Temporal Regularization for Robust Motion Compensation in Reduced Dose Cardiac-Gated Spect Images

Motion compensation is an effective approach for noise suppression and motion blur reduction in cardiac gated SPECT imaging. In this work, we investigate the potential benefit of using a deep learning network for motion compensation in a sequence of gated images throughout the cardiac cycle in the presence of large inter-subject variability and imaging degrading factors. We make use a set of clinical acquisitions from 130 subjects and quantify the motion compensation accuracy by variants of two known cascaded learning networks (namely VTN and VoxelMorph). The results in the experiments show that both networks can yield accurate compensation results in both standard dose and half dose studies. Specifically, VTN achieved a relative MSE of 0.0312 (full dose) and 0.0561 (half dose), compared to 0.0340 (full dose) and 0.0356 (half dose) for VoxelMorph. Both networks also outperformed the classical optical flow equation (OFE) method.

We investigated the performance of a post reconstruction dual respiratory and cardiac (R&C) motion compensation method for improved image quality of 4D cardiac gated small animal myocardial perfusion (MP) SPECT images. A normal mouse was injected with ∼8 mCi of Tc-99m sestamibi, anesthetized, fitted with ECG leads for cardiac gating signal acquisition, and placed on top of a pressure gauge bellow for respiratory motion measurements. A 2-hour list-mode dataset was acquired using a MILab small animal SPECT system fitted with a multi-pinhole collimator with 0.4 mm resolution in 5-minute sections. They were subsequently sorted for different acquisition times and reconstructed using a vendor provided OS-EM algorithm with simultaneous 6 respiratory and 8 cardiac equal-time gates over each motion cycle. Using a group-wise B-spline non-rigid image-based registration method, the deformation fields of the respiratory motion (respiratory motion) were estimated and applied to each cardiac phase for respiratory motion correction. Then, the respiratory motion compensated cardiac gated SPECT images were similarly used to estimate cardiac motion (cardiac motion) and later transformed to a reference frame and summed. Finally, the reference frame was inverse-transformed using the estimated cardiac motion to each of the 8 cardiac frames. The cardiac gated images with dual R&C motion compensation were compared to those without correction but with post-smoothing filter. The results showed the dual R&C motion compensation significantly reduced image noise level. At the same time, the image resolution was improved by 10% to 40% depending on the different acquisition times when compared with that obtained without motion compensation at the same image noise level. We conclude that dual R&C motion compensation provides significant reduction of noise level in 4D cardiac gated small animal MP SPECT images with minimum degradation of resolution. The improved image quality can be traded for reduction of acquisition time or radiation dose to the animal.

We present a post-reconstruction motion-compensated spatio-temporal filtering method for noise reduction in cardiac gated SPECT images. SPECT imaging suffers from low photon count due to radioactive dose limitations resulting in a high noise level in the reconstructed images. This is especially true in gated cardiac SPECT where the total number of counts is divided into a number of gates (time frames). Classical spatio-temporal filtering approaches, used in gated cardiac SPECT for noise reduction, do not accurately account for myocardium motion and brightening and therefore perform sub-optimally. The proposed post-reconstruction method consists of two steps: motion and brightening estimation and spatio-temporal motion-compensated filtering. In the first step we utilize a left ventricle model and a deformable mesh structure. The second step, which consists of motion-compensated spatio-temporal filtering, makes use of estimated myocardial motion to enable accurate smoothing. Additionally, the algorithm preserves myocardial brightening, a result of partial volume effect which is widely used as a diagnostic feature. The proposed method is evaluated quantitatively to assess noise reduction and the influence on estimated ejection fraction.

New Magnetic Resonance (MR) imaging applications include real time monitoring of temperature changes during cardiac radiofrequency ablations. MR-thermometry requires online robust motion compensation to cope with the complex motion of the heart resulting from respiratory activity and cardiac contraction (potentially in presence of arrhythmia), together with the presence of noise in MR images. We propose a method to adaptively and automatically tune parameters of motion compensation algorithms that use robustness function. The core of the method is the estimation of the probability density function (pdf) of the error for each pixel in a reference frame using the Rician noise pdf model in MRI. Then parameter map is derived from estimated pdf. The proposed method leads to better results than using a fixed control parameter of the robustness function, which should facilitate the use of such methods for clinical purpose.

