Abs ac
Re e ences
A mo ion-co ec ed econs uc ion o highly unde sampled pe usion CMR
INTRODUCTION: Fi s -pass pe usion ca diac MR (pCMR) acili a es he non-in asi e diagnosis o co ona y a e y
disease by acqui ing dynamic images du ing he apid passage o a con as bolus h ough he hea 1. A ade-o exis s
be ween hea co e age and empo al/spa ial esolu ion. To alle ia e his sho coming, unde sampled acquisi ions
ollowed by cons ained econs uc ion me hods ha e been p oposed. Also, he e is a g owing p e e ence o acqui ing
images du ing ee-b ea hing, as i o e s inc eased eliabili y and enhanced pa ien com o . In his con ex , he
inco po a ion o mo ion co ec ion (MoCo) app oaches becomes essen ial o bo h ee-b ea hing acquisi ions and o
co ec ing esidual mo ion due o impe ec b ea h-holding. Howe e , MoCo ends o deg ade as accele a ion inc eases.
This wo k p oposes a pipeline o a mo ion-co ec ed econs uc ion o highly unde sampled pCMR acquisi ions. A no el
igid MoCo s ep, o mula ed exclusi ely in k-space, is p oposed.
METHODS: Da a Th ee pa ien s unde wen a REST pCMR acquisi ion on a Philips 3T Achie a scanne . Acquisi ions
pa ame e s we e: in-plane esolu ion = 1.6 × 1.6 mm2, FOV = 320×320 mm2, slice hickness = 10 mm, 32 coils, and
scan ime = 1 min. A adial sampling was used wi h 10 spokes pe ame (accele a ion ac o ~20x); howe e , p io o
any o he s ep, he k-space was g idded on o a Ca esian g id. Sensi i i y maps we e es ima ed using ESPIRIT2.
P oposed app oach: The main s eps o he mo ion-co ec ed econs uc ion pipeline a e: 1) Rigid MoCo in k-space
wi h K-CC-MoCo3; his me hod pe o ms a pai wise egis a ion o mula ed exclusi ely in k-space wi h he no malized
c oss co ela ion as he egis a ion me ic de ined be ween a syn he ic e e ence and he k-spaces o each ame. To
ocus he minimiza ion on he hea egion, a ROVi coil-comp ession app oach4 is employed. 2) Low Rank + Spa se
(L+S) econs uc ion5 o he igidly mo ion-co ec ed k-space. 3) Pai -wise non- igid image-based MoCo o he
econs uc ed dynamic images o co ec o sligh esidual mo ion6.
RESULTS & DISCUSSION: Figu e 1 shows he esul s o he mo ion-co ec ed econs uc ions in a ep esen a i e
pa ien in which bulk mo ion caused by impe ec b ea h-holding is obse able. The esul s show ha he applica ion o
MoCo app oaches no ably educes blu ing when all he ames o he dynamic images a e combined. Also, we should
highligh ha he igid MoCo s ep in k-space becomes c ucial o an accu a e o e all MoCo; speci ically, he non- igid
mo ion-co ec ed images ha we e econs uc ed wi hou his i s s ep (Fig 1.C) p esen a poo e co ec ion compa ed
o he images ob ained wi h he h ee-s eps p oposed app oach (Fig 1.D).
ACKNOWLEDGMENTS Wo k suppo ed by “la Caixa” Founda ion and FCT wi h p ojec [LCF/PR/HR22/00533 ].
Elisa Moya-Sáez1*, Rosa-Ma ía Menchón-La a1,2, Ri a G. Nunes3, Te esa M. Co eia4,5, Ca los Albe ola-López1
1LPI, Uni e si y o Valladolid, Spain; 2Uni e sidad Poli écnica de Ca agena, Spain; 3ISR–Lisboa and Depa men o
Bioenginee ing, IST, Uni e si y o Lisbon, Po ugal; 4Cen e o Ma ine Sciences-CCMAR, Fa o, Po ugal; 5School o Biomedical
Enginee ing and Imaging Sciences, King's College London, UK *el[email p o ec ed]
Figu e 1 – Sum along ames o he econs uc ed dynamic images o he pipelines A) wi hou MoCo, B) wi h only igid k-space-
based MoCo. C) wi h only non- igid image-based MoCo, and D) wi h bo h igid (in k-space) and non- igid (in image-space) MoCo.
Also, in ensi y p o iles o E) y- di ec ion (yellow line, oo -head) and F) x- di ec ion (g een line, igh -le ).
1. Fai M.J., e al. J Ca dio asc Magn Reson. 2015;17(1),68; 2. Uecke M., e al. Magn Reson Med 2021;71(3),990-1001; 3. Moya-
Sáez E., e al. ISMRM 2025. 4. Kim D., e al. Magn Reson Med 2021; 86(1),197-212; 5. O azo, R., e al; Magn Reson Med
2015;73(3),1125-1136; 6. Menchón-La a R.M., e al. Signal P ocessing 2023;202,108771.
