Emerging diagnostic imaging technologies promise better detection of pre-clinical corneal ectasia
RV Display from OD with moderate clinical ectasia detected
The growing range of corneal imaging tools continues to evolve, revealing information that enhances the ability for the diagnosis of early forms of keratoconus, as well as for staging the disease and planning individualised treatment strategies for helping such patients, said Renato Ambrósio Jr MD, PhD, Rio de Janeiro, Brazil.
First of all, this is important to consider the nomenclature for such instrumentation that goes from ultrasound central corneal thickness (CCT) and Placido disk-based topography up to Scheimpflug 3D tomography.
While topography characterises the front surface, tomography depicts both front and back surfaces of the cornea, also providing thickness mapping. In addition, high-frequency ultrasound (HF-US) and optical coherence tomography (OCT) devices provide a layered tomographic outline, with the ability for epithelial mapping. Furthermore, corneal biomechanical parameters are also characterised by novel technologies, such as the ultra-high-speed Scheimpflug camera that monitors corneal deformation during air-puff non-contact tonometry, Prof Ambrósio told the 8th EuCornea Congress in Lisbon, Portugal.
His current understanding is that the refractive surgeon should aim for characterising the susceptibility of the cornea for biomechanical failure and ectasia progression when screening candidates for laser vision correction procedures. This cannot be determined only by classic CCT and corneal topography, while such procedures do provide relevant information. In fact, it requires an enhanced characterisation of the cornea with conscious use of the diagnostic information. Nevertheless, the challenge is to efficiently interpret the enormous amount of data from a variety of instruments.
A collaborative research was established between Instituto de Olhos Renato Ambrósio (Rio de Janeiro, Brazil) and Vinci Eye Clinic with Prof Paolo Vinciguerra MD and Riccardo Vinciguerra MD (Milano, Italy) and also Prof Cynthia Roberts PhD (Columbus, Ohio USA). The Corvis Biomechanical Index (CBI), available at the Vinciguerra Screening Report (VSR) of the Corvis ST (Oculus Optikgeräte GmbH, Wetzlar, Germany), was developed based on corneal thickness profile and deformation parameters, providing very high accuracy to separate normal (n=480) from clinical ectatic corneas (n=276) with an area under the curve (AUC) of 0.983 (cut-off value of 0.48, 95.7% specificity and 97.5% sensitivity).
Interestingly, the BAD-D (Belin-Ambrósio deviation index), available on the Oculus Pentacam, had in the same series an area under the curve of 0.999 (cut-off value of 1.95, 98.9% specificity and 99.2% sensitivity). Nevertheless, while these data confirm the high accuracy of isolated corneal tomographic and biomechanical assessments for detecting clinical ectasia, there was evidence for the need to go beyond to detect abnormalities in asymptomatic cases with very mild ectatic corneas. This is the case of the eyes with normal topography from patients with clinical ectasia in the other eye.
Ninety-six eyes with normal topography from very asymmetric ectasia cases (VAE-NT) were included in the work for the integration of Scheimpflug imaging from the Pentacam and Corvis ST (Oculus, Wetzlar, Germany). The Tomographic and Biomechanical Index (TBI), available in the ARV (Ambrósio, Roberts and Vinciguerra) display, was developed to integrate tomographic and biomechanical data using artificial intelligence, exceeding the accuracy to detect ectasia. The random forest method with leave-one-out cross-validation (RF/LOOCV) was the best model for the TBI, providing 100% sensitivity and specificity with a cut-off of 0.79 for distinguishing clinical ectasia (n=276) from normal corneas (n=480). The AUC of the TBI for the normal versus the VAE-NT cases (n=94) was 0.985, with a cut-off value of 0.29, providing 90.4% sensitivity and 96% specificity.
Interestingly, studies from Iran, India, Portugal and Germany were done for external validation of the TBI with similar results as the original study.
Corneal anatomical evaluation has further evolved into layered tomography, providing the ability to characterise structures, such as the epithelium, Bowman’s layer and Descemet’s membrane. This capability may help in early diagnosis of keratoconus, Prof Ambrósio said, citing original work developed by Professor Dan Z Reinstein MD, MA (Cantab), FRCOphth, using HF-US, and Professor David Huang MD, PhD, using OCT.
In the future, the inclusion of molecular biology is expected. Nevertheless, the current evolution we have in corneal diagnostic devices represent a true revolution, which is in constant evolution, concluded Prof Ambrósio.
Renato Ambrósio Jr: firstname.lastname@example.org