In this work, we present a four-dimensional reconstruction technique for cardiac gated SPECT images using a content-adaptive deformable mesh model. Cardiac gated SPECT images are affected by a high level of noise. Noise reduction methods usually do not account for cardiac motion and therefore introduce motion blur-an artifact that can decrease diagnostic accuracy. Additionally, image reconstruction methods typically rely on uniform sampling and Cartesian griding for image representation. The proposed method utilizes a mesh representation of the images in order to utilize the benefits of content-adaptive nonuniform sampling. The mesh model allows for accurate representation of important regions while significantly compressing the data. The content-adaptive deformable mesh model is generated by combining nodes generated on the full torso using pre-reconstructed emission and attenuation images with nodes accurately sampled on the left ventricle. Ventricular nodes are further displaced according to cardiac motion using our previously introduced motion estimation technique. The resulting mesh structure is then used to perform iterative image reconstruction using a mesh-based maximum-likelihood expectation-maximization algorithm. Finally, motion-compensated post-reconstruction temporal filtering is applied in the mesh domain using the deformable mesh model. Reconstructed images as well as quantitative evaluation show that the proposed method offers improved image quality while reducing the data size.

4D image reconstruction can significantly improve the image quality in cardiac-gated imaging using single-photon emission computed tomography (SPECT). However, it is also associated with increased computation complexity which prevents it from being widely used in the clinic. In this study, we investigate a post-processing approach for cardiac-gated SPECT images by using a 3D residual convolutional neural network (CNN). In our formulation, the network is trained to produce images that are comparable to 4D reconstruction. In the experiments, we demonstrated this approach on a set of 197 clinical acquisitions. The results show that the proposed CNN approach can effectively suppress the noise level in the reconstructed myocardium. It also outperforms two alternative post-processing methods, including a non-local means (NLM) filter previously developed for gated SPECT images.

PD-0143: 4D cone-beam computed tomography combining total variation regularization and motion compensation

Motion-compensated reconstruction has shown to be effective for suppressing imaging noise and reducing motion blur in cardiac-gated SPECT imaging. In this work we investigate the feasibility of using a cascaded learning network for motion correction in a sequence of gated images acquired at different phases of the cardiac cycle, which are known to suffer from both imaging degradation and large-extent motion deformation. To be realistic, we make use of a set of consecutive clinical acquisitions (from 130 subjects) which includes both normal and abnormal subjects. The quantitative results in the experiments indicate that the learning network can yield consistently more accurate compensation results than the classical optical flow equation (OFE) method both on standard dose and half dose studies. In particular, the learning network achieved a relative MSE of 0.021 (full dose) and 0.037 (half dose), compared to 0.035 (full dose) and 0.053 (half dose) for OFE.

In this study, we consider a filtering method for image sequence degraded by additive Gaussian noise and/or impulse noise (i.e., mined noise). In general, for the image sequence filtering, motion compensation (MC) method is required in order to obtain good filtering performance both in the still and moving regions of an image sequence. Nevertheless a heavy computation load is imposed on MC method and MC tends to get mistaken motion vector owing to additive noise. To overcome above drawbacks of MC, we have proposed a Video-Data Dependent Weighted Average (Video-DDWA) filter for image sequence restoration degraded by additive Gaussian noise. The Video-DDWA filter whose weights are controlled by some local information contain a motion information as a motion detector is shown that the motion information method is more effective tool than MC method for image sequence restoration. However Video-DDWA filter is not proper for removing the mixed noise. Therefore, we replace weighted average filters and a motion information of the Video-DDWA with weighted median filters and a mixed noise robust motion information, respectively. We propose this filter as a Video-Data Dependent Weighted Median (Video-DDWM) filter for removing mixed noise from image sequence. Through some simulations, the Video-DDWM filter is proven to be more effective both the restoration results and computation time than the 3D-DDWM filter with impulse robust MC for removing mixed noise from image sequence.

Motion compensation is an effective approach for noise suppression and motion blur reduction in cardiac-gated SPECT imaging. In this work, we investigate the potential benefit of applying motion compensation with a deep learning (DL) network for assessment of perfusion defects in gated images. In addition to evaluating motion-compensation accuracy on clinical acquisitions, we also conduct a receiver-operating characteristic (ROC) study to assess the detection performance of perfusion detects when DL motion compensation is used to generate the perfusion images. For this task we use a clinical model observer on a set of hybrid studies generated from clinical acquisitions in which the perfusion defects are introduced as ground truth. The results in the experiments demonstrate that DL motion compensation can yield higher detection accuracy than conventional ungated studies.

