Single photon and positron-emitting radiopharmaceuticals are widely used for tumor imaging. Among the applications in our institution are the study of tumor response to therapy using a variety of positron-emitting tracers and the therapy of recurrent lymphoma using 131I labeled antibodies. These applications require considerable improvement in the accuracy of in vivo isotope concentration estimates and in the determination of the confidence levels of those measurements. This application deals with optimization of imaging systems and the development of correction schemes for the sources of error in acquired tomographic image data. We will use simulations of imaging systems that include three-dimensional effects and experimental validation of techniques with phantom measurements. The selection of our oncology application areas was based on the large number of such studies ongoing in our institution and because these applications offer a stringent test of correction schemes in both single photon emission tomography (SPECT) and positron emission tomography (PET). We are developing a simulation system to study emission tomographs and their effect on quantitative imaging. The simulation system is designed to use complex objects (variable attenuation and isotope distributions) and to include realistic simulations of collimators and detectors. The software is being implemented with formal software engineering techniques to provide a transportable package, allowing others to implement the simulation system and our correction techniques with minimal effort. In the Monte Carlo portions of the software the use of forced detection, importance sampling, and stratification provides an increase in computational speed of up to 200 times that of conventional Monte Carlo approaches previously used in nuclear medicine. This efficient code is the basis for a correction technique that estimates scatter using the entire reconstructed image data set -- a three dimensional scatter correction. The principal efforts proposed are: 1) development of compute-efficient collimator and detector simulations and extension of the current simulation software; 2) development of simulation-based scatter correction techniques; 3) comparison of several scatter and attenuation correction schemes applicable to SPECT and PET tumor imaging; and 4) establishment of confidence levels for these corrections that include effects caused by the instability of the detector system. Correction schemes to be studied will include: weighted energy acquisition, Compton window subtraction, and deconvolution (including spatially variant formulations as they are developed by other investigators). The application of our correction techniques to human studies will be performed by our consultants as part of their ongoing efforts in tumor imaging.
Showing the most recent 10 out of 32 publications
