HighlY Constrained backPRojection Research Articles

Noise Reduction and Image Quality Improvement of Low Dose and Ultra Low Dose Brain Perfusion CT by HYPR-LR Processing

PurposeTo evaluate image quality and signal characteristics of brain perfusion CT (BPCT) obtained by low-dose (LD) and ultra-low-dose (ULD) protocols with and without post-processing by highly constrained back-projection (HYPR)–local reconstruction (LR) technique.Methods and MaterialsSimultaneous BPCTs were acquired in 8 patients on a dual-source-CT by applying LD (80 kV,200 mAs,14×1.2 mm) on tube A and ULD (80 kV,30 mAs,14×1.2 mm) on tube B. Image data from both tubes was reconstructed with identical parameters and post-processed using the HYPR-LR. Correlation coefficients between mean and maximum (MAX) attenuation values within corresponding ROIs, area under attenuation curve (AUC), and signal to noise ratio (SNR) of brain parenchyma were assessed. Subjective image quality was assessed on a 5-point scale by two blinded observers (1:excellent, 5:non-diagnostic).ResultsRadiation dose of ULD was more than six times lower compared to LD. SNR was improved by HYPR: ULD vs. ULD+HYPR: 1.9±0.3 vs. 8.4±1.7, LD vs. LD+HYPR: 5.0±0.7 vs. 13.4±2.4 (both p<0.0001). There was a good correlation between the original datasets and the HYPR-LR post-processed datasets: r = 0.848 for ULD and ULD+HYPR and r = 0.933 for LD and LD+HYPR (p<0.0001 for both). The mean values of the HYPR-LR post-processed ULD dataset correlated better with the standard LD dataset (r = 0.672) than unprocessed ULD (r = 0.542), but both correlations were significant (p<0.0001). There was no significant difference in AUC or MAX. Image quality was rated excellent (1.3) in LD+HYPR and non-diagnostic (5.0) in ULD. LD and ULD+HYPR images had moderate image quality (3.3 and 2.7).ConclusionSNR and image quality of ULD-BPCT can be improved to a level similar to LD-BPCT when using HYPR-LR without distorting attenuation measurements. This can be used to substantially reduce radiation dose. Alternatively, LD images can be improved by HYPR-LR to higher diagnostic quality.

Myocardial Perfusion Magnetic Resonance Imaging Using Sliding-Window Conjugate-Gradient HYPR Methods in Canine With Stenotic Coronary Arteries

First-pass perfusion magnetic resonance imaging (MRI) is a promising technique for detecting ischemic heart disease. However, the diagnostic value of the method is limited by the low spatial coverage, resolution, signal-to-noise ratio (SNR), and cardiac motion-related image artifacts. A combination of sliding window and conjugate-gradient HighlY constrained back-PRojection reconstruction (SW-CG-HYPR) method has been proposed in healthy volunteer studies to reduce the acquisition window for each slice while maintaining the temporal resolution of 1 frame per heartbeat in myocardial perfusion MRI. This method allows for improved spatial coverage, resolution, and SNR. In this study, we use a controlled animal model to test whether the myocardial territory supplied by a stenotic coronary artery can be detected accurately by SW-CG-HYPR perfusion method under pharmacological stress. Results from 6 mongrel dogs (15-25 kg) studies demonstrate the feasibility of SW-CG-HYPR to detect regional perfusion defects. Using this method, the acquisition time per cardiac cycle was reduced by a factor of 4, and the spatial coverage was increased from 2 to 3 slices to 6 slices as compared with the conventional techniques including both turbo-Fast Low Angle Short (FLASH) and echoplanar imaging (EPI). The SNR of the healthy myocardium at peak enhancement with SW-CG-HYPR (12.68 ± 2.46) is significantly higher (P < 0.01) than the turbo-FLASH (8.65 ± 1.93) and EPI (5.48 ± 1.24). The spatial resolution of SW-CG-HYPR images is 1.2 × 1.2 × 8.0 mm, which is better than the turbo-FLASH (1.8 × 1.8 × 8.0 mm) and EPI (2.0 × 1.8 × 8.0 mm). Sliding-window CG-HYPR is a promising technique for myocardial perfusion MRI. This technique provides higher image quality with respect to significantly improved SNR and spatial resolution of the myocardial perfusion images, which might improve myocardial perfusion imaging in a clinical setting.

