Siemens mMR Research Articles

Modelling the impact of injection time on the bolus shapes in PET-MRI AIF Conversion

With the introduction of combined PET/MRI systems, AIF conversion can be made under certain circumstances (see [1]). We propose a model that allows modification of the injection parameters in the AIF fit to account for differences caused by different injection durations [2]. Brain 18F-Choline PET and DSC-MRI data were obtained using Siemens mMR. The MR contrast agent was injected with a rate of 4ml/sec and the PET tracer was injected manually. Perfusion Mismatch Analyzer [3] was used to extract the MRI-AIF. Carotid arteries were segmented on a post contrast MPRAGE image. PET frames were registered onto this MPRAGE image using rigid registration and partial volume correction was done using the iterative Yang method [4]. The AIFs were fitted using a convolution of a ‘double Butterworth’ function, representing the injection, with a tri-exponential function representing the elimination [Eq. 1]. The bolus shape can be adjusted by changing Δτ (τ2 - τ1). This was tested with a population based MRI AIF [5], as well as with clinical data. 1 where For the population based input function, Figure ​Figure11 shows that when Δτ was increased, lower and wider peaks were seen, and with decreased Δτ, higher but narrower peaks were observed. Figure ​Figure22 shows that the function fits both clinical PET and MRI AIFs well. Values of τ1 and τ2 were changed to modify the MRI-AIF and Figure ​Figure33 shows the modified MRI-AIF together with the original fitted PET-AIF, normalized to their peaks. Two AIFs have similar peak shapes but start to differ at the elimination phase as Gd-DOTA and 18F-Choline have different tissue uptake rates. Figure 1 Simulated MRI-AIFs using Parker’s population-based input function refitted with the developed function. AIF shapes with different injection durations, Δτ is shown. Figure 2 The double Butterworth convolution function used to fit (a) DSC-MRI data and (b) 18F-Choline PET data together with a plot where the timescale of PET-AIF was limited to MRI-AIF’s to show different bolus widths. Figure 3 The MRI-AIF with modified τ1 and τ2 values plotted together with the PET-AIF. The MRI-AIF peak is scaled to PET-AIF’s peak. This enables conversion of the early part of the AIFs from one modality to another even if different injection protocols are used.

Sparsely sampled MR navigators as a practical tool for quality control and correction of head motion in simultaneous PET/MR.

We present a study using MR navigators sampled between other protocolled MR sequences during PET/MR brain scanning, to make motion-QC and motion correction (MC) feasible in clinical simultaneous PET/MR, obviating the need for special MR sequences with interleaved (dense) navigators. During 30 min [11C]-PiB (453±148 MBq) PET listmode scans of 29 patients on the Siemens mMR, a set of 10 3D navigator volumes was acquired (2D EPI 3.0x3.0x3.0 mm3 voxels, 64x64 matrix, 36 slices, TE 30 ms, TR 3,000 ms) in between other scheduled MR acquisitions at 6 time points. The rigid motion between navigators was calculated (all motions given are for a point in the cortex 6 cm from the scanners cFOV). The 3 patients with largest motion (11, 8 and 8 mm) and the only 2 PiB-positive patients with motion > 4 mm (8 and 6 mm motion) were reconstructed into 6 frames with frames being split at the midpoint between navigator start times (framing [mean±SD]: 259±8, 306±8, 191±2, 342±2, 546±1, 157±19 seconds) and then averaged to one image after MC. A blinded evaluation comparing non-MC and MC PiB images was performed by a nuclear medicine physician. The average maximum motion magnitude was 3.9±2.4 mm (1 to 11 mm). The visual evaluation of the 5 MC patients rated 3/5 non-MC images blurred (average score: 3.4±0.5 on a scale of 1-5) and 0/5 MC images blurred (score: 4.0±0.0) with no correlation between motion magnitude and rating. A slight, overall decrease in blurring and an increase in image quality rating with MC was found, but with no impact on clinical outcome. The effect of MC is limited with PiB, but with other tracers (e.g. FDG), longer scans, smaller ROIs or larger motion magnitudes, navigators are easily used for motion-QC and MC in clinical practice.

