Abstract Archives of the RSNA, 2012
Assessment of the Potential Improvement in CT Number Accuracy Using a Raw-data Based Metal Artifact Reduction Technique in Phantom and Patient Data
Scientific Formal (Paper) Presentations
Presented on November 30, 2012
Presented as part of SST15: Physics (Image-guided Radiation Therapy)
Katharine Grant PhD, Presenter: Employee, Siemens AG
Rainer Raupach PhD, Abstract Co-Author: Employee, Siemens AG
Esther Meyer, Abstract Co-Author: Employee, Siemens AG
Michael Yaszemski MD, PhD, Abstract Co-Author: Nothing to Disclose
Thomas G. Flohr PhD, Abstract Co-Author: Employee, Siemens AG
Bernhard Schmidt PhD, Abstract Co-Author: Employee, Siemens AG
Implanted hardware (high z materials such as metal and ceramics) affect the accuracy of CT numbers used for radiation therapy planning. Here, we assess the performance of a novel raw data based algorithm (FSNMAR: Meyer et al. SPIE 2012) in improving the accuracy of CT numbers in the presence of metal implants, compared to standard weighted filtered back projection (FBP) and existing techniques for reducing metal artifacts (MAR).
Several different high z material components including joint and dental prosthesis were arranged in acrylic water phantoms mimicking the head and torso. Materials tested included titanium alloys, surgical grade stainless steel, gold, ceramic and an amalgam. All 6 phantoms containing hardware were scanned at routine dose and maximum dose (high mAs) using an appropriate base protocol (routine head for dental, routine pelvis for hips, etc.) utilizing both single and dual energy protocols. Each data set was reconstructed with FBP, MAR and FSNMAR.
In phantoms, the FSNMAR algorithm consistently reduced the standard deviation of the Hounsfield units (HU – CT numbers) surrounding the hardware and also provided qualitative improvements in image quality. Not only did FNSMAR provide improvements in image quality compared to standard MAR methods, but more impressively, when compared to standard FBP, FSNMAR reduced the SD of CT numbers by: 84% for a stainless steel spine screw base plate; 72% for titanium spine screws; 85% for a titanium femoral revision; and 78% in the real tooth with Amalgam.
FSNMAR consistently reduces artifacts that arise from high atomic number (z) materials without increasing image noise. Compared to MAR and FBP, FSNMAR provides a significant reduction in CT number variations in the presence of metal hardware, improving CT number, and thus, radiation therapy planning accuracy. FSNMAR also has the potential to improve image quality and diagnostic accuracy in CT images when metal hardware is present.
Metal artifacts in CT images lead to an unwanted shift in CT numbers, directly impacting radiation therapy planning. FSNMAR substantially improves the accuracy of CT values and thus, therapy planning.
Assessment of the Potential Improvement in CT Number Accuracy Using a Raw-data Based Metal Artifact Reduction Technique in Phantom and Patient Data. Radiological Society of North America 2012 Scientific Assembly and Annual Meeting, November 25 - November 30, 2012 ,Chicago IL. http://archive.rsna.org/2012/12025045.html