Alireza Karimi*, Najme Meimani, Reza Razaghi, Seyed Mohammadali Rahmati, Khosrow Jadidi and Mostafa Rostami
Background: Keratoconus is recognized by asymmetrical thinning and bulging of the cornea, resulting a distortion in the surface of the cornea. Keratoconus also alters the biomechanical properties of the cornea, which can be an indicator of the healthy and keratoconus eyes. This study was aimed at employing a combination of clinical data, finite element method (FEM), and artificial neural network (ANN) to establish a novel biomechanical-based diagnostic method for the keratoconus eyes.
Methods: To do that, the clinical-biomechanical parameters of 40 healthy and 40 keratoconus eyes were obtained via the Pentacam and non-contact tonometer (Corvis ST, Oculus Optikgeräte, Wetzlar, Germany) devices. Intraocular pressure (IOP) was measured using a Goldmann applanation tonometer as well as Corvis. According to the geometry of the cornea, the FE model of each cornea was made and the same boundary and loading conditions were applied not only to confirm the FE model in terms of the biomechanical parameters, but also to calculate the amount of von Mises stress in the apex of the cornea. The clinical-biomechanical data of the Corvis along with the von Mises stresses were then incorporated into the ANN algorithm to distinguish the healthy and keratoconus corneas on a basis of the resulted von Mises stresses. The proposed programming code, according to the input data from the Corvis, enabled to predict whether the cornea is keratoconus or not. Finally, to verify the results of the proposed method, 155 individuals were examined.
Results: The clinical and biomechanical results of the Corvis revealed that the healthy corneas have a higher thickness compared to the keratoconus ones. No significant differences were observed among the IOPs, 1st applanation length, and pick distance in the highest concavity. The 2nd applanation length and radius in the highest concavity of the healthy cornea were higher than the keratoconus ones. Conversely, the 1st and 2nd applanation velocities and deformation amplitudes of the keratoconus corneas were higher than the healthy ones. The FE results also showed higher stresses for the healthy corneas compared to the keratoconus ones. The ANN was also well verified since it demonstrated more than 95.5% accuracy on diagnosing the keratoconus eyes.
Conclusions: These findings have implications not only for identifying the keratoconus corneas as an important clinical and surgical tool for eye care professionals, but also for providing both a quantitative and an accurate approach to the problem of understanding the biomechanical nature of keratoconus.
Cornea, Keratoconus, Corvis, Finite element, Artificial Neural Network.
Department of Mechanical Engineering, Ky0ushu University, 744 Motooka, Nishi-ku, Fukuoka 819-0395, Basir Eye Health Research Center, Tehran, Basir Eye Health Research Center, Tehran, Department of Biomedical Engineering, Amirkabir University of Technology, Tehran 15875, Department of Chemistry, Shahid Beheshti University of Medical Sciences, Tehran 19839, Department of Biomedical Engineering, Amirkabir University of Technology, Tehran 15875