Related Experiment Video
Updated: Jun 25, 2026
14:27
Identification of Disease-related Spatial Covariance Patterns using Neuroimaging Data
Published on: June 26, 2013
A pattern similarity scheme for medical image retrieval.
Dimitris K Iakovidis1, Nikos Pelekis, Evangelos E Kotsifakos
1University of Athens, Ilisia, Greece. dimitris.iakovidis@ieee.org
Summary
This study introduces a new method for efficient medical image retrieval using the PAtterns for Next generation DAtabase systems (PANDA) framework. The approach uses clustering to find semantically meaningful patterns in radiographic images for better database searching.
Area of Science:
- Medical Imaging
- Database Systems
- Computer Science
Background:
- Efficient retrieval of medical images from large databases is crucial for clinical decision-making.
- Current content-based medical image retrieval (CB-MIR) methods often struggle with semantic interpretation and scalability.
Purpose of the Study:
- To propose a novel and efficient scheme for content-based medical image retrieval.
- To leverage the PAtterns for Next generation DAtabase systems (PANDA) framework for pattern representation and management in medical images.
- To enable unsupervised semantic interpretation of retrieval results.
Main Methods:
- Block-based low-level feature extraction from medical images.
- Clustering of feature space using an expectation-maximization algorithm to form semantically meaningful patterns.
- Exploitation of the PANDA framework's 2-component property for cluster similarity estimation.
Main Results:
- The proposed scheme demonstrated efficient and effective application for medical image retrieval from large datasets.
- The method successfully generated higher-level, semantically meaningful patterns from low-level image features.
- Unsupervised semantic interpretation of retrieval results was achieved.
Conclusions:
- The novel scheme provides an efficient and effective solution for content-based medical image retrieval.
- The PANDA framework facilitates robust pattern representation and management for medical image databases.
- The approach offers potential for extension with knowledge representation methodologies for enhanced medical image analysis.
