Assessing patient dose in interventional fluoroscopy using patient-dependent hybrid phantoms

Interventional fluoroscopy uses ionizing radiation to guide small instruments through blood vessels or other body pathways to sites of clinical interest. The technique represents a tremendous advantage over invasive surgical procedures, as it requires only a small incision, thus reducing the risk of infection and providing for shorter recovery times. The growing use and increasing complexity of interventional procedures, however, has resulted in public health concerns regarding radiation exposures, particularly with respect to localized skin dose. Tracking and documenting patient-specific skin and internal organ dose has been specifically identified for interventional fluoroscopy where extended irradiation times, multiple projections, and repeat procedures can lead to some of the largest doses encountered in radiology. Furthermore, inprocedure knowledge of localized skin doses can be of significant clinical importance to managing patient risk and in training radiology residents.;In this dissertation, a framework is presented for monitoring the radiation dose delivered to patients undergoing interventional procedures. The framework is built around two key points, developing better anthropomorphic models, and designing clinically relevant software systems for dose estimation. To begin, a library of 50 hybrid patient-dependent computational phantoms was developed based on the UF hybrid male and female reference phantoms. These phantoms represent a different type of anthropomorphic model whereby anthropometric parameters from an individual patient are used during phantom selection. The patient-dependent library was first validated and then used in two patient-phantom matching studies focused on cumulative organ and local skin dose. In terms of organ dose, patient-phantom matching was shown most beneficial for estimating the dose to large patients where error associated with soft tissue attenuation differences could be minimized. For small patients, inherent difference in organ size and location limited the effectiveness of matching. For skin dose, patient-phantom matching was found most beneficial for estimating the dose during lateral and anterior-posterior projections. Patient-sculpting of the patient.s outer body contour was also investigated for use during skin dose estimation and highlighted as a substantial step towards better patient-specificity.;In order to utilize the models for actual patient dosimetry, two programs were developed based on the newly released Radiation Dose Structured Report (RDSR). The first program allows for the visualization of skin dose by translating the reference point air kerma to the location of the patient.s skin characterized by a computational model. The program represents an innovative tool that can be used by the interventional physician to modify behavior when clinically appropriate. The second program operates by automatically generating an input file from the RDSR which can then be run within a Monte Carlo based radiation transport code. The program has great potential for initiating and promoting the concept of 'cloud dosimetry', where patient-specific radiation transport is performed off-site and returned via the internet. Both programs are non-proprietary and transferable, and also incorporate the most advanced computational phantoms developed to date. Using the tools developed in this work, there exist a tangible opportunity to improve patient care with the end goal being a better understanding of the risk/benefit relationship that accompanies the medical use of ionizing radiation.