Michael Manhart

Dynamic Iterative Reconstruction for Interventional 4-D C-Arm CT Perfusion Imaging

Tissue perfusion measurement using C-arm angiography systems capable of CT-like imaging (C-arm CT) is a novel technique with potentially high benefit for catheter guided treatment of stroke in the interventional suite. However, perfusion C-arm CT (PCCT) is challenging: the slow C-arm rotation speed only allows measuring samples of contrast time attenuation curves (TACs) every 5-6 s if reconstruction algorithms for static data are used. Furthermore, the peak values of the TACs in brain tissue typically lie in a range of 5-30 HU, thus perfusion imaging is very sensitive to noise. We present a dynamic, iterative reconstruction (DIR) approach to reconstruct TACs described by a weighted sum of basis functions. To reduce noise, a regularization technique based on joint bilateral filtering (JBF) is introduced. We evaluated the algorithm with a digital dynamic brain phantom and with data from six canine stroke models. With our dynamic approach, we achieve an average Pearson correlation (PC) of the PCCT canine blood flow maps to co-registered perfusion CT maps of 0.73. This PC is just as high as the PC achieved in a recent PCCT study, which required repeated injections and acquisitions.

Denoising and artefact reduction in dynamic flat detector CT perfusion imaging using high speed acquisition: first experimental and clinical results.

Flat detector CT perfusion (FD-CTP) is a novel technique using C-arm angiography systems for interventional dynamic tissue perfusion measurement with high potential benefits for catheter-guided treatment of stroke. However, FD-CTP is challenging since C-arms rotate slower than conventional CT systems. Furthermore, noise and artefacts affect the measurement of contrast agent flow in tissue. Recent robotic C-arms are able to use high speed protocols (HSP), which allow sampling of the contrast agent flow with improved temporal resolution. However, low angular sampling of projection images leads to streak artefacts, which are translated to the perfusion maps. We recently introduced the FDK-JBF denoising technique based on Feldkamp (FDK) reconstruction followed by joint bilateral filtering (JBF). As this edge-preserving noise reduction preserves streak artefacts, an empirical streak reduction (SR) technique is presented in this work. The SR method exploits spatial and temporal information in the form of total variation and time-curve analysis to detect and remove streaks. The novel approach is evaluated in a numerical brain phantom and a patient study. An improved noise and artefact reduction compared to existing post-processing methods and faster computation speed compared to an algebraic reconstruction method are achieved.

C-arm CT Perfusion Imaging in the Interventional Suite

Diagnostic perfusion imaging using CT or MRI has been available for several years. One of its applications is acute stroke diagnosis. Interventional perfusion imaging using C-arm CT is a novel field of research. It could provide perfusion information during catheter-guided stroke treatment in order to optimize patient management. In this survey article, the clinical and technical background of this topic are described and first results from in-vivo measurements are presented.

A realistic digital phantom for perfusion C-arm CT based on MRI data

CTP is an important imaging modality for diagnosis of ischemic stroke, which is computed from of a series of consecutive CT-scans during the injection of contrast agent. Contrast flow at any point in space can be tracked as minor changes in intensity over a period of about 40 seconds to one minute, represented as a time-attenuation curve (TAC) for every voxel. This work presents an isotropic, dense, physiologically realistic and dynamic brain phantom for CT perfusion. The phantom is based on MRI scans of a volunteer and is freely available for download.

Iterative denoising algorithms for perfusion C-arm CT with a rapid scanning protocol

Tissue perfusion measurement using C-arm angiography systems capable of CT-like imaging (C-arm CT) is a novel technique with potentially high benefit for catheter-guided treatment of stroke in the interventional suite. New rapid scanning protocols with increased C-arm rotation speed enable fast acquisitions of C-arm CT volumes and allow for sampling the contrast flow with improved temporal resolution. However, the peak contrast attenuation values of brain tissue lie typically in a range of 5-30 HU. Thus perfusion imaging is very sensitive to noise. In this work we compare different denoising algorithms based on the algebraic reconstruction technique (ART) and introduce a novel denoising technique, which requires only iterative filtering in volume space and is computationally much more attractive. Our evaluation using a realistic digital brain phantom shows that all methods improve the perfusion maps perceptibly compared to Feldkamp-type (FDK) reconstruction. The volume-based technique performs similarly to the ART-based methods: the Pearson correlation of reference and reconstructed blood flow maps increases from 0.61 for the FDK method to 0.81 for the best ART method and to 0.79 for the volume-based method. Furthermore results from a canine stroke model study are shown.

