Whole Reconstruction-Free System Design for Direct Positron Emission Imaging from Image Generation to Attenuation Correction
Background Introduction
A hundred years ago, Hevesy first proposed using radioactive tracers as biological markers in plants, later validated through experiments in rats. This discovery propelled the development of nuclear medicine and molecular imaging in the biomedical field, making it possible to quantitatively visualize biological processes at the molecular level. Among the many imaging techniques, Single Photon Emission Computed Tomography (SPECT) and Positron Emission Tomography (PET) are particularly important as they can quantitatively detect biological functions and metabolism by marking compounds. During the development of these technologies, integrating anatomical information through X-ray Computed Tomography (CT) or Magnetic Resonance Imaging (MRI) further improved the accuracy of diagnostics and data correction.
However, a major limitation of existing systems is the time consumption and noise propagation in the image reconstruction process. In light of this, researchers in recent years have begun exploring next-generation molecular imaging methods that do not rely on mathematical reconstruction steps, namely Direct Positron Emission Imaging (DPEI). DPEI uses ultra-fast Time-of-Flight (TOF) detectors to directly locate the signal source by detecting differences in 511 keV gamma rays. This technology has garnered widespread attention due to its space-saving and flexible geometric properties. To further enhance the practical applicability of the DPEI system in clinical practice, this study introduces a completely new non-reconstruction imaging method, Direct µCompton Imaging, aiming to achieve complete non-reconstruction imaging from image generation to attenuation correction.
Research Origin
This paper, written by scientists Yuya Onishi, Fumio Hashimoto, Kibo Ote, and Ryosuke Ota, was published in the May 2024 issue of the journal “IEEE Transactions on Medical Imaging.” The research work was primarily conducted at the Central Research Laboratory of Hamamatsu Photonics K.K in Japan.
Research Process
Direct µCompton Imaging Algorithm
Direct µCompton Imaging is a method that generates radioactive gamma rays through an external positron source and directly locates Compton scatter positions using ultra-fast TOF detectors. The primary process includes generating positron collisions and measuring the arrival time and position of gamma rays at the detector, using this data to mathematically solve the location of Compton scattering events.
In the study, the Compton scatter positions are solved using the Newton method, recording the xyz positions and time information based on the interaction between gamma rays and the detector. By deriving the geometric relationships through formulas and using event-by-event geometric correction factors, i.e., the Klein-Nishina formula, a µCompton image with anatomical information is finally generated.
Direct Positron Emission Imaging Algorithm
DPEI does not require a collimator and generates three-dimensional images based on the same TOF detector geometry. It only needs to record the xyz position and time of each gamma photon at the detector to mathematically solve the position of annihilation events.
Attenuation Correction
To correct the γ-ray attenuation in DPEI images, the study used µCompton images. First, the µCompton images are converted into three-dimensional attenuation coefficient distributions (µmap), then the attenuation correction factors (ACF) are calculated, and ultimately attenuation-corrected DPEI images are generated.
Experimental Setup
The feasibility of the entirely non-reconstruction imaging system was evaluated through Monte Carlo simulations. The detector used a single block of bismuth germanate crystal, measuring 200×200 mm² in size and 5 mm in thickness. The virtual setup in the simulation included three types of materials: air, water, and bone. Positron sources were uniformly injected into the water sample, and multiple simulations were conducted to evaluate the performance of µCompton and DPEI imaging. Simulated brain samples were also included to approximate clinical scenarios.
Data Analysis
The image intensity generated by Direct µCompton Imaging reflects the probability of Compton scattering, correlating linearly with attenuation coefficients. The experiment analyzed various parameters through simulations of different elements and tissues, showing that µCompton images can be converted into linear attenuation coefficient distributions, and the geometrically corrected images can distinguish materials of different densities.
DPEI images were attenuation-corrected using µCompton images, improving the quantitative accuracy of the images. The fusion effect of DPEI and µCompton images was evaluated under different detector conditions, indicating that combining both methods can simultaneously acquire functional and anatomical information, enhancing the utility and diagnostic accuracy of the images.
Research Results
The experiments demonstrated that the images generated by Direct µCompton Imaging can be fused with DPEI images, and by combining the two imaging methods, both functional and anatomical information could be simultaneously obtained, enhancing the diagnostic accuracy of nuclear medicine clinical examinations. Additionally, using µCompton images for attenuation correction significantly improved the quantitative precision of DPEI images.
In the simulated brain sample experiments, the central region of the uncorrected DPEI images showed significant signal attenuation, whereas the images restored the expected quantitative values after attenuation correction using µCompton images, displaying a 4:1 contrast in brain gray and white matter.
Research Conclusion
This paper demonstrates the feasibility of a completely non-reconstruction imaging system from image generation to attenuation correction through the development of Direct µCompton Imaging. Monte Carlo simulations showed that such a multimodal imaging system could acquire hybrid images of function and anatomy, and using µCompton images for attenuation correction significantly improved the quantitative accuracy of DPEI images. The realization of the entirely non-reconstruction imaging system provides a new perspective and application prospects for molecular imaging, offering significant technical support for the future development of nuclear medicine and molecular imaging fields.
Research Highlights
- Novelty: The proposed Direct µCompton Imaging method is pioneering in the field of non-reconstruction imaging, capable of directly generating anatomical information.
- Practicality: The integrated imaging system not only improves the quantitative accuracy of images but also makes simultaneous acquisition of functional and anatomical information possible, increasing clinical application potential.
- Technical Breakthrough: Utilizing advanced ultra-fast TOF detectors and deep learning technology, it showcases immense potential and development space in nuclear medicine imaging technology.