A novel similarity learning method via relative comparison for content-based medical image retrieval

Abstract

Nowadays, the huge volume of medical images represents an enormous challenge towards health-care organizations, as it is often hard for clinicians and researchers to manage, access, and share the image database easily. Content-based medical image retrieval (CBMIR) techniques are employed to facilitate the above process. It is known that a few concrete factors, including visual attributes extracted from images, measures encoding the similarity between images, user interaction, etc. play important roles in determining the retrieval performance. This paper concentrates on the similarity learning problem of CBMIR. A novel similarity learning paradigm is proposed via relative comparison, and a large database composed of 5,000 images is utilized to evaluate the retrieval performance. Extensive experimental results and comprehensive statistical analysis demonstrate the superiority of adopting the newly introduced learning paradigm, compared with several conventional supervised and semi-supervised similarity learning methods, in the presented CBMIR application.

Fig. 1

An illustration of concordant and discordant pairs in KT measure (Eq. 1) for image retrieval

Fig. 2

An illustration of approximating a discrete signum function (in blue) via a continuous hyperbolic tangent function (in red)

Fig. 3

Standard images with grades indicating the severity of nuclear cataract disease within slit-lamp images according to Wisconsin Cataract Grading System

Fig. 4

An illustration of lens structure

Fig. 5

a The influence of different values of learning rate . b The influence of different values of iteration times T

Fig. 6

Examples of retrieving slit-lamp images based on queries with diverse grades via different methods

Fig. 6

Examples of retrieving slit-lamp images based on queries with diverse grades via different methods

Fig. 6

Examples of retrieving slit-lamp images based on queries with diverse grades via different methods

Fig. 6

Examples of retrieving slit-lamp images based on queries with diverse grades via different methods

Fig. 7

Precision–recall curves of methods retrieving medical images depicting various degrees of severity of the nuclear cataract disease (from left to right, up to down: with grades 1, 2, 3, and 4)

Fig. 8

a Averaged precision–recall curves on all experimental results. b Box plot of precision of all experimental results

Fig. 9

Summary of learned weights on different dimensions of the constructed feature space according to Table 2