Modern Trends In Hyperspectral Image Analysis A Review
Alysha
The site is secure.
The ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.
Modern hyperspectral imaging systems produce huge datasets potentially conveying a great abundance of information; such a resource, however, poses many challenges in the analysis and interpretation of these data. Deep learning approaches certainly offer a great variety of opportunities for solving classical imaging tasks and also for approaching new stimulating problems in the spatial-spectral domain. This is fundamental in the driving sector of Remote Sensing where hyperspectral technology was born and has mostly developed, but it is perhaps even more true in the multitude of current and evolving application sectors that involve these imaging technologies. The present review develops on two fronts: on the one hand, it is aimed at domain professionals who want to have an updated overview on how hyperspectral acquisition techniques can combine with deep learning architectures to solve specific tasks in different application fields. On the other hand, we want to target the machine learning and computer vision experts by giving them a picture of how deep learning technologies are applied to hyperspectral data from a multidisciplinary perspective. The presence of these two viewpoints and the inclusion of application fields other than Remote Sensing are the original contributions of this review, which also highlights some potentialities and critical issues related to the observed development trends.
A common feature of many new analytical techniques that allows fast and non-destructive analysis of poorly-water-soluble drug is that they generate a large amount of data with a multivariate character within a short time frame, which in turn highlights the need for advanced data analytical methods in extracting information from the complex data set. The current review critically examines how spectroscopy and imaging techniques can be utilized for fast and non-destructive characterization of solid state poorly water-soluble drug formulations. The first part of the present review describes the basics behind many of the currently used methods including Raman, near infrared (NIR), infrared (IR) spectroscopy and X-ray powder diffractometry in characterizing poorly water soluble drugs. Key emphasis was placed on a critical review of the currently used spectral preprocessing methods, and the influence of selected preprocessing on spectral data sets is exemplified. Further the existing uni- and multivariate spectral data analytical methods in analyzing complex spectral data sets are reviewed, covering estimation of spectral peak moments, peak modeling, variations of Principal Component Analysis (PCA), variations of Partial Least Squares (PLS) analysis and Multivariate Curve Resolution (MCR). The second part of the present review discusses hyperspectral imaging, UV imaging, optical microscopy imaging and process imaging methods suitable for characterization of poorly water-soluble solid state drug formulations. Image analytical techniques suitable for analyzing hyperspectral image data set are described. Further, the application of various image analytical techniques leading to the estimation of nucleation and crystal growth rates from polarized light microscopy is described.
Hyperspectral imaging (HSI) is a promising technology that can provide valuable support for the advancement of the medical field. Bibliometrics can analyze a vast number of publications on both macroscopic and microscopic levels, providing scholars with essential foundations to shape future directions. The purpose of this study is to comprehensively review the existing literature on medical hyperspectral imaging (MHSI). Based on the Web of Science (WOS) database, this study systematically combs through literature using bibliometric methods and visualization software such as VOSviewer and CiteSpace to draw scientific conclusions. The analysis yielded 2,274 articles from 73 countries/regions, involving 7,401 authors, 2,037 institutions, 1,038 journals/conferences, and a total of 7,522 keywords. The field of MHSI is currently in a positive stage of development and has conducted extensive research worldwide. This research encompasses not only HSI technology but also its application to diverse medical research subjects, such as skin, cancer, tumors, etc., covering a wide range of hardware constructions and software algorithms. In addition to advancements in hardware, the future should focus on the development of algorithm standards for specific medical research targets and cultivate medical professionals of managing vast amounts of technical information.
Copyright 2023 Jiang, Ma, Tan, Yang, Jiao and Xu. This is an open-access article distributed under the terms of the Creative Commons Attribution License (CC BY). The use, distribution or reproduction in other forums is permitted, provided the original author(s) and the copyright owner(s) are credited and that the original publication in this journal is cited, in accordance with accepted academic practice. No use, distribution or reproduction is permitted which does not comply with these terms.
Thank you for visiting nature.com. You are using a browser version with limited support for CSS. To obtain the best experience, we recommend you use a more up to date browser (or turn off compatibility mode in Internet Explorer). In the meantime, to ensure continued support, we are displaying the site without styles and JavaScript.
