Purpose: Imaging in the presence of implants (instrumentation and prostheses) presents a notoriously difficult challenge to CT because of photon starvation and beam hardening. To alleviate these limitations, a statistical reconstruction approach that includes knowledge of implant shape and composition was previously reported. This work extends the approach to modeling of photon transport, including polychromatic x‐ray beams and scatter, and evaluates the method in simulated and real data. Methods: Previous work on Known‐Component Reconstruction (KCR) is first extended to include a polyenergetic beam (KCR‐POLY). The method simultaneously estimates the unknown background volume and the position of implants with known attenuation and shape. Simulations included an anthropomorphic knee with a Co‐Cr‐Mo implant and system model for an extremities CT system (110 kVp+0.2 mm Cu). Experimental validation was performed on an imaging bench in which a Titanium spine fixation rod (65 mm long, 5.5 mm diameter) was imaged within a 20.5 cm diameter water cylinder (120 kVp+0.2 mm Cu) in geometry simulating an interventional C‐ arm. Results: The polyenergetic system model was essential to high image quality in KCR reconstructions of large, highly attenuating implants such as knee prostheses and spine instrumentation, where standard penalized‐ likelihood and monoenergetic variants of KCR fail. The first application of KCR‐POLY in real data demonstrates the potential of the algorithm in practice, reducing or eliminating artifacts and restoring image uniformity. Conclusions: The KCR‐POLY algorithm yielded major reduction in metal artifacts, owing both to a priori component knowledge (the implant) and account of the polyenergetic beam, object attenuation, and x‐ray scatter. Ongoing research focuses on improvements to the registration algorithm, scatter, and experimental studies with complex, deformable implants. The work supports application of CT to a range of applications conventionally prohibited by metal implants ‐ e.g. surgical guidance or diagnostic imaging of joints with prostheses. This work was supported in part by NIH 2R01‐CA‐112163.
ASJC Scopus subject areas
- Radiology Nuclear Medicine and imaging