Related Experiment Video
Updated: Oct 4, 2025
13:44
Detection of Architectural Distortion in Prior Mammograms via Analysis of Oriented Patterns
Published on: August 30, 2013
43.1K
Local bit-plane neighbour dissimilarity pattern in non-subsampled shearlet transform domain for bio-medical image
Hilly Gohain Baruah1, Vijay Kumar Nath1, Deepika Hazarika1
1Department of Electronics and Communication Engineering, School of Engineering, Tezpur University, Napaam, Tezpur, Assam 784028, India.
Mathematical Biosciences and Engineering : MBE
|February 9, 2022
Summary
A new descriptor, non-subsampled shearlet transform local bit-plane neighbour dissimilarity pattern (NSST-LBNDP), enhances biomedical image retrieval. This method improves accuracy and recall compared to existing techniques.
Area of Science:
- Medical Imaging
- Computer Vision
- Signal Processing
Background:
- Biomedical image retrieval is crucial for medical diagnosis and research.
- Existing descriptors often lack efficiency in capturing fine details and directional information.
- The need for robust and accurate image retrieval methods in medical applications is growing.
Purpose of the Study:
- To introduce a novel descriptor, the non-subsampled shearlet transform local bit-plane neighbour dissimilarity pattern (NSST-LBNDP), for improved biomedical image retrieval.
- To leverage the strengths of NSST for translational invariance and directional sensitivity.
- To enhance feature extraction by incorporating bit-plane slicing and local neighbour dissimilarity patterns.
Main Methods:
- Decomposition of input images using non-subsampled shearlet transform (NSST).
- Introduction of non-linearity via local energy features of NSST coefficients, normalized to 8-bit values.
- Decomposition of normalized subband features into bit-plane slices and encoding using neighbour dissimilarity relationships.
Main Results:
- The proposed NSST-LBNDP descriptor demonstrated superior performance in experiments.
- Evaluated on computed tomography (CT) and magnetic resonance imaging (MRI) datasets.
- Achieved higher average retrieval precision (ARP) and average retrieval recall (ARR) compared to recent descriptors.
Conclusions:
- NSST-LBNDP offers a significant advancement in biomedical image retrieval.
- The descriptor effectively captures fine details and anisotropic information, leading to enhanced retrieval accuracy.
- The method shows promise for clinical applications requiring efficient and precise image retrieval.