Dynamic PET Denoising with HYPR Processing

HighlY constrained backPRojection (HYPR) is a promising image-processing strategy with widespread application in time-resolved MRI that is also well suited for PET applications requiring time series data. The HYPR technique involves the creation of a composite image from the entire time series. The individual time frames then provide the basis for weighting matrices of the composite. The signal-to-noise ratio (SNR) of the individual time frames can be dramatically improved using the high SNR of the composite image. In this study, we introduced the modified HYPR algorithm (the HYPR method constraining the backprojections to local regions of interest [HYPR-LR]) for the processing of dynamic PET studies. We demonstrated the performance of HYPR-LR in phantom, small-animal, and human studies using qualitative, semiquantitative, and quantitative comparisons. The results demonstrate that significant improvements in SNR can be realized in the PET time series, particularly for voxel-based analysis, without sacrificing spatial resolution. HYPR-LR processing holds great potential in nuclear medicine imaging for all applications with low SNR in dynamic scans, including for the generation of voxel-based parametric images and visualization of rapid radiotracer uptake and distribution.

TU-E-201B-07: Image Noise and Radiation Dose Reduction in Spectral CT

Purpose: Spectral (multiple energy) CT can characterize material composition using the x‐ray spectral information. However, imagenoise and radiation dose is a concern as photons are distributed into multiple energy bins. Therefore, the intrinsic trade‐off between number of energy bins and photon numbers in each bin results in either too noisy images, or too high radiation dose. The purpose of this study is to break this trade‐off and to reduce noise and/or radiation dose in spectral CT.Method and Materials: A semi‐anthropomorphic thoracic phantom was scanned with multiple energies (80,100, 120 and 140 kVps) on a dual source CT scanner. Images were reconstructed at each beam‐energy separately using commercial software. The reconstructed images were processed wtih the Highly Constrained Back‐Projection (HYPR) algorithm using the averaged image of all 4 kVp scans as the composite image.CT numbers and noise were measured inside three circular ROIs representing calcium, water, and soft tissue. Noise and dose reduction using HYPR compared with commercial software was calculated. Results: The same CT numbers were obtained using both algorithms while approximately 50% of noise reduction was achieved using HYPR. Based upon the relationship between noise and radiation dose, this was equivalent to 4 times dose reduction given the same imagenoise. These results could be interpreted as noise reduction given the same radiation dose, dose reduction given the same imagenoise, or a combination of these two. Conclusion: We have presented a method to reduce imagenoise and/or radiation dose in spectral CT. Using HYPR, images at each individual energy bin have SNR similar as the composite image which uses all x‐ray photons. This breaks the tradeoff between number of energy bins and number of photons in each bin. Therefore, the intrinsic problem of excessive noise and/or radiation dose in spectral CT has been addressed.

CMR2009: 9.02: Increasing imaging time of 124I‐labeled NM404 using highly constrained backprojection (HYPR)

CMR2009: 9.02: Increasing imaging time of ¹²⁴I‐labeled NM404 using highly constrained backprojection (HYPR)

Renal Perfusion and Hemodynamics: Accurate in Vivo Determination at CT with a 10-Fold Decrease in Radiation Dose and HYPR Noise Reduction

