No full-text available
To read the full-text of this research,
you can request a copy directly from the authors.
SU-E-T-286: A Case Study of More Accurate Phase Re-Assignment for 4DCT Images
Medical Physics
Abstract
Purpose: We present a case study of an improved phase re?assignment approach for a challenging 4DCT image data set on a commercial system.Methods: The Varian RPM system was used to produce the respiratory traces for phase binning the raw 4DCT images acquired from a GE LightSpeedRT CT scanner. The RPM system uses a predictive approach to generate a *.vxp file containing details of the respiratory signal for correlation with the raw 4DCT images. The *.vxp file was read in using a program written in C++, and respiratory phases were re?assigned using an approach which retrospectively detects individual max and min of the respiratory signal. The re?binned file was used to reprocess the raw 4DCT data and generate a set of new phase binned images, which were compared with those from the original *.vxp files.Results: For one challenging clinical case, we demonstrate how the current predictive method can inadvertently assign a grossly incorrect phase after a highly irregular breathing cycle has occurred. The new image sets were seen to be significantly more accurate in representation of particular smooth internal structures. Seven additional patients were retrospectively studied using the same re?binning approach, with none of these patients demonstrating significant image quality improvement. This suggests that the vendor's currently employed phase assignment method produces reasonable results for the majority of scenarios, but may suffer significantly for unusual cases where extreme breathing anomalies occur. Conclusions: The vendor's standard phase assignment method which uses a predictive breathing cycle algorithm can be challenged in instances where highly anomalous breathing cycles of much longer period are encountered. This can introduce a rippling, or ?piano key?, effect on subsequent images. We presented a simple retrospective method for improving the accuracy of phase binning in these circumstances and demonstrated improved image fidelity on one challenging clinical case.
... In the case of irregular breathing, if one is to rewrite the VXP file with user determined max and min values instead of those set by RPM using its predictive method, generated 4DCT datasets can be seen to be much more accurate with smooth internal structures. 18 Breathing irregularity is certainly more difficult for RGSC to trace properly with phantom studies demonstrating measurable resulting distortions in target shape, position, and density. 9 However, these studies are either phantom based or only include data from a small number of patients. ...
... This has not been shown before for a large patient sample. VXP inaccuracy with marking of minima and maxima is in agreement with studies by Wang et al. 18 and Shi et al. 10 The uncertainty in frequency within the scan due to breathing rate variation is 37% with the average difference between marked and actual frequency having a standard deviation of almost 10%. This is somewhat surprising since the operator can review and edit the peaks and troughs before VXP export. ...
... In addition to persons not correctly reviewing the breathing trace, actual (calculated in this study) frequency variations were less than that was marked which agrees with the conclusion by Wang et al. who found the predictive filter of RPM to incorrectly assign phase. 18 Wang's study only featured a very small patient sample so this study improves upon those by testing the accuracy of the minima/maxima assignment more thoroughly through a larger patient sample, thus supplying better inference based on the breathing practices. This study is well positioned to supply uncertainties in breathing trace parameters for any subsequent patient. ...
4DCT simulation is a popular solution for radiotherapy simulation of lung cancer patients as it allows the clinician to gain an appreciation for target motion during the patient breathing cycle. Resultant binning of images and production of the 4DCT dataset relies heavily on the recorded breathing trace; but quality assurance is not routinely performed on these and there lacks any substantial recommendations thereof.
An application was created for Windows in C# that was able to analyze the VXP breathing trace files from Varian RPM/RGSC and quantify various metrics associated with the patient breathing cycle. This data was then used to consider errors in voluming of targets for several example cases in order to justify recommendations on quality assurance.
For 281 real patient breathing traces from 4DCT simulation of lung targets, notable differences were found between RGSC and application calculations of phase data. For any new patient without individualized QA, the average marked phase calculation (which is used for 4DCT reconstruction) is only accurate to within 19% of the actual phases. The error in BPM within the scan due to breathing rate variation is 37%. The uncertainty in amplitude due to breathing variation is 34% in the mean. Phase uncertainty leads to misbinning which we have shown can lead to missing 66% of the target for gated treatment. Variation in inhalation/exhalation level leads to voluming errors which, without individualized QA, can be assumed to be 11% (PTV is smaller than actual).
Without individualized quality assurance of patient breathing traces, large uncertainties have to be assumed for metrics of both phase and amplitude, leading to clinically significant uncertainties in treatment. It is recommended to perform individualized quality assurance as this provides the clinician with an accurate quantification of uncertainty for their patient.
ResearchGate has not been able to resolve any references for this publication.