Unsupervised band selection of medical hyperspectral images guided by data gravitation and weak correlation.

• An automated band selection framework for band clustering, outlier noisy band removal, and intercluster ranking was proposed based on the CCE principle. • Data gravitation was defined to simulate the interrelationships between bands. Based on it, the most representative band is selected from each cluster to ensure sufficient information and internal correlations of the selected band. To the best of our knowledge, this is the first time that data gravitation theory has been applied to band selection. • An entropy-containing similarity matrix was constructed, and a weak correlation intracluster band-sorting strategy was developed, which effectively reduces redundancy while ensuring band information. The final ranking was implemented using a customised S-shaped strategy. • Experiments on human brain MHSIs demonstrated that DGWCR achieves superior performance compared to several state-of-the-art band selection methods and deep learning-based methods. Medical hyperspectral images (MHSIs) are used for a contact-free examination of patients without harmful radiation. However, high-dimensionality images contain large amounts of data that are sparsely distributed in a high-dimensional space, which leads to the "curse of dimensionality" (called Hughes' phenomenon) and increases the complexity and cost of data processing and storage. Hence, there is a need for spectral dimensionality reduction before the clinical application of MHSIs. Some dimensionality-reducing strategies have been proposed; however, they distort the data within MHSIs. To compress dimensionality without destroying the original data structure, we propose a method that involves data gravitation and weak correlation-based ranking (DGWCR) for removing bands of noise from MHSIs while clustering signal-containing bands. Band clustering is done by using the connection centre evolution (CCE) algorithm and selecting the most representative bands in each cluster based on the composite force. The bands within the clusters are ranked using the new entropy-containing matrix, and a global ranking of bands is obtained by applying an S-shaped strategy. The source code is available at https://www.github.com/zhangchenglong1116/DGWCR. Upon feeding the reduced-dimensional images into various classifiers, the experimental results demonstrated that the small number of bands selected by the proposed DGWCR consistently achieved higher classification accuracy than the original data. Unlike other reference methods (e.g. the latest deep-learning-based strategies), DGWCR chooses the spectral bands with the least redundancy and greatest discrimination. In this study, we present a method for efficient band selection for MHSIs that alleviates the "curse of dimensionality". Experiments were validated with three MHSIs in the human brain, and they outperformed several other band selection methods, demonstrating the clinical potential of DGWCR. [ABSTRACT FROM AUTHOR]