To prospectively evaluate the accuracy of computed tomographic (CT) perfusion measurements of renal hemodynamics and function obtained by using images acquired with one-tenth the typical radiation dose and postprocessed with a highly constrained back-projection (HYPR)-local reconstruction (LR) noise-reduction technique. This study was approved by the institutional Animal Care and Use Committee. Two consecutive CT perfusion acquisitions were performed in 10 anesthetized pigs over 180 seconds by using routine (80 kV, 160 mAs) and one-tenth (80 kV, 16 mAs) dose levels. Images obtained with each acquisition were reconstructed with identical parameters, and the one-tenth dose images were also processed with a HYPR-LR algorithm. Attenuation changes in kidneys were determined as a function of time to form time-attenuation curves (TACs). Extended gamma-variate curve-fitting was performed, and regional perfusion, glomerular filtration rate, and renal blood flow were calculated. Image quality was evaluated (in 10 pigs), and the agreement for renal perfusion and function between the routine dose and the one-tenth dose HYPR-LR images was determined (for 20 kidneys) by using statistical methods. Statistical analysis was performed by using the paired t test, linear regression, and Bland-Altman analysis. TACs obtained with the one-tenth dose were similar to those obtained with the routine dose. Statistical analysis showed that there were no significant differences between the routine dose and the one-tenth dose acquisitions in renal perfusion and hemodynamic values and that there were slight but statistically significant differences in some values with the one-tenth dose HYPR-LR-processed acquisition. The image quality of the one-tenth dose acquisition was improved by using the HYPR-LR algorithm. Linear regression and Bland-Altman plots showed agreement between the images acquired by using the routine dose and those acquired by using the one-tenth dose with HYPR-LR processing. A 10-fold dose reduction at renal perfusion CT imaging can be achieved in vivo, without loss of accuracy. The image quality of the one-tenth dose images could be improved to be near that of the routine dose images by using the HYPR-LR noise-reduction algorithm. http://radiology.rsna.org/lookup/suppl/doi:10.1148/radiol.2531081677/-/DC1.

Radiation dose reduction in time-resolved CT angiography using highly constrained back projection reconstruction

Recently dynamic, time-resolved three-dimensional computed tomography angiography (CTA) has been introduced to the neurological imaging community. However, the radiation dose delivered to patients in time-resolved CTA protocol is a high and potential risk associated with the ionizing radiation dose. Thus, minimizing the radiation dose is highly desirable for time-resolved CTA. In order to reduce the radiation dose delivered during dynamic, contrast-enhanced CT applications, we introduce here the CT formulation of HighlY constrained back PRojection (HYPR) imaging. We explore the radiation dose reduction approaches of both acquiring a reduced number of projections for each image and lowering the tube current used during acquisition. We then apply HYPR image reconstruction to produce image sets at a reduced patient dose and with low image noise. Numerical phantom experiments and retrospective analysis of in vivo canine studies are used to assess the accuracy and quality of HYPR reduced dose image sets and validate our approach. Experimental results demonstrated that a factor of 6–8 times radiation dose reduction is possible when the HYPR algorithm is applied to time-resolved CTA exams.

Ultrashort TE spectroscopic imaging (UTESI) using complex highly‐constrained backprojection with local reconstruction (HYPR LR)

Recently, the highly-constrained backprojection (HYPR) and HYPR with local reconstruction (HYPR LR) methods have been introduced to reconstruct magnitude images from a series of highly undersampled data while preserving high spatial and temporal resolution and high signal-to-noise ratio (SNR) in applications with spatiotemporal correlations. However, these conventional HYPR algorithms are limited to the generation of magnitude images and, therefore, have limitations in their potential applications. In this work, the HYPR LR algorithm has been modified to extend the use of algorithms in the HYPR family to applications that require processing of complex data, such as MR chemical shift imaging (CSI) or spectroscopic imaging. The proposed method processes the magnitude information the same way as in original HYPR LR processing. In addition, it improves the phase images by subtracting the phase map of a synthesized composite image. The feasibility and efficiency of this algorithm has been demonstrated on CSI of cortical bone, Achilles tendon, and a healthy volunteer on a clinical 3T scanner.

CE‐MRA of the lower extremities using HYPR stack‐of‐stars

To investigate the properties of HYPR (HighlY constrained back PRojection) processing-the temporal fidelity and the improvements of spatial/temporal resolution-for contrast-enhanced MR angiography in a pilot study of the lower extremities in healthy volunteers. HYPR processing with a radial three-dimensional (3D) stack-of-stars acquisition was investigated for contrast-enhanced MR angiography of the lower extremities in 15 healthy volunteers. HYPR images were compared with control images acquired using a fast, multiphase, 2D Cartesian method to verify the temporal fidelity of HYPR. HYPR protocols were developed for achieving either a high frame update rate or a minimal slice thickness by adjusting the acquisition parameters. HYPR images were compared with images obtained using 3D TRICKS, a widely used protocol in dynamic 3D MRA. HYPR images showed good temporal agreement with 2D control images. In comparison with TRICKS, HYPR stack-of-stars demonstrated higher spatial and temporal resolution. High radial undersampling factors for each time frame were permitted, typically approximately 50 to 100 compared with fully sampled radial imaging. In this feasibility study, HYPR processing has been demonstrated to improve the spatial or temporal resolution in peripheral CE-MRA.

