Medical Physics Seminar – Monday, October 29, 2012
Radiation Dose Reduction via iterative reconstruction methods: current status and future directions
Ji Tang (invited by Dr. Guang-Hong Chen)
Radiation dose reduction in CT exams has become a top priority in CT research because of concerns over the potential risks associated with the use of ionizing radiation. A variety of dose reduction strategies have been implemented by the joint effort of CT manufacturers, medical physicists, and radiologists in current clinical practice. However, further reduction in radiation dose is still needed, and the iterative image reconstruction methods have been demonstrated to hold promises for further reducing radiation dose down to 50%-10% (depending on the algorithm to be used) of the radiation dose needed today. Recently, almost all major CT vendors have released vendor-specific iterative reconstruction products to reduce radiation dose by about 50%, but the diagnostic performance of these dose reduction software packages is still needed to be determined in prospective clinical studies. Additionally, preliminary clinical studies have also demonstrated that a model based statistical image reconstruction will have promise to reduce radiation dose down to 20%-30%, but the image reconstruction time is still prohibitively long (~hours) for routine clinical deployment. Recently, the Prior Image Constrained Compressed Sensing (PICCS) algorithm developed at UW-Madison has shown great potential for reducing radiation dose down to 10%-30% while image reconstruction time is reduced down to a clinically acceptable range (1-2 minutes) for the same image volume of 300~400 slices. In this seminar, we plan to present an overview of the current iterative reconstruction methods used by each vendor and the PICCS method developed at UW-Madison; we will also present our initial results from the prospective clinical trial carried out by the UW-Madison team, and finally to present the potential challenges and future directions in iterative image reconstruction methods for future ultra-low radiation dose CT practices.