Anterior segment optical coherence tomography (AS-OCT) image analysis methods and applications: A systematic review

https://doi.org/10.1016/j.compbiomed.2022.105471Get rights and content

Highlights

  • Systematic literature review to identify and classify AS-OCT image analysis methods.

  • Summary of key methods, results and findings are provided for each study included in the review.

  • Three main categories were identified: glaucoma assessment, corneal segmentation, and anterior segment biometry.

  • The application of deep learning methods is becoming more popular and provides great performance improvement.

Abstract

Background

Anterior segment optical coherence tomography (AS-OCT) constitutes an important imaging modality to examine the anterior eye, which is commonly used in research and clinical practice. Since its introduction, a range of image analysis methods have been developed to quantify these images using different analysis techniques for various applications. This systematic review aims to provide an in-depth summary and to classify image analysis techniques found in the literature applied to AS-OCT images.

Methods

Scopus and Engineering Village databases were searched to retrieve relevant studies up to and including January 2022. Customized search statements were used along with cross reference and hand search techniques to ensure a complete coverage. Performance metrics were extracted, analyzed, and compared (when possible).

Results

Three main application categories were identified: glaucoma assessment, corneal segmentation, and anterior segment biometry. These three categories constitute 66% of the total studies reported in this review. Studies were also analyzed by year of publication, and since 2019 deep learning methods were favored over traditional programming or machine learning methodologies. Overall, the AS-OCT image analysis field is less developed compared to posterior segment OCT imaging.

Conclusion

This review presents the state of the art in the field of AS-OCT image analysis. It highlights the opportunities for future areas of research, such as the expansion of DL methods and the extension to specific clinical areas that have received limited attention including surgical monitoring, contact lenses, and specific clinical conditions such as keratoconus and corneal lesions.

Introduction

Optical coherence tomography (OCT) is a non-invasive imaging technique capable of visualizing biological samples with micron level resolution [1]. From its first in vivo application, using a time-domain configuration [2], the technology has progressively evolved towards the Fourier domain (FD) [3], which encompasses both swept-source and spectral-domain configurations and allows for faster scanning times, higher axial resolution [4], and an enhanced signal-to-noise ratio [5]. Consequently, FD-OCT configurations have become a standard clinical and research tool in the diagnosis and monitoring of ocular disease [6].

OCT imaging of the anterior segment of the eye (AS-OCT), was introduced in 1994 [7], and has continued to advance with respect to resolution, tissue penetration, and analysis techniques [6]. AS-OCT images provide fundamental information for clinicians and researchers to understand the anatomical features of the anterior segment and related disorders [8]. This imaging technique also facilitates the timely diagnosis of corneal and ocular surface pathologies (e.g. keratoconus, anterior chamber angle closure, ocular surface lesions, and iris tumors [9]), and the ability to monitor the effect of specific treatments [8,10]. Although some clinical information can be extracted from the simple visual inspection of OCT images (i.e. gross morphological changes), the precise objective quantification of ocular structures is desirable to detect subtle anatomical variations over time or in comparison to normative data. Since manual processes used to extract information from AS-OCT images are time-consuming, and liable to suffer from subjective judgement errors [11,12], a range of automatic image analysis techniques have been developed.

Due to their objectivity, repeatability, and accuracy, automated image analysis methods have shown excellent potential for diagnosis (i.e. classification) and segmentation tasks [13,14], and have been applied across a range of medical disciplines [15,16]. Image analysis methods can be broadly categorized into three main branches based on the underlying algorithm: traditional programming (TP), machine learning (ML), and deep learning (DL). Generally, TP utilizes methods with a fixed set of rules for image analysis; while in ML and DL, image analysis rules are autonomously developed during the training phase. In recent years, an increasing number of studies have focused on DL methods, which provide state of the art performance and are comparable to manual analyses by trained human observers [15,17].

To date, the vast majority of the OCT literature in the ophthalmic field has focused on techniques and algorithms used in retinal image analysis [[18], [19], [20], [21], [22], [23]]. While AS-OCT has received less attention, a number of review articles have been published with an emphasis on clinical applications [8,10,24] and specific diseases such as glaucoma [5]. Recently, Ting et al. [25] reviewed anterior segment image analysis, focusing on DL methodologies; however, to the best of our knowledge a comprehensive review of AS-OCT image analysis techniques has not been published. This systematic review provides an in-depth summary of the type of image analysis techniques and methods that have been applied to AS-OCT images.

This systematic review provides an overview of the state of the art in AS-OCT image analysis techniques and is a useful source of information about how automated analyses of AS-OCT images have been utilized in clinical and research applications to date. The review clearly highlights the recent shift in analysis methods towards AI, especially DL methods. This trend is consistent with other medical imaging fields and other OCT imaging applications (i.e. retinal OCT) [18]. This article also provides the reader with information about classification and segmentation techniques including different types of analysis (TP, ML and DL) and their performance. In contrast to previous reviews which have focused on specific ocular pathologies such as glaucoma, or methodologies such as classifiers or deep learning techniques [21,25,26], this review focuses on the technical aspects of various image analysis processes applied to AS-OCT images. Studies included in this review have been categorized based on their main application and the type of algorithms used.