Fast dynamic reconstruction algorithm with joint bilateral filtering for perfusion C-arm CT

Tissue perfusion measurement using C-arm angiography systems is a novel technique with potential high benefit for catheter-guided treatment of stroke in the interventional suite. However, perfusion C-arm CT (PCCT) is challenging: the slow C-arm rotation speed only allows measuring samples of contrast time attenuation curves (TACs) every 5 - 6 s if reconstruction algorithms for static data are used. Furthermore, the peaks of the tissue TACs typically lie in a range of 5 - 30 HU, thus perfusion imaging is very sensitive to noise. We present a dynamic, iterative reconstruction (DIR) approach to reconstruct TACs described by a weighted sum of linear spline functions. The optimization problem is solved using an appropriate initialization and a Landweber-based optimization strategy with a modified backprojection step. To reduce noise a novel regularization technique based on joint bilateral filtering (JBF) is introduced. The algorithm is evaluated using simulation data created with a dynamic cylindrical phantom, a realistic digital brain phantom and real measured data from an animal study with a canine stroke model. Results indicate that the DIR algorithm qualitatively and quantitatively improves reconstructed TACs and perfusion maps compared to classical Feldkamp (FDK) reconstruction. For the brain phantom study the Pearson correlation (PC) of the reconstructed cerebral blood flow (CBF) maps to the ground truth increased from 0.82 (FDK) to 0.87 (DIR). For the canine study the PC of the CBF maps to coregistered perfusion CT maps increased from 0.61 (FDK) to 0.73 (DIR).

Guided noise reduction with streak removal for high speed flat detector CT perfusion

Tissue perfusion measurement using C-arm angiography systems capable of CT-like imaging (flat detector CT (FD-CT)) is a novel technique with high potential benefit for catheter-guided treatment of stroke in the interventional suite. New high speed protocols (HSP) with increased C-arm rotation speed enable fast acquisitions of FD-CT volumes and allow for sampling the contrast flow with improved temporal resolution. However, the peak contrast attenuation values of brain tissue typically lie in a range of 5-30 HU. Thus perfusion imaging is very sensitive to noise. Recently we introduced the FDK-JBF denoising technique based on Feldkamp (FDK) reconstruction followed by denoising in volume space using joint bilateral filtering (JBF). In the evaluation FDK-JBF achieved comparable results to algebraic techniques, but is computationally less costly. Yet the angular sampling of the projection data in the HSP is coarse, which leads to streak artifacts in the reconstructed volumes. Mask volumes are subtracted from the contrast agent enhanced (bolus) volumes and the streak artifacts are subtracted out if the patient does not move during the acquisition. However, in case of motion the streak artifacts will not be identical in the mask and bolus volumes. We show that these streaks can lead to severe artifacts in the perfusion maps and describe a novel technique for streak removal (SR), which is based on streak detection by using time-contrast curve analysis.We evaluated the FDK-SR-JBF algorithm in a phantom and a patient study and show that noise and streaks can be reduced within a short computation time.

