Joint estimation; TOF-PET
Emission-based Joint Estimation of Patient and Hardware Attenuation Distributions for Hybrid PET/MR Imaging (#3271)
T. Heußer1, Y. Berker1, 2, M. T. Freitag3, M. Kachelrieß1
1 German Cancer Research Center (DKFZ), Medical Physics in Radiology, Heidelberg, Germany
Attenuation correction (AC) for both patient and hardware is required for accurate PET quantification. The current clinical value of hybrid PET/MR imaging is limited because MR-based AC (MRAC) does not properly account for bone attenuation and neglects attenuation of flexible hardware components (headphones, flexible body coils). To improved PET quantification for PET/MR, we developed a new reconstruction approach combining ideas of our recently published methods for emission-based patient and hardware AC. The new algorithm, xMR-MLAA, is based on maximum-likelihood reconstruction of attenuation and activity (MLAA), and estimates patient and hardware attenuation in an interleaved manner. The proposed method was evaluated for clinical head scans acquired with a Siemens Biograph mMR. Conventional T1-weighted MR images were used to derive the patient support as well as anatomical prior information on the patient attenuation distribution. Standard MRAC results in activity underestimation in the brain, caused both by neglecting headphone attenuation and by treating bone as soft tissue during AC. Across three FDG patients included in this study, average brain activity underestimation was 14.8% compared to CTAC. Using xMR-MLAA, bone attenuation information could be accurately recovered while preserving air cavities like the nasal sinuses and the inner ears. Moreover, accurate estimates of the headphone attenuation were obtained. Average brain activity underestimation with xMR-MLAA was reduced to 3.1% compared to CTAC.
Keywords: PET/MR, Attenuation Correction, MLAA
A Differential PET Image Reconstruction Method for Improved Sensitivity of Tau Protein Deposition in Alzheimer Disease Progression Monitoring (#2262)
A. Ihsani1, J. Dutta2, K. Johnson3, J. A. Becker3, J. Ouyang1, G. El Fakhri1
1 Massachusetts General Hospital & Harvard Medical School, Gordon Center for Medical Imaging, Radiology, Boston, Massachusetts, United States of America
Alzheimer Disease (AD) is a neurodegenerative disease that reduces the cognitive and physical abilities of affected individuals. Recent literature suggests that the progression of AD may be monitored by observing hyperphospholylated Tau protein accumulation in the form of neurofibrillary tangles in the brain. AV1451 (a PET radiotracer) binds to all forms Tau protein, but highlights regions where there is excessive buildup, therefore AV1451 seems adequate to monitor the progression of AD. Currently, two (or more) scans of the same patient are taken years apart, and progression of Tau protein deposition is evaluated by point-wise subtraction of independently reconstructed images. These difference images are overly noisy and very sensitive to alignment artifacts. We propose to reconstruct the difference image jointly from two scans for increased sensitivity to subtle changes in Tau buildup. The difference images resulting from the proposed approach show potential in improving the sensitivity to Tau deposition compared to the existing approach, ultimately enabling shorter time intervals between consecutive scans.
Keywords: Alzheimer Disease, tau protein, AV1451, T807, joint image reconstruction
Joint Reconstruction of Stress Imaging and Reversible Map for Cardiac SPECT Imaging (#3850)
X. Lai1, Y. Petibon1, G. El Fakhri1, J. Ouyang1
1 Masschussetts General Hospital and Harvard Medical School, Radiology Department, Gordon Center for Medical Imaging,, Boston, United States of America
Rest-Stress perfusion SPECT imaging protocol is widely used to identify cardiac diseases. In this protocol, a patient is imaged twice: once when the patient is at rest and the other when the heart is at stress either by excise or pharmacologically. One critical step for cardiac disease diagnosis is to identify the reversible defects, which shows perfusion deficit in the stress image while it is not in the rest one. Besides visually findings the reversible defects, the reversible map is generated by registering, normalizing, and subtracting the two imaging volumes. In this work, we applied a joint reconstruction method recently developed in our group. Both stress imaging and revisible map are jointly reconstructed while preserving Poisson noise modeling, which is critical to the image quality of reconstructed SPECT images. This joint reconstruction approach circumvents the need to subtract coregistered stress and rest images as in the conventional approach, significantly improved imaging qualities, and could benefit the clinical diagnosis.
Keywords: Cardiac SPECT Imaging
Constant determination and simultaneous activity and attenuation estimation using TOF data and single events (#2172)
T. Feng1, J. Wang1, W. Zhun1, H. Li1
1 UIH America, Inc, Houston, Texas, United States of America
Maximum likelihood activity and attenuation estimation (MLAA) using time-of-flight (TOF) data can determine activity and attenuation up to a constant. Prior knowledge is widely used for the determination of the constant. However, it may result in quantitation errors due to the discrepancy between prior and truth. The goal of this study is to develop a method to obtain accurate estimation of the constant and hence determine patient specific, quantitatively correct activity and attenuation.