Hyperspectral imaging is a frontier in the field of medical imaging technology. It enables the simultaneous collection of spectroscopic and spatial data. Structural and physiological information encoded in these data can be used to identify and localise typically elusive biomarkers. Studies of retinal hyperspectral imaging have provided novel insights into disease pathophysiology and new ways of non-invasive diagnosis and monitoring of retinal and systemic diseases. This review provides a concise overview of recent advances in retinal hyperspectral imaging.
Hyperspectral imaging has its origins in geospatial sciences [1] but has evolved over several decades, with expanding applications in art conservation [2], food quality and safety control [3, 4], pharmaceuticals [5], agriculture [6], forensics [7, 8] and medicine [9, 10].
Hyperspectral imaging is like conventional photography, however instead of using a single broadband (white) light flash, a series of images is captured across a continuous range of discrete wavelengths of interest. These images are then stacked to yield a three-dimensional data set called a hypercube, which is comprised of two spatial dimensions (x and y) and one spectral dimension (λ) (Fig. 1) [11]. Each location, or pixel in the hypercube has its own spectral signature (reflectance as a function of wavelength) indicative of its composition [12] and can be further interrogated to identify the individual constituents (referred to as endmembers).
A hyperspectral imaging camera (labelled HS) uses narrow bandwidth tuneable light source to illuminate the retina typically in less than one second. The reflected light from the retina is then collected by an image sensor. There are various modes of acquiring a hyperspectral image, but typically different frames are obtained by scanning the source wavelengths to generate a data cube (HS Cube). Each image has both spatial and spectral information and each pixel has a corresponding spectral signature. Images are analysed using deep learning image analysis methods. A convolutional neural network (CNN) is illustrated here. A CNN model is composed of input, hidden and output layers. Hidden layers are fully connected and consist of multiple stacks of convolution, pooling and activation, enabling automated feature extraction, classification and regression. Training a CNN model requires accurate ground-truth data, depicted here as positron emission tomography (PET), optical coherence tomography (OCT) and optical coherence tomography- angiography (OCTA) scans for illustrative purposes. Outputs of this model include disease biomarkers.
Hyperspectral image data cubes are complex and require processing prior to analysis [11, 13]. Each pixel in a hyperspectral image consists of a mixture of reflectance spectra of endmembers [13, 18]. Spectral unmixing is a commonly used analysis technique. Unmixing aims to isolate the individual endmembers present at a given location, and to quantify the abundance of each endmember in each pixel at that location. In the past, mathematical models that account for unmixing have been constructed and used with some success [16, 17]. More recently, approaches utilising artificial intelligence have been popular because of their ability to process and analyse large amounts of highly complex data [16, 19, 20]. A comprehensive discussion of hyperspectral imaging data analysis techniques is beyond the scope of this review and can be found elsewhere [21, 22].
Retinal hypoxia is a key component of numerous ocular diseases including diabetic retinopathy [32], retinal vein occlusion [15, 29], and some forms of glaucoma [30, 31]. Whilst fluorescein angiography (FA) and OCT-angiography (OCTA) are used as surrogate markers of hypoxia, high resolution and non-invasive retinal oximetry methods have the potential to significantly improve the detection, prognostication and management of these diseases [1].
The characteristic absorption spectra of oxy- and deoxy-haemoglobin serve as the basis of pulse oximetry and non-contact retinal oximetry. Oxymap T1 (Oxymap, Reykjavk, Iceland), the reference device for retinal oximetry, uses dual wavelength imaging to estimate oxygen saturation values for higher order vessels [32, 33]. Hyperspectral imaging has several potential advantages over dual wavelength oximetry. The use of many spectral channels can facilitate noise reduction and improved signal detection [30]. This can enable the development of more complex oximetry models and potentially enable pixel-level oximetry across the image field [17]. Studies of a range of retinal diseases indicate that hyperspectral imaging may enable high resolution retinal oximetry to improve our understanding of disease pathophysiology [15, 31, 34,35,36,37,38].
c80f0f1006