Methods of measuring the iridocorneal angle in tomographic images of the anterior segment of the eye

INTRODUCTION This paper presents the problem of automatic measurement of the iridocorneal angle in tomographic images of the anterior segment of the eye. It includes the results of the comparison of well-known methods for measuring the iridocorneal angle with new methods, proposed in this paper. All these methods concern tomographic image analysis and processing. MATERIAL AND METHOD In total, approximately 100'000 tomographic images (from about 6'000 patients) were analysed. They were obtained using two devices: SOCT Copernicus (Optopol Tech. SA, Zawiercie, Poland) and Visante OCT (Carl Zeiss Meditec, Inc, Dublin, California, USA). The patients, aged 12 to 78 years with varying degrees of the iridocorneal angle pathology, were from the region of Silesia, Poland. The images were in DICOM or RAW formats and analysed in the software developed by the authors for the purposes of this study. RESULTS The results indicate that the measurement method proposed by the authors, which is based on the calculation of the minimum distance between the iris and the cornea in the adopted area, is the most accurate. For this method sensitivity was 0.88, specificity 0.89 and the area under the Receiver Operating Characteristic curve (AUC) was 0.88. The other known methods for measuring the iridocorneal angle gave worse results, that is, for example, for the measurement of the distance between the iris and the cornea AUC = 0.87, sensitivity = 0.86 and specificity = 0.71. For another well-known method of measuring the iridocorneal angle AUC = 0.77, sensitivity = 0.82 and specificity = 0.61. CONCLUSIONS The study proved that the proposed method of measuring the minimum distance between the iris and the cornea within the adopted area is the most effective in the classification of the iridocorneal angle in patients with a high degree of pathology of all the compared measurement methods based on tomographic images. However, it requires fully automated measurement.