This document describes a hybrid intelligent system for diagnosing myocardial perfusion from cardiac images. It proposes using a backpropagation neural network to classify cardiac SPECT images as normal or abnormal, but notes the neural network's performance is poor. It then adds a fuzzy logic system to model a cardiologist's risk assessment. The fuzzy system takes the neural network's two output values as inputs and outputs a risk level. Fuzzy sets and rules are defined based on a cardiologist's expertise. Heuristics are also added to decrease or increase risk levels based on perfusion levels in images. The hybrid system aims to better model a cardiologist's diagnosis than the neural network alone.