Knowledge discovery approach to automated cardiac SPECT diagnosis

The paper describes a computerized process of myocardial perfusion diagnosi s from cardiac single proton emission computed tomography (SPECT) images us ing data mining and knowledge discovery approach. We use a six-step knowled ge discovery process. A database consisting of 267 cleaned patient SPECT im ages (about 3000 2D images), accompanied by clinical information and physic ian interpretation was created first. Then, a new user-friendly algorithm f or computerizing the diagnostic process was designed and implemented. SPECT images were processed to extract a set of features, and then explicit rule s were generated, using inductive machine learning and heuristic approaches to mimic cardiologist's diagnosis. The system is able to provide a set of computer diagnoses for cardiac SPECT studies, and can be used as a diagnost ic tool by a cardiologist. The achieved results are encouraging because of the high correctness of diagnoses. (C) 2001 Elsevier Science B.V. All right s reserved.