Deconvolution of Free-Breathing Y90 PET Images for the Calculation of Y90 Dose Distribution for Radioembolization of Liver Cancer
N Yu*, H Qu, P Xia, The Cleveland Clinic Foundation, Cleveland, OHSU-E-J-188 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: Recent progress in the use of post treatment Y90 PET (post-PET) imaging has made possible to quantitatively analyze the delivered tumor dose distribution for radioembolization of liver cancer with Y90 microspheres. Patient breathing may affect the calculated dose in clinically meaningful ways. The purpose of this work is to present a method to deconvolve post-PET to obtain dose and to validate the results.
Methods: Eleven patients were chosen for this retrospective study. Patient breathing patterns were obtained from fluoroscopy images acquired during treatment. The probability density functions that characterized the breathing patterns in the superior-posterior direction were developed and used to deconvolute the post-PET to generate the deconvoluted PET images (dec-PET), which minimizes the breath motion effect into dose calculation. The Van-Cittert algorithm was used for the deconvolution. To validate our method, we re-convolved the dec-PET with the same breathing kernel to generate post-PET-Ver. The delivered Y90 dose distributions were calculated with these three PET images using a previously published dose kernel. Selected isodose lines and dose volume histograms were analyzed. In particular, the dose to 95% volume encompassing 120 Gy isodose line from the post-PET for the three sets of PET images, D95,pet were compared.
Results: The motion amplitude in the superior/inferior direction for all the patients was 0.62 -2.0 cm. The DVHs for the dec-PET dose deviated significantly from the DVHs of the post-PET dose whereas the DVHs from the dec-PET -ver nearly overlapped with the DVHs of post-PET dose. For the 120 Gy dose line from the post-PET, D95,dec/D95,post was 0.89 ±0.08, D95,post2/D95,post was 0.98±0.01
Conclusion: We demonstrated that the deconvolution was reversible and we can use this method to remove breathing motion effect. Using this deconvolution algorithm, Y90 dose can be accurately calculated from deconvoluted post-PET with the breathing motion effect minimized.