An inverse optimization method for patient-specific CT organ dose reduction involving tube current modulation

Abstract

Three specific tasks are: First, to develop a procedure for computing organ doses with existing TCM scheme using a recently developed Monte Carlo code, ARCHER, and then using patient-specific anatomical information, such as the scout image of the CT scan or from the patient-specific phantom to generate necessary data for optimization. Second, to develop an optimization algorithm and software tool for new TCM scheme that meets the criteria of minimizing the organ risks while maintaining the CT image quality. Third, to develop a method of evaluating the CT image quality in terms of noise using the GE CatSim software for a given set of TCM scheme. The results show that there is a tradeoff between the image quality and the dose and risk reduction. Aiming to maintain good image quality, the risk-based TCM sets are generated to reduce risks and doses to concerned organs by 15% to 19% while the noise is maintained within 3% increase. Aiming for large dose and risk reduction, the risk-based TCM sets are generated with 31% reduction for the liver, 38% reduction for the stomach, and 46% reduction for female breasts, but the noises are increased by around 15%. In summary, the established framework can generate risk-based TCM sets capable of reducing cancer risks to organs, while maintaining the image noise level. The framework is flexible in the dose reduction and the image quality maintenance, making it a potentially useful tool for personalized TCM generation.

The number of CT scans performed in U.S. each year has increased substantially in the past 20 years, causing concerns to many national and international organizations. The CT radiation doses can lead to stochastic effects such as cancer, and methods to reduce the CT radiation dose are of great interest to the radiology community. Currently, techniques such as Automatic-Exposure-Control (AEC) or Tube-Current-Modulation (TCM) do not fully consider patient-specific organ doses, and there is no algorithm to allow for automatic inverse optimization of the total risk associated with a group of organs of interest. The objective of this PhD work is to develop a patient-specific and organ dose based computational framework that can be used to adjust the TCM for the purposes of reducing the patient stochastic cancer risks for a group of organs.