Non rigid respiratory motion correction in whole body PET/MR imaging

Respiratory motion in PET/MR imaging leads to reduced quantitative and qualitative image accuracy. Correction methodologies include the use of respiratory synchronized gated frames which lead to low signal to noise ratio (SNR) given that each frame contains only part of the count available throughout an average PET acquisition. In this work, 4D MRI extracted elastic transformations were applied to list-mode data either inside the image reconstruction or to the reconstructed respiratory synchronized images to obtain respiration corrected PET images. Five patient datasets acquired on the SIEMENS mMR PET/MR system were used. T1-weighted 4D MR images were registered to the end expiration MR image using an elastic B-spline registration algorithm to derive deformation matrices used to account for the respiratory motion. The derived matrices were subsequently applied during the List Mode OSEM PET image reconstruction of the original emission list-mode data. The corrected images were compared with those produced by applying the elastic transformation after reconstructing PET frames followed by summing the realigned gated frames, as well as with uncorrected motion averaged images. Significant improvement was obtained by using any of the two respiratory motion correction techniques compared to uncorrected motion average images (>30% differences in tumor position and size and 25% SNR increase). Both correction approaches lead to nearly equivalent improvements (differences of <8% and <6% for the lesion FWHM and the SNR respectively). A list-mode reconstruction based respiratory motion correction for PET has been implemented and its performance evaluated. This approach was based on the use of elastic transformations derived from 4D MRI during PET image reconstrucion. Our results show significant respiratory motion compensation when compared to the motion average PET images, with slightly better contrast and SNR compared to an equivalent 4D PET image space elastic motion correction method.

Towards coronary plaque imaging using simultaneous PET-MR: a simulation study

Coronary atherosclerotic plaque rupture is the main cause of myocardial infarction and the leading killer in the US. Inflammation is a known bio-marker of plaque vulnerability and can be assessed non-invasively using fluorodeoxyglucose-positron emission tomography imaging (FDG-PET). However, cardiac and respiratory motion of the heart makes PET detection of coronary plaque very challenging. Fat surrounding coronary arteries allows the use of MRI to track plaque motion during simultaneous PET-MR examination. In this study, we proposed and assessed the performance of a fat-MR based coronary motion correction technique for improved FDG-PET coronary plaque imaging in simultaneous PET-MR. The proposed methods were evaluated in a realistic four-dimensional PET-MR simulation study obtained by combining patient water–fat separated MRI and XCAT anthropomorphic phantom. Five small lesions were digitally inserted inside the patients coronary vessels to mimic coronary atherosclerotic plaques. The heart of the XCAT phantom was digitally replaced with the patient's heart. Motion-dependent activity distributions, attenuation maps, and fat-MR volumes of the heart, were generated using the XCAT cardiac and respiratory motion fields. A full Monte Carlo simulation using Siemens mMR's geometry was performed for each motion phase. Cardiac/respiratory motion fields were estimated using non-rigid registration of the transformed fat-MR volumes and incorporated directly into the system matrix of PET reconstruction along with motion-dependent attenuation maps. The proposed motion correction method was compared to conventional PET reconstruction techniques such as no motion correction, cardiac gating, and dual cardiac-respiratory gating. Compared to uncorrected reconstructions, fat-MR based motion compensation yielded an average improvement of plaque-to-background contrast of 29.6%, 43.7%, 57.2%, and 70.6% for true plaque-to-blood ratios of 10, 15, 20 and 25:1, respectively. Channelized Hotelling observer (CHO) signal-to-noise ratio (SNR) was used to quantify plaque detectability. CHO-SNR improvement ranged from 105% to 128% for fat-MR-based motion correction as compared to no motion correction. Likewise, CHO-SNR improvement ranged from 348% to 396% as compared to both cardiac and dual cardiac-respiratory gating approaches. Based on this study, our approach, a fat-MR based motion correction for coronary plaque PET imaging using simultaneous PET-MR, offers great potential for clinical practice. The ultimate performance and limitation of our approach, however, must be fully evaluated in patient studies.