Systems and methods for low-count quantitative projection domain SPECT imaging
Inventors
Jha, Abhinav Kumar • LI, Zekun • Thorek, Daniel • Rahman, Md Ashequr
Assignees
Interested in licensing this patent?
MTEC can help explore whether this patent might be available for licensing for your application.
Publication Number
US-12582361-B2
Publication Date
2026-03-24
Expiration Date
Abstract
A system for low-count quantitative single-photon emission computed tomography (LC-QSPECT) is provided. The system is programmed to a) store a computer tomography (CT) scan of a subject being examining including a plurality of defined volumes of interest (VOIs) of the subject being examined; b) model a system matrix based on the stored CT, wherein the model describes the probability that photons emitted from each of the defined VOIs are detected in different projection bins, wherein a plurality of projection bins are defined around the subject and the defined VOIs; c) adjust the model with analysis of stray-radiation noise around the subject; d) detect, by the one or more sensors, one or more photons being emitted by an alpha-particle-emitting isotope; and e) execute the adjusted model with the one or more detected photons as inputs to determine a source VOI of the detected photons.
Core Innovation
The disclosed invention provides LC-QSPECT/MEW-PDQ for low-count quantitative single-photon emission computed tomography in an α-particle radiopharmaceutical therapy setting. It uses computer tomography (CT) scan information to define regional defined volumes of interest (VOIs) for a subject rather than using voxel grids. In this framework, the detected photons are used to determine a source VOI for the regional uptake from low-count photon measurements.
The invention constructs a projection-domain system matrix in which the probability that photons emitted from each defined VOI are detected in different projection bins is modeled. The model is explicitly adjusted with stray-radiation-related noise, including a Poisson noise model with a mean noise term. The resulting adjusted model is executed to determine the source VOI using an iterative maximum-likelihood (MLEM-like) estimation approach.
The disclosure further addresses quantification in the presence of overlap/crosstalk between αRPT daughter isotopes by using multiple-energy-window projection-domain quantification (MEW-PDQ). It jointly models the coupled contributions across multiple energy windows while accounting for noise and decay, including specifically 227Th and 223Ra. This enables direct projection-domain estimation of regional uptake while accounting for partial-volume effects and avoiding reconstruction information loss.
Claims Coverage
The relevant independent claims are directed to an LC-QSPECT system and a corresponding method/device, with four primary inventive features. Across the independent claims, the core claim scope is centered on CT-defined VOIs with boundaries different from voxels, projection-bin system-matrix modeling of detection probability, adjustment for stray-radiation noise, and estimation of a source VOI from detected photons emitted by an alpha-particle-emitting isotope.
CT-defined VOIs with non-voxel boundaries
Store a computer tomography (CT) scan of a subject being examining including a plurality of defined volumes of interest (VOIs) of the subject being examined, and define the plurality of defined VOIs based on the CT scan such that the plurality of defined VOIs are different than and have different boundaries from any plurality of voxels defining the subject to be examined.
Projection-bin system-matrix probability modeling
Model a system matrix based on the stored CT scan, wherein the model describes the probability that photons emitted from each of the plurality of defined VOIs are detected in different projection bins, wherein a plurality of projection bins are defined around the subject and the plurality of defined VOIs.
Stray-radiation-noise adjusted system matrix
Adjust the model with analysis of stray-radiation noise around the subject.
Source VOI determination from detected alpha-particle photons
Detect, by one or more sensors, one or more photons being emitted by an alpha-particle-emitting isotope within the subject, and execute the adjusted model with the one or more detected photons as inputs to determine a source VOI of the one or more detected photons.
Together, the independent claims cover a CT-based VOI approach with non-voxel boundaries, projection-bin system-matrix modeling of photon detection probability, explicit adjustment for stray-radiation noise, and determining a source VOI from detected photons emitted by an alpha-particle-emitting isotope.
Stated Advantages
Reduced partial volume effects (PVEs) and reduced partial-volume bias.
Avoidance of reconstruction information loss.
Improved effective sensitivity by using multiple-energy-window quantification.
Documented Applications
Low-count quantitative single-photon emission computed tomography (LC-QSPECT) for α-particle radiopharmaceutical therapy, including quantification involving 227Th and 223Ra through multiple-energy-window projection-domain quantification (MEW-PDQ).
Interested in licensing this patent?