Multiperspective Recognition Applied to the Computer-Aided Medical Diagnosis - A Comparative Study of Methods
Abstract
One of the most frequently used methods for computer-aided diagnosis is pattern recognition. The classical pattern recognition method assigns a given pattern to one and only one class from a given set of classes. In contrast, multiperspective classification is a process in which an object undergoes several classification tasks. Each task denotes recognition from a different point of view and with respect to different sets of classes. This work presents several different approaches to the algorithmization of multiperspective diagnosis and the implied decision algorithms. Several decision algorithms are presented for three different approaches to multiperspective classification (i.e., direct, decomposed independent, and decomposed dependent). These algorithms are the probabilistic (empirical Bayes) algorithm, the nearest-neighbor algorithm, fuzzy method, and artificial neural networks of the back propagation and counter propagation types. The proposed approaches and algorithms were applied to the computer-aided diagnosis of chronic renal failure and medical decisions in non-Hodgkins lymphoma. Results show that the best diagnostic outcome was achieved through use of the back propagation neural network. The probabilistic algorithm and the counter propagation neural network were less effective, and the fuzzy logic algorithm was the worst performer of all. (3 tables, 5 refs.)
Document Details
- Document Type
- Technical Report
- Publication Date
- Oct 25, 2001
- Accession Number
- ADA412703
Entities
People
- Edward Puchala
- Marek W. Kurzynski
Organizations
- Wrocław University of Science and Technology