Section snippets

Traditional programming

Traditional programming encompasses a broad range of image analysis techniques with the general approach that the image analysis program follows a predefined set of rules. In AS-OCT images, this approach has been successfully applied to several applications, including corneal segmentation [27], artifact and noise removal [28], and the diagnosis of angle closure [29]. Quality image enhancement techniques, including de-noising, can be effectively addressed with TP methods since the noise can be

Eligibility criteria and search strategy

The search strategy and selection of publications were conducted in accordance with the Preferred Reporting Items for Systematic Reviews and Meta-Analyses (PRISMA) guidelines [40,41]. Literary sources included Scopus and Engineering Village (Compendex, Inspec) databases. Cross reference and hand search techniques were also used to ensure complete coverage of articles. All original peer-reviewed publications including AS-OCT image processing with detailed information of algorithm or analysis

Classification

Twenty-three classification research studies were identified. These were further subdivided according to the main purpose of the study into three different applications: angle closure detection, corneal graft detachment detection, and quality assessment.

Angle closure detection

The clinical sign of closure of the anterior chamber angle is a precursor to elevation of the intraocular pressure and glaucoma. The anterior chamber angle is the route through which circulating aqueous humor enters the trabecular meshwork and

Discussion

This article summarizes the state of the art in AS-OCT image analysis methods, encompassing traditional image processing techniques to the more contemporary deep learning methods. A total of 91 papers were classified and sequentially presented in this review, from which three major applications have been identified: angle closure glaucoma detection, corneal segmentation, and anterior segment biometry. Fig. 6 provides a breakdown of the distribution of the included studies in terms of their main

Limitations and future work

To develop powerful image analysis methods, particularly DL methods, diverse datasets with a large number of annotated images are required. As observed in this review, most studies have a limited number of images [87,12], and the differences between studies datasets (instruments, protocol and imaging acquisition) makes direct comparisons difficult. For example, if we consider two corneal segmentation studies, but only one has used high resolution images [93], a direct comparison between studies

Conclusions

This systematic literature review provides an extensive report of the state of the art in AS-OCT image analysis with a central focus on the technical aspects of image analysis techniques. The studies are classified according to their application and programing paradigm, with three main categories identified (angle closure glaucoma detection, corneal segmentation, and anterior segment biometry) which constitutes almost 70% of all studies included. Although in recent years there has been a shift

References (154)

  • H. Fu

    AGE challenge: angle closure glaucoma evaluation in anterior segment optical coherence tomography

    Med. Image Anal.

    (Dec. 2020)
  • Q. Zhou

    Anterior chamber angle classification in anterior segment optical coherence tomography images using hybrid attention based pyramidal convolutional network

    Biomed. Signal Process Control

    (Jul. 2021)
  • H. Hao

    Angle-closure assessment in anterior segment OCT images via deep learning

    Med. Image Anal.

    (Apr. 2021)
  • S.I. Niwas

    Complex wavelet based quality assessment for AS-OCT images with application to Angle Closure Glaucoma diagnosis

    Comput. Methods Progr. Biomed.

    (2016)
  • L. Wang

    EE-Net: an edge-enhanced deep learning network for jointly identifying corneal micro-layers from optical coherence tomography

    Biomed. Signal Process Control

    (Jan. 2022)
  • J.G. Fujimoto

    Optical coherence tomography: technology and applications

  • D. Huang

    Optical coherence tomography

    Science

    (1991)
  • N. Maeda

    Optical coherence tomography for corneal diseases

    Eye Contact Lens

    (2010)
  • J. Maslin et al.

    Anterior segment imaging in glaucoma: an updated review

    Indian J. Ophthalmol.

    (2015)
  • J. Fujimoto et al.

    The development, commercialization, and impact of optical coherence tomography

    Investig. Ophthalmol. Vis. Sci.

    (2016)
  • J.A. Izatt

    Micrometer-scale resolution imaging of the anterior eye in vivo with optical coherence tomography

    Arch. Ophthalmol.

    (1994)
  • S.B. Han et al.

    Applications of anterior segment optical coherence tomography in cornea and ocular surface diseases

    J. Ophthalmol.

    (2016)
  • J.L.B. Ramos et al.

    Clinical and research applications of anterior segment optical coherence tomography - a review

    Clin. Exp. Ophthalmol.

    (2009)
  • N. Venkateswaran et al.

    Optical coherence tomography for ocular surface and corneal diseases: a review

    Eye Vis.

    (2018)
  • D. Alonso-Caneiro et al.

    Automatic Segmentation of Choroidal Thickness in Optical Coherence Tomography

    (2013)
  • F. LaRocca et al.

    Robust Automatic Segmentation of Corneal Layer Boundaries in SDOCT Images Using Graph Theory and Dynamic Programming

    (2011)
  • D. Alonso-Caneiro et al.