Reproducibility of Parenchymal Blood Volume Measurements Using an Angiographic C-arm CT System

RATIONALE AND OBJECTIVES Intra-procedural measurement of hepatic perfusion following liver embolization continues to be a challenge. Blood volume imaging before and after interventional procedures would allow identifying the treatment end point or even allow predicting treatment outcome. Recent liver oncology studies showed the feasibility of parenchymal blood volume (PBV) imaging using an angiographic C-arm system. This study was done to evaluate the reproducibility of PBV measurements using cone beam computed tomography (CBCT) before and after embolization of the liver in a swine model. MATERIALS AND METHODS CBCT imaging was performed before and after partial bland embolization of the left lobe of the liver in five adult pigs. Intra-arterial injection of iodinated contrast with a 6-second x-ray delay was used with a two-sweep 8-second rotation imaging protocol. Three acquisitions, each separated by 10 minutes to allow for contrast clearance, were obtained before and after embolization in each animal. Post-processing was carried out using dedicated software to generate three-dimensional (3D) PBV maps. Two region-of-interest measurements were placed on two views within the right and left lobe on each CBCT 3D PBV map. Variation in PBV for scans acquired within each animal was determined by the coefficient of variation and intraclass correlation. A Wilcoxon signed-rank test was used to test post-procedure reduction in PBV. RESULTS The CBCT PBV maps showed mean coefficients of variation of 7% (range: 2%-16%) and 25% (range:  13%-34%) for baseline and embolized PBV maps, respectively. The intraclass correlation for PBV measurements was 0.89, demonstrating high reproducibility, with measurable reduction in PBV displayed after embolization (P = 0.007). CONCLUSIONS Intra-procedural acquisition of 3D PBV maps before and after liver embolization using CBCT is highly reproducible and shows promising application for obtaining intra-procedural PBV maps during locoregional therapy.

Double Your Views - Exploiting Symmetry in Transmission Imaging.

For a plane symmetric object we can find two views - mirrored at the plane of symmetry - that will yield the exact same image of that object. In consequence, having one image of a plane symmetric object and a calibrated camera, we can automatically have a second, virtual image of that object if the 3-D location of the symmetry plane is known. In this work, we show for the first time that the above concept naturally extends to transmission imaging and present an algorithm to estimate the 3-D symmetry plane from a set of projection domain images based on Grangeats theorem. We then exploit symmetry to generate a virtual trajectory by mirroring views at the plane of symmetry. If the plane is not perpendicular to the acquired trajectory plane, the virtual and real trajectory will be oblique. The resulting X-shaped trajectory will be data-complete, allowing for the compensation of in-plane motion using epipolar consistency. We evaluate the proposed method on a synthetic symmetric phantom and, in a proof-of-concept study, apply it to a real scan of an anthropomorphic human head phantom.

Dynamic Measurement of Arterial Liver Perfusion With an Interventional C-Arm System

Purpose Objective intraprocedural measurement of hepatic blood flow could provide a quantitative treatment end point for locoregional liver procedures. This study aims to validate the accuracy and reproducibility of cone-beam computed tomography perfusion (CBCTp) measurements of arterial liver perfusion (ALP) against clinically available computed tomography perfusion (CTp) measurements in a swine embolization model. Methods Triplicate CBCTp measurements using a selective arterial contrast injection were performed before and after complete embolization of the left lobe of the liver in 5 swine. Two CBCTp protocols were evaluated that differed in sweep duration (3.3 vs 4.5 seconds) and the number of acquired projection images (166 vs 248). The mean ALP was measured within identical volumes of interest selected in the embolized and nonembolized regions of the perfusion map generated from each scan. Postembolization CBCTp values were also compared with CTp measurements. Results The 2 CBCTp protocols demonstrated high concordance correlation (0.90, P < 0.001). Both CBCTp protocols showed higher reproducibility than CTp in the nontarget lobe, with an intraclass correlation of 0.90 or greater for CBCTp and 0.83 for CTp (P < 0.001 for all correlations). The ALP in the embolized lobe was nearly zero and hence excluded for reproducibility. High concordance correlation was observed between the CTp and each CBCTp protocol, with the shorter CBCTp protocol reaching a concordance correlation of 0.75 and the longer achieving 0.87 (P < 0.001 for both correlations). Conclusions Dynamic blood flow measurement using an angiographic C-arm system is feasible and produces quantitative results comparable to CTp.