Statistical Signal Processing for Radio-Nucleide Tomography

Radio-nucleide imaging is one of the principal diagnostic imaging modalities for detecting, localizing, and asessing metabolic abnormalities in human tissue. Examples include single photon emission computed tomography (SPECT) and positron emission tomography (PET) for cancer detection. In these biomedical imaging modalities a radioactive tracer is injected into the patient and gamma ray emissions are detected on the surfaces of multiple radiation detectors placed around the body. These detectors acquire tomographic projections from which reconstruction of the spatial distribution of the tracer can be performed. As the radioactive decay of the tracer forms a random spatial Poisson process, multisensor statistical signal processing theory can be used to provide reliable detection/estimation algorithms and to specify lower bounds on attainable algorithm performance. In this talk we will discuss applications of this theory to the design of radio-nucleide imaging instruments and associated image reconstruction algorithms.