Comparative analysis of subjective and objective clinical pterygium grading based on its redness appearance

• Main Authors: Hilmi, Mohd Radzi, Che Azemin, Mohd Zulfaezal, Mohd Kamal, Khairidzan
• Format: Conference or Workshop Item
• Language: English; English
• Published: ICVS2015 Organizing Committee 2015
• Subjects: RE Ophthalmology
• Online Access: http://irep.iium.edu.my/72286/7/IMG-1770.JPG; http://irep.iium.edu.my/72286/8/72286_Comparative%20analysis%20of%20subjective.pdf; http://irep.iium.edu.my/72286/; http://www.vision.riec.tohoku.ac.jp/ICVS2015/committee.html

Pterygium is wing-shaped, vascular and fleshy growth which encroaches from bulbar conjunctiva and progresses towards central cornea. Pterygium has been established as one of the factors which cause induced-corneal astigmatism, alteration of anterior corneal parameter, and changes in corneal topography. However, lack of research address the clinical agreement of subjective clinical grading based on its redness appearance, severity and properties of the tissue. Thus, this research aimed to assess both subjective and objective clinical agreement based on its redness appearance and mimic it with human color perception. In this research work, 68 pterygium fibrovascular images were captured in a standardized magnification and illumination. All images were stored in JPEG format and randomize sorted. The images were compared and graded based on reference images described in previous study by two observers. The time interval between intra and inter-session were set at least a month apart. Similar sets of images were used to mimic human color perception. Mimicking human color perception has commonly employed RGB color space, which is inadequate. When the CIELab features were extracted in various color space, it showed better representation of human perception (correlation coefficient = 0.68) compared to using vesselness filter (correlation coefficient = 0.43). This research work also found that for the subjective clinical grading, both intra-observer intraclass correlation (0.957, 95% CI: 0.930 – 0.973) (P = 0.001) and inter-observer (0.937, 95% CI: 0.898 – 0.961) (P = 0.001) shows excellent agreement respectively. In conclusion, this research work found that intra and inter- observer correlation were excellent, which proved that the individual variability is minimal in normal color perception. In addition, this research work also found that the CIELab can serve as a basis for future work on to automate pterygium clinical grading based on tissue redness which may reduce clinician bias and set a new standard for clinical pterygium assessment.