Sparse Representation-Based Discriminative Metric Learning for Brain MRI Image Retrieval.

Guohua Zhou,Tongguang Ni,Bing Lu,Xuelong Hu

doi:10.3389/fnins.2021.829040

Guohua Zhou, Tongguang Ni + Show 2 more

Open Access

https://doi.org/10.3389/fnins.2021.829040

Copy DOI

Abstract

Magnetic resonance imaging (MRI) can have a good diagnostic function for important organs and parts of the body. MRI technology has become a common and important disease detection technology. At the same time, medical imaging data is increasing at an explosive rate. Retrieving similar medical images from a huge database is of great significance to doctors’ auxiliary diagnosis and treatment. In this paper, combining the advantages of sparse representation and metric learning, a sparse representation-based discriminative metric learning (SRDML) approach is proposed for medical image retrieval of brain MRI. The SRDML approach uses a sparse representation framework to learn robust feature representation of brain MRI, and uses metric learning to project new features into the metric space with matching discrimination. In such a metric space, the optimal similarity measure is obtained by using the local constraints of atoms and the pairwise constraints of coding coefficients, so that the distance between similar images is less than the given threshold, and the distance between dissimilar images is greater than another given threshold. The experiments are designed and tested on the brain MRI dataset created by Chang. Experimental results show that the SRDML approach can obtain satisfactory retrieval performance and achieve accurate brain MRI image retrieval.

Highlights

The number of tumor patients has increased rapidly in recent years
This paper proposes sparse representation-based discriminative metric learning (SRDML) approach for brain Magnetic resonance imaging (MRI) image retrieval
A Laplacian matrix L is constructed based on the matrix P, In order to better explain the proposed model, first, we introduce some important terms: (1) a labeled training image set X = [x1, x2, ..., xn] ∈ Rd×n; (2) the testing image set contains images available for the retrieval of the target brain MRI images; (3) each image in testing image set is considered as the query image

Summary

INTRODUCTION

The number of tumor patients has increased rapidly in recent years. Tumors have become one of the most common diseases in the world. This paper proposes sparse representation-based discriminative metric learning (SRDML) approach for brain MRI image retrieval. The advantages of SRDML are as follows: (1) the local information retention term of coding coefficients maintains the semantic correlation and visual similarity of brain MRI images in the projection space. SRDML combines the sparse representation, pairwise constraint of coding coefficients, and locality constraint of atoms together to yield an image retrieval approach. Algorithm 1 SRDML approach Input: Similar image pair subset and dissimilar image pair subset in X, parameters α, β, and γ ; Output: coding coefficient matrix Y, dictionary B, and metric matrix M; 1: Initializing B and Y with the K-SVD algorithm, M with the identity matrix; 2: Calculating graph Laplacian matrix L using Eqs 8 and 9; 3: while t ≤ maximum number of iterations Tmax. N where n represents the number of retrieved images of the same type of brain tumor in the dataset, rank(i) represents the ranking number of the i-th retrieved image of the same type of brain tumor in the search results

EXPERIMENTS

CONCLUSION

Findings

DATA AVAILABILITY STATEMENT