Cardiac motion estimation from medical images: a regularisation framework applied on pairwise image registration displacement fields

Abstract

Accurate cardiac motion estimation from medical images such as ultrasound is important for clinical evaluation. We present a novel regularisation layer for cardiac motion estimation that will be applied after image registration and demonstrate its effectiveness. The regularisation utilises a spatio-temporal model of motion, b-splines of Fourier, to fit to displacement fields from pairwise image registration. In the process, it enforces spatial and temporal smoothness and consistency, cyclic nature of cardiac motion, and better adherence to the stroke volume of the heart. Flexibility is further given for inclusion of any set of registration displacement fields. The approach gave high accuracy. When applied to human adult Ultrasound data from a Cardiac Motion Analysis Challenge (CMAC), the proposed method is found to have 10% lower tracking error over CMAC participants. Satisfactory cardiac motion estimation is also demonstrated on other data sets, including human fetal echocardiography, chick embryonic heart ultrasound images, and zebrafish embryonic microscope images, with the average Dice coefficient between estimation motion and manual segmentation at 0.82–0.87. The approach of performing regularisation as an add-on layer after the completion of image registration is thus a viable option for cardiac motion estimation that can still have good accuracy. Since motion estimation algorithms are complex, dividing up regularisation and registration can simplify the process and provide flexibility. Further, owing to a large variety of existing registration algorithms, such an approach that is usable on any algorithm may be useful.

Cardiac forces regulate zebrafish heart valve delamination by modulating Nfat signaling

“…A previous study approximating the heart as a linear tube has shown that the absence of red blood cells leads to a decrease in maximal wall shear stress of 2 to 3 times at 48 hpf [ 13 ]. Using a recently developed motion estimation algorithm [ 40 ], we were able to extract the position and motion of the cardiac wall (and blood cells in wild-type controls) in our movies of the beating heart and model wall shear stress in wild-type and gata1 mutants at 65 hpf. At this stage, when heartbeat is faster and stronger, we found that maximal wall shear stress at the AVC is approximately 3 to 4 times lower in gata1 mutants ( S10 Fig ).…”

“…For the gata1 mutants, XYTZ images of the heart were captured at a frame rate of 100 frames per second for 80 frames, and the beating heart was realigned using BeatSync2.0 [ 79 ]. We track the ventricle volumes across time with a motion estimation algorithm [ 40 ] and calculate flow rates via the rate of volume changes of the ventricle. Subsequently, AVC shear rate is estimated with the assumption of a parabolic flow profile across a maximum inscribed circle in the cross section of the AVC, thus providing conservatively maximum wall shear rate values in the gata1 mutants.…”

“…A previous study approximating the heart as a linear tube has shown that the absence of red blood cells leads to a decrease in maximal wall shear stress of 2 to 3 times at 48 hpf [13]. Using a recently developed motion estimation algorithm [40], we were able to extract the position and motion of the cardiac wall (and blood cells in wild-type controls) in our movies of the beating heart and model wall shear stress in wild-type and gata1 mutants at 65 hpf. At this stage, when heartbeat is faster and stronger, we found that maximal wall shear stress at the AVC is approximately 3 to 4 times lower in gata1 mutants (S10 Fig) . Gata1 mutant hearts appear normal at 48 hpf and, unlike gata1 morphants, which have increased numbers of AVC endocardial cells [41], we found no differences in the number of AVC endocardial cells in gata1 mutants compared to controls (S11A and S11B Fig) . Searching for possible delamination defects, we found that we could screen out the majority of gata1 mutants with early superior AV valve defects by removing embryos with pericardial edema at 65 hpf (S11C-S11H Fig) . Imaging the beating heart of gata1 mutants in the Tg(fli1a:gal4ff; UAS:Kaede) background at 80 hpf, we found that only 17% of gata1 mutant superior AV valves manage to delaminate by 80 hpf compared to 92% of gata1 control superior AV valves (S12A and S12B Fig, S7 and S8 Movies).…”

Spatiotemporal Bayesian Regularization for Cardiac Strain Imaging: Simulation and In Vivo Results

“…Please note that in case of patients (e.g. arrhythmia) heart movements might be irregular mostly impacting cyclic periodicity [72] however smoothness assumption in a small local neighborhood should still hold based on findings from feasibility studies reported by Elen et al [10]. Moreover, in the preferred spatiotemporal regularization scheme (STBR-2), the spatial neighborhood size is always larger than temporal neighborhood size in our implementation thus ensuring that the algorithm is not biased by temporal information.…”

Fluid mechanics of the left atrial ligation chick embryonic model of hypoplastic left heart syndrome

“…vmtk. org), and then a validated cardiac motion estimation algorithm (Wiputra et al 2020) was used to propagate the 3D reconstruction to all time points. Similar to previous work (Ho et al 2019b), the threshold values for segmentation were iteratively determined so that volumetric calculations from motion tracking would show minimum retrograde flow across the atrioventricular junction, which was a feature demonstrated in the literature for both normal and LAL embryos (Tobita and Keller 2000), and which we could verify with Doppler measurements in some embryos.…”