    Morphological Changes in the Conjunctiva, Episclera and Sclera Following Short-Term Miniscleral Contact Lens Wear in Rigid Lens Neophytes

    (2016)
  • S. Kulkarni et al.

    Artificial intelligence in medicine: where are we now?

    Acad. Radiol.

    (2019)
  • A. Esteva

    A guide to deep learning in healthcare

    Nat. Med.

    (2019)
  • N. Porporato et al.

    Recent Advances in Anterior Chamber Angle Imaging

    (2020)
  • H. Rabbani et al.

    A review of algorithms for segmentation of optical coherence tomography from retina

    J. Med. Signals Sens.

    (2013)
  • T. Dsw

    Artificial intelligence and deep learning in ophthalmology

    Br. J. Ophthalmol.

    (Feb. 2019)
  • S. S et al.

    Ophthalmic diagnosis using deep learning with fundus images - a critical review

    Artif. Intell. Med.

    (Jan. 2020)
  • S. Sengupta et al.

    Ophthalmic Diagnosis and Deep Learning - A Survey

    (2018)
  • S.B. Wang et al.

    Anterior Segment Optical Coherence Tomography and its Clinical Applications

    (2019)
  • D.S.J. Ting

    Artificial intelligence for anterior segment diseases: emerging applications in ophthalmology

    Br. J. Ophthalmol.

    (2020)
  • A. Singh

    Deep learning classifiers for automated detection of gonioscopic angle closure based on anterior segment OCT images

    Lect. Notes Comput. Sci. (Including Subser. Lect. Notes Artif. Intell. Lect. Notes Bioinformatics)

    (Sep. 2019)
  • J.A. Eichel et al.

    A novel algorithm for extraction of the layers of the cornea

  • D. Williams et al.

    Fast segmentation of anterior segment optical coherence tomography images using graph cut

    Eye Vis

    (2015)
  • H. Fu

    Segmentation and quantification for angle-closure glaucoma assessment in anterior segment oct

    IEEE Trans. Med. Imag.

    (2017)
  • R.Y. Choi et al.

    Introduction to machine learning, neural networks, and deep learning

    Transl. Vis. Sci. Technol.

    (2020)
  • S.I. Niwas

    Reliable feature selection for automated angle closure glaucoma mechanism detection

    J. Med. Syst.

    (2015)
  • Z. Burgansky-Eliash

    Optical coherence tomography machine learning classifiers for glaucoma detection: a preliminary study

    Investig. Ophthalmol. Vis. Sci.

    (2005)
  • M. Treder et al.

    Using deep learning in automated detection of graft detachment in Descemet membrane endothelial keratoplasty: a pilot study

    Cornea

    (2019)
  • J. Ouyang et al.

    Accurate tissue interface segmentation via adversarial pre-segmentation of anterior segment OCT images

    Biomed. Opt Express

    (2019)
  • T. Schlegl et al.

    Unsupervised anomaly detection with generative adversarial networks to guide marker discovery

    (2017)
  • M. Pekala et al.

    OCT segmentation via deep learning: a review of recent work

    (2019)
  • J. Kugelman

    Automatic choroidal segmentation in OCT images using supervised deep learning methods

    Sci. Rep.

    (2019)
  • M.J. Page

    The PRISMA 2020 statement: an updated guideline for reporting systematic reviews

    BMJ

    (Mar. 2021)
  • D. Moher et al.

    Preferred reporting items for systematic reviews and meta-analyses: the PRISMA statement

    BMJ

    (Jul. 2009)
  • Cited by (9)

    • Patch-based CNN for corneal segmentation of AS-OCT images: Effect of the number of classes and image quality upon performance

      2023, Computers in Biology and Medicine
      Citation Excerpt :

      They provide a potentially useful platform for the segmentation of AS-OCT images but have not yet been explored in depth. Given the lack of studies in this area [23], a patch-based CNN architecture has been developed and applied to segment three corneal boundaries present in AS-OCT scans: the anterior surface of the epithelium (EP), the posterior surface of Bowman's layer (BL), and the posterior surface of the endothelium (EN). In addition, the CNN was tested using two different patch classification categories (four (4C) and eight (8C) classes).

    • Deep learning in retinal optical coherence tomography (OCT): A comprehensive survey

      2022, Neurocomputing
      Citation Excerpt :

      A range of review papers have described the broad application of DL techniques to medical image analysis [12–15]. Various publications have also provided reviews of DL methods in ophthalmology, which have broadly described DL classification and segmentation methods, and how these methods can be applied for retinal diseases (diabetic retinopathy, age-related macular degeneration, and glaucoma) and or different imaging modalities (fundus and OCT images) [16,8,17–21]. This paper provides a focused review on the state of the art of OCT image segmentation with an emphasis on DL methods, highlighting the technical aspects of DL methods for semantic segmentation of retinal layers, and anomalies in OCT images.

    View all citing articles on Scopus
    View full text