Hybrid linearization scheme for X-ray CT beam hardening correction

Inventors

Gao, Hewei • Sun, Zhihui

Assignees

RefleXion Medical Inc

Abstract

Disclosed herein are methods for reducing beam-hardening artifacts in CT imaging using a mapping operator that comprises a hybrid spectral model that incorporates air scan X-ray intensity data acquired at two different effective mean energies. In one variation, the air scan X-ray intensity data acquired during a calibration session is combined with an ideal spectral model for each X-ray detector to derive the hybrid spectral mode. A mapping operator based on the hybrid spectral model is used to correct beam-hardening artifacts in the acquired CT projection data. In some variations, the mapping operator is a lookup table of monochromatic (corrected) projection values, and the acquired CT projection data is used to calculate the index of the lookup table entry that contains the corrected projection value that corresponds with the acquired CT projection data.

Core Innovation

A phantom-less calibration method for a CT imaging system is provided to reduce beam-hardening artifacts. The method performs a first air scan at a first effective mean energy and a second air scan at a second effective mean energy higher than the first, and acquires first and second X-ray intensity data for each X-ray detector rather than using a calibration phantom.

From the first and second X-ray intensity data, a virtual filter is calculated for each X-ray detector as a selected material having a thickness. The thickness is determined using a low-to-high ratio of measured X-ray intensities, and the virtual-filter thickness is based on an effective mean-energy spectra model and a virtual-filter attenuation coefficient of the selected material.

A mapping operator is then calculated for each X-ray detector, where the mapping operator uses a hybrid spectral model that converts a polychromatic projection value to a monochromatic projection value. The mapping operator is based on the virtual filter and is implemented as a conversion function that can be realized with lookup tables indexed by discretized polychromatic projection values, optionally using different lookup tables for different CT scan energy levels.

The hybrid spectral model combines an ideal spectral model with the virtual filter in order to approximate the monochromatic projection representation and thereby reduce beam-hardening artifacts. In described implementations, the patent reports a substantial reduction of ring, bowtie, and cupping artifacts and improved CT number uniformity.

Claims Coverage

The document provides one independent claim directed to phantom-less calibration of a CT imaging system to reduce beam-hardening artifacts, using per-detector virtual filters derived from two air scans at different effective mean energies and a hybrid spectral-model-based mapping operator that converts polychromatic to monochromatic projection values.

Across the independent claim and its refinements, the core coverage is directed to deriving detector-wise virtual filters from two air scans at different effective mean energies using an LHR-based thickness determination, and then using a hybrid spectral model, optionally LUT-based and energy-level-specific, to convert polychromatic projection values to monochromatic projection values to reduce beam-hardening artifacts.

Stated Advantages

Documented Applications