MASAI: A high order connectionist system for 3D complex images scanning

We present in this paper a visual connectionist system for medical image interpretation. We propose an original method that gets rid of some invariance problems thanks to a strong dialogue between two cooperative systems of data processing, operating at two different levels. Scale invariances are ta...

Full description

Saved in:
Bibliographic Details
Main Authors: Marchal, Patrick, Anton, Jean-Luc, Guyot, Frederic
Format: Conference Proceeding
Language:English
Subjects:
Artificial neural networks
Encoding
Image edge detection
Image resolution
Image segmentation
Pattern recognition
Online Access:Request full text
Tags: Add Tag
No Tags, Be the first to tag this record!
Description
Summary:We present in this paper a visual connectionist system for medical image interpretation. We propose an original method that gets rid of some invariance problems thanks to a strong dialogue between two cooperative systems of data processing, operating at two different levels. Scale invariances are taken into account through a multi-scale initial encoding. Based on previous works, the system is able to learn and recognize complex 3D patterns. It provides a top-down scanning, and allows a recursive pattern description.
DOI:10.1109/IEMBS.1992.5761341