Undersampled radial MR acquisition and highly constrained back projection (HYPR) reconstruction: Potential medical imaging applications in the post‐Nyquist era

During the past several years there has been extensive study of alternative MR acquisition strategies such as spiral and radial. Vastly undersampled imaging with projections (VIPR) is a three-dimensional (3D) radial acquisition that provides acceptable images while violating the Nyquist theorem by factors of up to several hundred. For applications like magnetic resonance angiography (MRA), VIPR provides sparse data sets with incoherent artifacts that satisfy the requirements of emerging reconstruction approaches like iterative image norm minimization (compressed sensing) and highly constrained back projection (HYPR). All of these tools can be used in combination with parallel imaging to provide extremely high acceleration factors in MRI. In this review we do not attempt to do justice to the many exciting developments in the general field of constrained reconstruction but focus on preliminary results using VIPR and HYPR for non-Cartesian, Nyquist-violating MRI and the extension of HYPR processing to a broad range of medical imaging applications in which the acquisitions satisfy the Nyquist theorem but lack sufficient signal-to-noise ratio (SNR), leading to the possibility of radiation reduction, increased ultrasound resolution and field-of-view, and improved dynamic display of radiotracers.

Simulation of relative temporal resolution of time‐resolved MRA sequences

Time-resolved MRA sequences are typically characterized by the display frame rate. However, true temporal resolution should be defined in a manner analogous to spatial resolution; it is not the ability of a sequence to update rapidly but rather the ability to discern changes that occur within a small time that should characterize temporal resolution. For view-shared methods like Keyhole and time-resolved imaging of contrast kinetics (TRICKS), regions of k-space from multiple time frames are combined to form a single reconstructed time frame. This often causes the time needed to acquire all k-space data points to be significantly longer than the displayed frame time, resulting in a poor frequency response. Simulated here are the temporal impulse response and temporal frequency response (TFR) curves of three time-resolved MRA methods, including the recently introduced highly-constrained backprojection local reconstruction (HYPR LR) method. It is found that the HYPR LR reconstruction method exhibits a better TFR for a larger spectrum of temporal and spatial frequencies than the Keyhole and TRICKS methods.

Evaluation of temporal and spatial characteristics of 2D HYPR processing using simulations

Highly constrained back-projection (HYPR) is a data acquisition and reconstruction method that provides very rapid frame update rates and very high spatial resolution for a time series of images while maintaining a good signal-to-noise ratio and high image quality. In this study we used simulations to evaluate the temporal and spatial characteristics of images produced using the HYPR algorithm. The simulations demonstrate that spatial accuracy is well maintained in the images and the temporal changes in signal intensity are represented with high fidelity. The waveforms representing signal intensity as a function of time obtained from regions-of-interest placed in simulated objects track the true curves very well, with variations from the truth occurring only when objects with very different temporal behavior are very close to each other. However, even when objects with different temporal characteristics are touching, their influences on each other are small.

Improved waveform fidelity using local HYPR reconstruction (HYPR LR).

The recently introduced HYPR (HighlY constrained backPRojection) method allows reconstruction of serial images from highly undersampled data. In HYPR, individual timeframes are obtained via unfiltered backprojections of normalized sinograms using anatomical constraints provided by a composite image. Here we develop the idea of constraining the backprojected data further to a series of local regions of interest in order to decrease the corruption of local information by distant signals. HYPR LR (local reconstruction) permits the use of a longer temporal window in the formation of the composite image, resulting in increased signal-to-noise ratio and quantitative reconstruction accuracy. Unlike HYPR, the new HYPR LR method can be applied to images acquired with arbitrary k-space trajectories. It is suitable for a broad range of medical imaging applications involving serial changes in image sequence, offering exciting new opportunities in the future.

