Text-based multi-dimensional medical images retrieval according to the features-usage correlation
1Department of Medical Informatics, Faculty of Medical Sciences, Tarbiat Modares University, Tehran, Iran. aa.safaei@modares.ac.ir.
Medical & Biological Engineering & Computing
|August 20, 2021
Summary
A novel text-based multi-dimensional indexing technique enhances medical image search engines by considering feature correlations. This approach improves both the efficiency and effectiveness of retrieving medical images for clinical and research use.
Area of Science:
- Medical Imaging
- Information Retrieval
- Computer Science
Background:
- The volume of medical images is rapidly increasing, necessitating powerful search engines.
- Existing search engines struggle to efficiently retrieve relevant medical images for clinical and research purposes.
Purpose of the Study:
- To propose a text-based multi-dimensional medical image indexing technique.
- To enhance the effectiveness and efficiency of medical image retrieval systems.
Main Methods:
- Utilized data mining for quantitative association pattern discovery on user query history.
- Fragmented features into subsets based on pairwise feature correlation (Affinity).
- Applied hierarchical clustering to create feature hierarchies, forming a multi-dimensional index structure.
Main Results:
- The proposed indexing technique significantly improves retrieval effectiveness and efficiency.
- Experimental evaluations demonstrated superior performance compared to existing methods like Lucene and Terrier.
- Analysis showed reduced memory usage and improved time complexity.
Conclusions:
- The developed multi-dimensional indexing technique offers a significant advancement for medical image search engines.
- Considering feature correlations semantically enhances retrieval precision.
- The hierarchical structure optimizes resource utilization for better efficiency.