The depth dependent attenuation factors in single events break the symmetry between activity and attenuation factors. We showed mathematically that in a 2D case, with the combination of TOF information and single events, a unique solution of attenuation and activity can be achieved, minus some special cases such as a scan of a point source. An iterative algorithm was developed to reconstruct activity and attenuation estimation utilizing TOF data and single events. Each iteration has three steps: 1) updating activity distribution using MLEM approach with TOF data;2) updating attenuation map using MLTR with non-TOF data;3)correcting constant coefficient in both activity distribution and attenuation map with single event data. No prior knowledge was applied in the iteration. 2D analytical simulation with no noise was generated to validate our method, in which both scatter information and randoms information were assumed to be known. Percentage root mean square error was used for quantitative analysis of the image reconstruction.
The simulated results showed that the new three-step iterative approach successfully recovered both activity distribution and attenuation map. The quantitative accuracy of activity is within 1% and the accuracy of attenuation is within 4% after 100 iterations.
In this paper, we showed that the attenuation and activity can be uniquely acquired with both TOF and single data without other prior information. The 2D derivation and simulation can extend to 3D as well.
Keywords: MLAA, Constant, Activity, attenuation, TOF, single events
Axial Fourier Rebinning for Time-of-Flight PET (#3893)
Y. Li1, S. Matej1, S. D. Metzler1
1 Penn, Philadelphia, PA, United States of America
Fully 3D time-of-flight (TOF) PET scanners offer the potential of substantially improved image quality in clinical PET imaging. The main challenges of 3D TOF PET imaging are the data storage with either list-mode or binned formats, and the reconstruction time using iterative algorithms. Previously, we derived the Fourier rebinning and consistency equations (FORCEs), and showed 3D TOF data can be fully characterized by the consistency equations. In this work, we present an exact Fourier rebinning for 3D TOF data based on the axial consistency equation to dramatically reduce the data storage and the reconstruction time. Starting from pre-corrected 3D TOF data, the axial Fourier rebinning algorithm estimates a 2D TOF sinogram for each transverse slice without information loss. Since the 3D TOF data are axially truncated, we provide provide a solution to estimate the missing portion in the oblique TOF projection data. The proposed axial Fourier rebinning for TOF data (axFRT) can take advantage of all the 3D TOF data statistics, and the rebinning 2D TOF data can then be reconstructed using any algorithm for 2D or 2.5D TOF reconstructions. The axFRT algorithm allows the axial data sets being rebinned independently, and there are tens of thousands of such data sets which can naturally take advantage of the massively parallel processors to dramatically speedup the rebinning. We show numerical simulations to demonstrate that axFRT produces accurate and unbiased rebinned sinograms even for TOF PET with large axial acceptance angle. The axFRT will be particularly useful for 3D TOF PET with large axial field of view for PET imaging applications including dynamic, whole- or total-body imaging.
Keywords: Axial Fourier rebinning, image reconstruction, time-of-flight, positron emission tomography (PET)
Can Time-of-Flight Information be Used to Mitigate Detector Induced Blur in PET Reconstruction? (#1969)
M. Toussaint1, 2, J. - P. Dussault1, R. Lecomte2
1 Université de Sherbrooke, Department of Computer Science, Sherbrooke, Québec, Canada
Time of Flight in Positron Emission Tomography (ToF-PET) provides multiple advantages in terms of image contrast, but normally has no incidence on spatial resolution, which is limited by the detector size and other physical factors. We hypothesize that ultra-fast ToF has the potential to mitigate the influence of detector blur by providing more accurate spatial information that could actually improve the PET image resolution. Let us consider the case where blurring due to ToF is far less than the blur induced by the detectors. In such case, the blur observed in a projection would mainly result from the detectors. By appropriately exploiting the more accurate spatial information provided by the ToF response, it should be possible to reduce the impact of the detector blur on PET resolution. We propose a novel approach to exploit ToF information in the reconstruction process in order to test the hypothesis. Two simplified acquisition configurations were considered with a fixed ToF resolution. The first one is built such that the detector blur at the center of the camera is greater than the ToF resolution, while the reverse is considered in the second case. Image spatial resolution was evaluated by imaging a hot spot phantom and observing which spots can be resolved. In the first case, the ToF model was able to resolve hot spots that are way smaller than the detector resolution. In the second case, the ToF reconstruction achieved a similar resolution as the classical PET reconstruction, as expected. This proof of concept must be further investigated with more realistic datasets generated by a validated simulator such as openGATE. In the advent of a practical implementation, the proposed ToF reconstruction paradigm would potentially make higher resolution achievable in clinical PET, or make it possible to use larger crystals without loss of resolution.
Keywords: Time-of-Flight, PET Image Reconstruction, Spatial Resolution, Iterative Algorithm