Abstract
Purpose :
Anterior segment optical coherence tomography (AS-OCT) is a developing technology finding increasing uses in clinical practice. Swept-source lasers machines have reduced the scanning times, making it possible to obtain high quality large volume scans of the anterior segment. The assessment and documentation of cornea-opacifying lesions such as keratitis, dystrophies and corneal scarring has typically been subjective using slit-lamp biomicroscopy or by photography. Human and technical factors can make obtaining serial photographs under standardized conditions difficult thereby making comparative analyses unreliable. Additionally, this only yields 2-dimensional (2D) information. We present the development of an AS-OCT technique to quantify corneal lesions in 3D to address this unmet need.
Methods :
Five female patients aged between 16 and 37 with corneal lesions received AS-OCTs using the the Casia II whilst attending the Birmingham & Midland Eye Centre. A semi-automated Matlab algorithm was developed to compute the lesion volume. Patient OCT images were deconvolved and high-pass filtered to optimise contrast and enhance edges. The region of interest was selected and lesion segmented using Matlab’s connectivity algorithm to extract the lesion. The final lesion volume in three dimensions was reconstructed in relation to the scanning resolution of the instrument. The algorithm was tested against a manual scar assessment performed by an ophthalmologist.
Results :
Four patients were diagnosed with acute keratitis, of which 1 was herpetic, the other 3 were presumed bacterial, and one chronic post-traumatic corneal scar. Of the 2 patients who had microbiological samples taken, no growth was detected. The lesions demonstrated varying location, morphology and intensity. The Matlab algorithm successfully segmented the corneal lesion in every patient scan. An average lesion volume of 0.052 mm3 was found within the 5 cases, ranging from 0.014 mm3 to 0.088 mm3. These were compared to manually segmented reconstructions. No significant difference (p>0.05) was found between the scar volume computed by the algorithm and the volume manually reconstructed by the experienced ophthalmologist.
Conclusions :
Our analysis protocol enabled 3D quantitative assessment of these lesions with this pilot data set. This tool has the potential to be utilized in a variety of contexts and for corneal pathology.
This abstract was presented at the 2019 ARVO Imaging in the Eye Conference, held in Vancouver, Canada, April 26-27, 2019.