Iterative projection reconstruction of time‐resolved images using highly‐constrained back‐projection (HYPR)

Highly-constrained back-projection (HYPR) is a technique for the reconstruction of sparse, highly-undersampled time-resolved image data. A novel iterative HYPR (I-HYPR) algorithm is presented and validated in computer simulations. The reconstruction method is then applied to cerebral perfusion MRI simulated as a radial acquisition and contrast-enhanced angiography of the head to assess feasibility in accelerating acquisitions requiring high temporal resolution and accurate representation of contrast kinetics. The I-HYPR algorithm is shown to be more robust than standard HYPR in these applications in which the sparsity condition is not met or in which quantitative information is required. Specifically, iterative reconstruction of undersampled perfusion and contrast-enhanced angiography data improved accuracy of the representation of contrast kinetics and increased the temporal separation of arterial and venous contrast kinetics. The I-HYPR reconstruction may have important diagnostic applications in settings requiring high temporal resolution and quantitative signal dynamics. Because I-HYPR allows relaxation of the sparsity requirements for the composite frame, the iterative reconstruction can enable novel acquisition strategies that independently optimize the quality of the composite and temporal resolution of the dynamic frames.

Time‐resolved MR angiography with limited projections

A method for reconstruction of time-resolved MRI called highly-constrained backprojection (HYPR) has been developed. To evaluate the HYPR reconstruction in relation to data sparsity and temporal dynamics, computer simulations were performed, investigating signal modulations under different situations that reflect dynamic contrast-enhanced MR angiography (MRA). In vivo studies were also performed with gadolinium diethylenetriamine pentaacetic acid (Gd-DTPA) for abdominal MRA in a canine model to demonstrate the application of HYPR for three-dimensional (3D) time-resolved MRA. When contrast dynamics vary over space, large vessels (e.g., veins) tend to introduce signal interference to small vessels (e.g., arteries) in HYPR, particularly when the vessels are in close proximity. The enhancement of background tissue signals may also alter the arterial and venous temporal profiles in HYPR. However, the artifacts are manifest as intensity modulation rather than structural interference, and therefore have little impact on structural diagnosis. Increasing the number of projections per time point increases temporal blur while reducing corruption of temporal behavior from adjacent tissues. Uniformly interleaved acquisition order, such as the bit-reversed order, is important to reduce artifacts. With high signal-to-noise ratio (SNR) and limited artifacts, HYPR reconstruction has potential to greatly improve time-resolved MRA in clinical practice.

TH‐D‐330A‐09: Factor of Ten Dose Reduction in CT Perfusion Imaging

Purpose: To reduce by a factor of ten the patient dose in contrast enhanced CT Neuro Perfusion protocols while maintaining SNR and physiologic information relative to standard full dose CT Neuro Perfusion protocols. Method and Materials: Our group's newly developed undersampling algorithm for contrast enhanced MRI, HighlY constrained back Projection (HYPR), is applied to a contrast enhanced CT Neuro Perfusion protocol. By using temporally redundant information in the time series, a reduction in the number of required projection angles for each time frame is possible. The time series data is acquired using interleaved and equally spaced projections. A composite image consisting of projection information from either the entire time series or a limited number of time frames is first constructed by the standard filtered back projection technique. Next, a HYPR time frame is reconstructed by multiplying the composite image by backprojections for each projection angle in the time frame after normalizing them by the corresponding projection from the composite image. Results: HYPR time frames with similar SNR to fully sampled images and containing equivalent physiologic information for the creation of perfusion parameter maps have been obtained. The simulated reduction in the number of projection angles for the studies completed would result in a reduction in dose by a factor of 10. Computed parametric maps of CBV, CBF, and MTT are qualitatively similar to that of full dose maps, and arterial and venous input functions show excellent agreement with full dose input functions. Conclusion: The application of HighlY constrained back Projection (HYPR) to contrast enhanced CT Neuro Perfusion protocols allows a potential reduction by an order of magnitude to the delivered patient dose while maintaining SNR and physiologic information in the form of CBV, CBF, and MTT maps and arterial, venous and tissue input functions.