Open AccessThe Automatic Polar Map Creation for Myocardial Perfusion SPECT Analysis Using Image Registration Combined with Feature Extraction
Author(s) -
Gabriel Patzer,
Marcelo Do Nascimento,
Marcus Vinícius Simões,
Jeovane Honório Alves,
Lauro WichertAna,
Paulo de Azevedo-Marques,
Lucas De Oliveira
Publication year - 2015
Language(s) - English
Resource type - Conference proceedings
DOI - 10.5753/sbcas.2015.10360
Subject(s) - artificial intelligence , image registration , computer science , feature extraction , computer vision , feature (linguistics) , gold standard (test) , pattern recognition (psychology) , polar , software , image (mathematics) , radiology , medicine , physics , philosophy , linguistics , astronomy , programming language
This paper describes a technique for an automatic polar map creation from myocardial perfusion SPECT images. This exam is widely used in post-infarction patient evaluations, in order to foretell the outcome and left ventricular function. This exam is difficult to interpret, since it is a 3D representation of the heart. Use of polar maps intends to simplify analysis of the exam, converting the 3D image into a 2D plot. The technique developed is based on a combination of image registration and feature detection. For this study, an overall of 31 cases were tested, with the results compared with the gold standard software. The correlation calculated between techniques was 0.76 in the worst case and 0.98 in the best case. Index Terms: Image Registration, Myocardial Infarction, Polar Map, SPECT.