In nuclear medicine, cardiac gated SPECT images are known to suffer from significantly increased noise owing to limited data counts. Consequently, spatial (and temporal) smoothing has been indispensable for suppressing the noise artifacts in SPECT reconstruction. However, recently we demonstrated that the benefit of spatial processing in motion-compensated reconstruction of gated SPECT (aka 4-D) could be outweighed by its adverse effects on the myocardium, which included degraded wall motion and perfusion defect detectability. In this work, we investigate whether we can alleviate these adverse effects by exploiting an alternative spatial smoothing prior in 4-D based on image total variation (TV). TV based prior is known to induce piecewise smoothing which can preserve edge features (such as boundaries of the heart wall) in reconstruction. However, it is not clear whether such a property would necessarily be beneficial for improving the accuracy of the myocardium in 4-D reconstruction. In particular, it is unknown whether it would adversely affect the detectability of perfusion defects that are small in size or low in contrast. In our evaluation study, we first use Monte Carlo simulated imaging with 4-D NURBS-based cardiac-torso (NCAT) phantom wherein the ground truth is known for quantitative comparison. We evaluated the accuracy of the reconstructed myocardium using a number of metrics, including regional and overall accuracy of the myocardium, accuracy of the phase activity curve (PAC) of the LV wall for wall motion, uniformity and spatial resolution of the LV wall, and detectability of perfusion defects using a channelized Hotelling observer (CHO). For lesion detection, we simulated perfusion defects with different sizes and contrast levels with the focus being on perfusion defects that are subtle. As a preliminary demonstration, we also tested on three sets of clinical acquisitions. From the quantitative results, it was demonstrated that TV smoothing could further reduce the error level in the myocardium in 4-D reconstruction along with motion-compensated temporal smoothing. In contrast to quadratic spatial smoothing, TV smoothing could reduce the noise level in the LV at a faster pace than the increase in the bias level, thereby achieving a net decrease in the error level. In particular, at the same noise level, TV smoothing could reduce the bias by about 30% compared to quadratic smoothing. Moreover, the CHO results indicate that TV could also improve the lesion detectability even when the lesion is small. The PAC results show that, at the same noise level, TV smoothing achieved lower temporal bias, which is also consistent with the improved spatial resolution of the LV in reconstruction. The improvement in blurring effects by TV was also observed in the clinical images.

In this article, the authors present a motion-compensated spatiotemporal processing algorithm to reduce noise in cardiac gated SPECT. Cardiac gated SPECT data are particularly noisy because the acquired photon data are divided among a number of time frames (gates). Classical spatial reconstruction and processing techniques offer noise reduction but they are usually applied on each frame separately and fail to utilize temporal correlation between frames. In this work, the authors present a motion-compensated spatiotemporal postreconstruction filter offering noise reduction while minimizing motion-blur artifacts. The proposed method can be used regardless of the type of image-reconstruction method (analytical or iterative). The between-frame volumetric myocardium motion is estimated using a deformable mesh model based on the model of the myocardial surfaces. The estimated motion is then used to perform spatiotemporal filtering along the motion trajectories. Both the motion-estimation and spatiotemporal filtering methods seek to maintain the wall brightening seen during cardiac contraction. Wall brightening is caused by the partial volume effect, which is usually viewed as an artifact; however, wall brightening is a useful signature in clinical practice because it allows the clinician to visualize wall thickening. Therefore, the authors seek in their method to preserve the brightening effect. The authors find that the proposed method offers better noise reduction than several existing methods as quantitatively evaluated by signal-to-noise ratio, bias-variance plots, and ejection fraction analysis as well as on tested clinical data. The proposed method mitigates for noise in cardiac gated SPECT images using a postreconstruction motion-compensated filtering approach. Visual as well as quantitative evaluation show considerable improvement in image quality.

