Cardiac MRI view classification using autoencoder
- 1 December 2014
- conference paper
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- No. 21566097,p. 125-128
Abstract
The growing interest of using cardiac Magnetic Resonance Imaging (MRI) to assess the heart function and structure results in creating huge cardiac image databases. Due to the lack of standard meta-data description of the images, content-based classification of the cardiac images is essential to manage such databases. In particular, cardiac view classification is becoming an important stage for medical image analysis; efficient content-based retrieval as well as CAD systems. The major challenge in such classification lies in the large variability in image appearance caused by variation of patient-specific geometry, heart deformations, and disease conditions. In this work, a fully automated view classification of cardiac MRI images is presented. The method uses the autoencoder system for automatic feature extraction from a training dataset. A softmax classifier is used to determine the view from the selected features. Several autoencoder systems have been investigated in order to select the most suitable architecture for the problem. The results show that the proposed method outperforms current systems with orientation classification accuracy for Cine, SENC and Tagged MRI Imaging sequences was 96.7%, 91.98%, 100% respectively.Keywords
This publication has 10 references indexed in Scilit:
- Classification of cardiac magnetic resonance image type and orientation2014
- Representation Learning: A Review and New PerspectivesIEEE Transactions on Pattern Analysis and Machine Intelligence, 2013
- Content‐based retrieval of brain tumor in contrast‐enhanced MRI images using tumor margin information and learned distance metricMedical Physics, 2012
- Automatic real-time view detection2009
- Content based image retrieval in medical applications: an improvement of the two-level architecture2009
- Automatic Cardiac View Classification of Echocardiogram2007
- Image-Based Multiclass Boosting and Echocardiographic View Classification2006
- Pattern Recognition and Machine Learning2006
- Automatic view recognition in echocardiogram videos using parts-based representation2004
- Nonlinear principal component analysis using autoassociative neural networksAIChE Journal, 1991