Machine Learning Algorithms to Detect Subclinical Keratoconus: Systematic Review

Abstract

Background: Keratoconus is a disorder characterized by progressive thinning and distortion of the cornea. If detected at an early stage, corneal collagen cross-linking can prevent disease progression and further visual loss. Although advanced forms are easily detected, reliable identification of subclinical disease can be problematic. Several different machine learning algorithms have been used to improve the detection of subclinical keratoconus based on the analysis of multiple types of clinical measures, such as corneal imaging, aberrometry, or biomechanical measurements.

Objective: The aim of this study is to survey and critically evaluate the literature on the algorithmic detection of subclinical keratoconus and equivalent definitions.

Methods: For this systematic review, we performed a structured search of the following databases: MEDLINE, Embase, and Web of Science and Cochrane Library from January 1, 2010, to October 31, 2020. We included all full-text studies that have used algorithms for the detection of subclinical keratoconus and excluded studies that did not perform validation. This systematic review followed the PRISMA (Preferred Reporting Items for Systematic Reviews and Meta-Analyses) recommendations.

Results: We compared the measured parameters and the design of the machine learning algorithms reported in 26 papers that met the inclusion criteria. All salient information required for detailed comparison, including diagnostic criteria, demographic data, sample size, acquisition system, validation details, parameter inputs, machine learning algorithm, and key results are reported in this study.

Conclusions: Machine learning has the potential to improve the detection of subclinical keratoconus or early keratoconus in routine ophthalmic practice. Currently, there is no consensus regarding the corneal parameters that should be included for assessment and the optimal design for the machine learning algorithm. We have identified avenues for further research to improve early detection and stratification of patients for early treatment to prevent disease progression.

Keywords: artificial intelligence; cornea; corneal disease; corneal imaging; corneal tomography; decision support systems; keratoconus; keratometry; machine learning; subclinical.

Figure 1

Filtering steps taken to accept or exclude studies in the systematic review.

Figure 2

Organizational diagram of relevant data types reported to be used for the detection of subclinical keratoconus.

Figure 3

Organizational diagram of relevant machine learning algorithms used for the detection of subclinical keratoconus.

Figure 4

Schematic diagram illustrating the 4 basic corneal parameters that can be measured using corneal imaging. (A) pachymetry. (B) displacement: distance between the apex of the cornea and the point of minimum thickness. (C) and (D) represent 2 methods of calculating the best-fit sphere (BFS). In (C) the BFS is fitted to both the normal peripheral posterior surface (blue) and the abnormal anterior protrusion of the central posterior surface (green). In (D) the BFS is fitted to only the normal peripheral posterior surface (blue) excluding the abnormal central posterior surface (green), leading to a larger relative elevation than in (C). (E) the smallest radius of curve of the astigmatic corneal surface corresponds to the largest refractive power (Kmax) and the largest radius of curve corresponds to the smallest refractive power (Kmin). CCT: central corneal thickness.

Figure 5

Heat maps of a subclinical keratoconus eye derived from Scheimpflug corneal imaging using the Pentacam HR device. The axial/sagittal map (A) depicts the curvature of the anterior corneal surface in dioptres and shows mild inferior steepening, while the pachymetry map (C) shows thinning in the same region. The front and back elevation maps (B and D, respectively) show a moderate increase in inferior elevation. BFS: best-fit sphere; OS: left eye.