A Framework for Medical Image Retrieval using Machine Learning and Directional Local Ternary Quantized Extrema Patterns

• Published in: 2019 International Conference on Smart Systems and Inventive Technology (ICSSIT) pp. 813 - 818
• Main Authors: Ashokkumar, C., Sathiamoorthy, S.
• Format: Conference Proceeding
• Language: English
• Published: IEEE 01.11.2019
• Subjects: Benchmark testing; Color histogram; Computed tomography; Directional local ternary quantized extrema pattern; Diseases; Feature extraction; Image retrieval; Medical diagnostic imaging; NEMA; Optimization; Particle swarm optimization; Support vector machine; Support vector machines
• DOI: 10.1109/ICSSIT46314.2019.8987930

In recent days, the application of medical images has been exponentially raised and it offers comprehensive information about the patient's health condition. It is applicable to diagnose the disease and also saved in a database for research purposes to get a detailed knowledge about the reason and remedial of the diseases. To retrieve the medical images effective in real time scenario, there is a great requirement of developing an innovative medical image indexing and retrieval model. This study introduces a new medical image retrieval (MIR) model using, Directional local ternary quantized extrema pattern (DLTerQEP). Next, particle swarm optimization (PSO) with support vector machine (SVM) based classifier is applied. The main concentration is provided for parameter optimization and feature selection for SVM with no reduction in the SVM classifier accuracy by the integration of PSO in the retrieval procedure. An analysis is made with the existing models to showcase the betterment of the DLTerQEPs with PSO-SVM model. On testing the presented model on the benchmark dataset, superior results are attained in terms of precision and recall.