration in radiotherapy

“…RTs and Dosimetrists are strongly involved in imaging before IR, have a key role in RIR, and a growing role in DIR. Survey data indicates that multidisciplinary peer review and audits would facilitate the safe operations of DIR, particularly when trained staff have responsibilities for managing DIR accuracy 4,5,30 supported by departmental safety, risk, and incident reporting processes.…”

Quantifying the accuracy of deformable image registration for cone‐beam computed tomography with a physical phantom

“…Moreover, some tests are not accompanied by readily available software to implement them. The literature on validation of image registration software, particularly DIR, is still in its infancy . Our physical phantom does not have inserts that can simulate all clinical scenerios, but it does provide a means for basic, end‐to‐end testing of the entire imaging system.…”