Accuracy quantification of a deformable image registration tool applied in a clinical setting

Abstract

The purpose of this study was to test the accuracy of a commercially available deformable image registration tool in a clinical situation. In addition, to demonstrate a method to evaluate the resulting transformation of such a tool to a reference defined by multiple experts. For 16 patients (seven head and neck, four thoracic, five abdominal), 30‐50 anatomical landmarks were defined on recognizable spots of a planning CT and a corresponding fraction CT. A commercially available deformable image registration tool, Velocity AI, was used to align all fraction CTs with the respective planning CTs. The registration accuracy was quantified by means of the target registration error in respect to expert‐defined landmarks, considering the interobserver variation of five observers. The interobserver uncertainty of the landmark definition in our data sets is found to be 1.2±1.1thinmathspacemm. In general the deformable image registration tool decreases the extent of observable misalignments from 4‐8 mm to 1‐4 mm for nearly 50% of the landmarks (to 77% in sum). Only small differences are observed in the alignment quality of scans with different tumor location. Smallest residual deviations were achieved in scans of the head and neck region (79%,≤thinmathspace4thinmathspacemm) and the thoracic cases (79%,≤thinmathspace4thinmathspacemm), followed by the abdominal cases (59%,≤thinmathspace4thinmathspacemm). No difference is observed in the alignment quality of different tissue types (bony vs. soft tissue). The investigated commercially available deformable image registration tool is capable of reducing a mean target registration error to a level that is clinically acceptable for the evaluation of retreatment plans and replanning in case of gross tumor change during treatment. Yet, since the alignment quality needs to be improved further, the individual result of the deformable image registration tool has still to be judged by the physician prior to application.PACS numbers: 87.57.nj, 87.57.N‐, 87.55.‐x

The suitability of common metrics for assessing parotid and larynx autosegmentation accuracy

“…From this, a DTA histogram can be produced. Several different metrics can be derived from this DTA histogram and some of the most commonly reported include the mean‐ and maximum‐DTA 14 , 20 and the 95%‐Hausdorff distance (95%‐HD), 21 , 22 which is defined as the 95th percentile of the DTA histogram. Although these metrics provide a measure of the distance between two structures, they too can be difficult to translate into clinical relevance (19) …”

Performance variations among clinically available deformable image registration tools in adaptive radiotherapy — how should we evaluate and interpret the result?

“…As such, for the first question of “how to evaluate,” assessment of both capacities, contour propagating and dose tracking, should be included. The current techniques for DIR evaluation generally fell into three categories: contour comparison, 21 , 22 , 23 , 24 , 25 landmark tracking, 5 , 23 and voxel‐based analysis 7 , 8 , 9 , 10 , 12 , 13 . Both the contour‐ and landmark‐based analyses can yield a skewed view of the overall registration accuracy, as they only test spatial accuracy in limited regions.…”

“…The ground‐truth deformations are typically derived from either physical image/dosimeter phantoms 15 , 27 , 28 , 29 , 30 or synthetic computerized data 5 , 26 . Of note, it is unrealistic to design physical phantoms to simulate every clinical scenario.…”

“…Previous studies to estimate the inherent accuracy of DIR have typically been performed only by comparing locations of anatomic landmarks or contours identified by physicians 5 , 6 , 7 , 8 . A recent work published by Hoffmann et al (5) evaluated the feasibility of a single commercial tool (Velocity AI, Velocity Medical Solutions, Atlanta, GA) in head and neck, thoracic and abdominal cases. A total of 30 to 50 landmarks were placed by physicians for a whole 3D volume for each case.…”

“…Their findings concluded that in spite of all algorithms resulting in comparable geometric matching, the choice of DIR implementation leads to uncertainties in dose warping for head and neck cancer. As several commercialized DIR tools are already available for clinical use, 5 , 12 , 13 , 14 there is an emerging and urgent need to investigate the reliability of these clinical tools in multiple applications in daily clinical routine. Previously, we have compared deformed vector fields of two DIR algorithms on both physical phantoms and virtual patient images (13) .…”