Multispectral Retinal Image Analysis (MRIA) for the Quantification of Macular Pigment

Antonio Calcagni, Iain Styles, Andrew Palmer, Elzbieta Claridge, Yuan Shen, Hannah Bartlet, Frank Eperjesi, Jonathan Gibson

Research output: Chapter in Book/Report/Conference proceedingConference contribution


Purpose: The aim of this study was to develop a method for quantification of (macular pigment (MP) from multispectral retinal images.

Methods: An imaging system was developed to rapidly acquire images of the macula at 6 wavelengths (507nm, 525nm, 552nm, 585nm, 596nm, 611nm). Spatial parametric maps of MP concentration were derived using computational methods based on a physics-based model of light interaction with the retinal tissues. The main advances made in this part of the project were the use of a computationally efficient Gaussian Process Emulation to replace the Monte Carlo simulation previously used; and statistical (Bayesian) formulation of the image analysis method, to account for noise and uncertainty. Quantitative parameters describing the distribution of MP from the parametric maps included the magnitude of the MP peak and a measure of its spread. These parameters were correlated with age, disease state, and a reference method [MP optical density (MPOD) from heterochromatic flicker photometry, HFP]. Ninety subjects took part in the study, 30 from each of the three groups: (1) Aged under 50 without AMD; (2) Aged 50 and over without AMD; (3) Aged 50 and over with AMD.

Results: Mean values for MP peak concentrations derived from MRIA were 0.47 [Standard Deviation (SD)=0.15], 0.50 (SD=0.16) and 0.51 (SD=0.14) for groups 1, 2 and 3 respectively, whereas those for HFP MPOD were 0.38 (SD=0.15), 0.44 (SD=0.16) and 0.53 (SD=0.17) respectively. The key findings of this project were: 1. Distribution of MP obtained from MRIA was in agreement with previous studies. 2. MRIA and HFP measurements were not correlated. 3. The correlation between age and MP levels was weak as measured by both techniques, in agreement with published data. 4. No statistically significant correlation was found between MP and the AMD group.

Conclusions: The quantity and distribution of MP shown in MRIA parametric maps was found to be in agreement with previous studies. No significant correlations were found between MRIA and HFP measurements of MP, possibly because the two techniques measure different quantities: pigment concentration (MRIA) and total optical density in the macula (HFP). In contrast with MRIA, HFP measurements are potentially affected by the variability of other pigments in, and adjacent to, the macula (hemoglobins, melanin). Further research is necessary to establish the factors that affect each method.
Original languageEnglish
Title of host publicationARVO 2013 Annual Meeting Abstracts
PublisherAssociation for Research in Vision and Ophthalmology
Number of pages1
Publication statusPublished - 2013


Dive into the research topics of 'Multispectral Retinal Image Analysis (MRIA) for the Quantification of Macular Pigment'. Together they form a unique fingerprint.

Cite this