The movement of cilia in the fallopian tubes (FTs) facilitates important processes involved in fertility, and abnormalities in cilia function are linked with diseases including endometriosis and pelvic inflammatory disease. For the first time, we demonstrate the use of optical coherence tomography (OCT) to create depth-resolved mapping of motile cilia locations and quantify cilia beat frequency (CBF) in human FT samples ex vivo. Segments of the FT ampulla were acquired from five patients following salpingectomy under an IRB approved protocol. The samples were longitudinally opened to expose the luminal surface for imaging. A sequence of at least 500 OCT images were acquired at 5-10 locations on each sample. To define the location of the motile cilia in the images, pixel-wise Fast Fourier Transform (FFT) analysis of intensity fluctuations with a sliding temporal window was performed on each image sequence. The frequencies corresponding to the physiological range of CBF (2-10 Hz) were selected for mapping, while the part of the FFT spectrum at higher frequencies (> 23 Hz) was used to define the noise threshold. The frequency with the highest FFT amplitude for each supra-threshold pixel was considered the CBF for this pixel and used to create a color-coded CBF map. The CBF map was overlaid with the OCT intensity image sequences to reveal cilia locations. Frequency histograms from the sliding window were examined to assess temporal consistency of the mapping and evaluate movement artifacts. OCT image sequences clearly showed the structure of FT plicae. The ciliated epithelium was obvious as a "shimmering" (rapidly changing intensity) layer atop plicae. Colored pixels on CBF maps visually aligned to these shimmering regions. Frequency histograms revealed that the image sequence peak CBF could be robustly determined, even in the presence of outliers attributable to table vibrations or bulk sample movement. OCT can provide depth-resolved maps of CBF in human ex vivo FT tissue. Potentially, this technique can aid in understanding cilia dynamics in the normal human FT over the menstrual cycle and across age, as well as in diseases that affect the FTs. Future work will be directed toward in vivo implementation including miniaturization and robust motion compensation.

PurposeSimultaneous multi-slice (SMS) balanced steady-state free precession (bSSFP) acquisition and iterative reconstruction can provide high spatial resolution and coverage for cardiac magnetic resonance (CMR) perfusion. However, respiratory motion remains a challenge for iterative reconstruction techniques employing temporal regularisation. The aim of this study is to evaluate an iterative reconstruction with integrated motion compensation for SMS-bSSFP first-pass myocardial stress perfusion in the presence of respiratory motion. MethodsThirty-one patients with suspected coronary artery disease were prospectively recruited and imaged at 1.5 T. A SMS-bSSFP prototype myocardial perfusion sequence was acquired at stress in all patients. All datasets were reconstructed using an iterative reconstruction with temporal regularisation, once with and once without motion compensation (MC and NMC, respectively). Three readers scored each dataset in terms of: image quality (1:poor; 4:excellent), motion/blurring (1:severe motion/blurring; 3:no motion/blurring), and diagnostic confidence (1:poor confidence; 3:high confidence). Quantitative assessment of sharpness was performed. The number of uncorrupted first-pass dynamics was measured on the NMC datasets to classify patients into ‘suboptimal breath-hold (BH)’ and ‘good BH’ groups. ResultsCompared across all cases, MC performed better than NMC in terms of image quality (3.5 ± 0.5 vs. 3.0 ± 0.8, P = 0.002), motion/blurring (2.9 ± 0.1 vs. 2.2 ± 0.8, P < 0.001), diagnostic confidence (2.9 ± 0.1 vs. 2.3 ± 0.7, P < 0.001) and sharpness index (0.34 ± 0.05 vs. 0.31 ± 0.06, P < 0.001). Fourteen patients with a suboptimal BH were identified. For the suboptimal BH group, MC performed better than NMC in terms of image quality (3.8 ± 0.4 vs. 2.6 ± 0.8, P < 0.001), motion/blurring (3.0 ± 0.1 vs. 1.6 ± 0.7, P < 0.001), diagnostic confidence (3.0 ± 0.1 vs. 1.9 ± 0.7, P < 0.001) and sharpness index (0.34 ± 0.05 vs. 0.30 ± 0.06, P = 0.004). For the good BH group, sharpness index was higher for MC than NMC (0.34 ± 0.06 vs 0.31 ± 0.07, P = 0.03), while there were no significant differences observed for the other three metrics assessed (P > 0.11). There were no significant differences between suboptimal BH MC and good BH MC for any of the reported metrics (P > 0.06). ConclusionsIntegrated motion compensation significantly reduces motion/blurring and improves image quality, diagnostic confidence and sharpness index of SMS-bSSFP perfusion with iterative reconstruction in the presence of motion.

Cardiac contraction motion compensation in gated myocardial perfusion SPECT: A comparative study