In Vivo Study of Retinal Transmission Function in Different Sections of the Choroidal Structure Using Multispectral Imaging

Abstract

Purpose.: The confounding factors of retinal transmission function are unclear. This study aimed to investigate retinal transmission function by evaluating the morphologic features of the retina and choroid in different section layers using multispectral imaging (MSI) in healthy eyes.

Methods.: Multispectral imaging was performed on 122 healthy eyes. In each subject, 11 images at 550 to 850 nm were collected. The exposed region of choroidal vessels in the images was divided into three regions and evaluated by four grades. The relationships between the axial length (AL)/age/sex and exposure grade were analyzed.

Results.: The 122 subjects were 20- to 77-years old with AL ranging from 22.23 to 31.41 mm. The choroidal vessels of 56% of subjects were initially exposed in 590-nm wavelength imaging. The exposure grade of choroidal vessels was positively correlated with the AL (P < 0.05) in subjects who were 20 to 40 years old. Among the participants whose AL was between 22.00 and 25.00 mm, the exposure grade was positively correlated with age (P < 0.05). This correlation was most significant at the wavelength of 590 nm (P < 0.001). Additionally, the AL/age values at 590 nm were significantly different among grade groups (P < 0.05). No statistical relationship was found between sex and exposure grade (P > 0.05).

Conclusions.: This study investigated the exposure grade of choroidal vessels as an indicator of the transmission function of the retina using MSI and provides reference exposure grades based on different AL/age groups of healthy subjects.

Various technologies are used in the examination of the ocular fundus to improve the convenience of ocular disease diagnosis for ophthalmologists. Specifically, visual acuity, visual field, and electrophysiological examination are the essential evaluation methods for measuring retinal functions. Ophthalmic fundus imaging reflects the morphologic features of the retina.^1,2 Fundus fluorescein angiography (FFA) and indocyanine green angiography (ICG) are mainly focused on the structural features of the ocular fundus vessels.^3,4 Fundus autofluorescence (FAF) provides information on the functions of RPE cells and photoreceptors.^5–8 Optical coherence tomography (OCT) has clearly elucidated the histologic morphology of the retina and choroid.^9,10 

Multispectral imaging (MSI) is a noninvasive technology initially used in clinical application.^11,12 Multispectral imaging generates a series of spectral slices throughout the entire depth of the retina and choroid using an extensive range of monochromatic light sources, enabling the view of different sectional planes of the retinal and choroidal structures en face.^13–16 This approach provides a new method with which to study the retinal and choroidal structures as well as their functions. Multispectral imaging enables the retinal and choroidal transmission function to be evaluated based on the differential light absorption (DLA), which is important for better understanding and diagnosis of ocular fundus diseases as well as enabling quantitative measurement. However, there is neither a well-developed method for interpreting and evaluating MSI results nor any data from healthy subjects using MSI examination as normal references. 

To acquire a better understanding of this new fundus examination method using MSI in evaluating retinal and choroidal transmission function, related confounding factors, such as age, sex, and axial length (AL), should be evaluated. Furthermore, to promote the use of MSI in clinical application, a rapid and simple assessment, especially a quantitative assessment, is needed to interpret the examination results. This prospective case series study was designed to perform MSI examination of eligible subjects with no ocular diseases except for refractive errors. The relationships between AL/age/sex and the exposure of choroidal vessels, which reflected the retinal transmission function, were analyzed in this study. 

Methods

Subjects

This present study was a prospective observational analysis of 122 eyes from 122 healthy subjects selected from among 200 Beijing residents who were randomly recruited from December 2013 to January 2015. All subjects underwent complete ocular examinations in the Department of Ophthalmology (Peking University People's Hospital, Beijing, China). The examinations were conducted by professional ophthalmologists and optometrists on both eyes to assure eligibility; these examinations included visual acuity testing, direct retinoscopy, IOP measurement, slit-lamp biomicroscopy, intraocular lens (IOL) Master score (Carl Zeiss Meditec, Jena, Germany), dilated fundus examination, and MSI (Annidis' RHA multispectral digital ophthalmoscopy, Ottawa, Canada). A detailed interview was also conducted to acquire information on each patient's family, medical, and surgery histories. Visual acuity was defined as the best-corrected value. The AL was measured twice per eye with an IOL Master, and the average was used as the final measurement. Aside from presenting with ocular abnormalities related to refractive errors, the included subjects lacked other pathologic features, including cataract, glaucoma, AMD, and other retinal degeneration diseases, intraocular surgery or laser therapy, and systemic diseases. The corrected visual acuity was required to reach 0.8 in eyes with normal AL. If both of the subject's eyes met the inclusion criteria, one eye was randomly selected for inclusion in the study in accordance with the alternating assignment. The entire research procedure followed the principles described in the Declaration of Helsinki for research involving human subjects, and written informed consent was acquired from all the involved participants. This project was approved by the institutional review board of Department of Ophthalmology (Peking University People's Hospital, Beijing, China). All examination results were recorded for the proposed analysis in the subsequent study. 

MSI Measurement Procedures

Multispectral imaging scanning was performed using Annidis' RHA multispectral digital ophthalmoscopy and digital fundus camera, which operated at wavelengths of 550, 580, 590, 620, 660, 690, 740, 760, 780, 810, and 850 nm. The operation pattern used in this project was the FullSpectrum mode, which provided up to 3-fold more flashes and wavelengths per eye than the FastScan mode and was more appropriate for the required analysis. Real-time pupil tracking was performed during the examination, and a single target was visible when looking into the lens of the MSI, which centered all the images between the macula and the optic nerve. Eleven images were generated in each of the 11 wavelengths (Fig. 1). 

Figure 1

 

Multispectral imaging images at different wavelengths. Multispectral imaging images at the wavelengths of 550, 580, 590, 620, 660, 690, 740, 760, 780, 810, and 850 nm are shown.

Morphologic Description and Analysis of Choroidal Vessels

Three professional observers (XL, LH, and YB) experienced in analyzing MSI images were invited to evaluate the morphologic features of choroidal vasculature in each slice. Majority determination was performed when there was disagreement. Image segmentation was implemented (Fig. 2). The image was divided into three regions, denoted R1, R2, and R3. R3 is a circle whose center is located at the fovea with the same diameter as the optic disc. R1 is an elliptical area around the optic disc, with the radius of the minor axis equaling the distance from the center of the optic disk to the edge of R3 and the diameter of the major axis equivalent to the interval between the top and bottom edges of the image. This area defines the region around the optic disk, in which the earliest exposure of choroidal vessels was frequently detected. The remainder area of the image is labeled R2. 

Figure 2

 

The method of MSI image segmentation. The image is divided into three regions, denoted R1, R2, and R3. R3 is a circle whose center is located at the fovea with the same diameter as the optic disc. R1 is an elliptical area around the optic disc, with the radius of the minor axis equal to the distance from the center of optic disk to the edge of R3 and the diameter of the major axis equivalent to the interval between the top and bottom edges of the image. This area defines the region around the optic disk, in which the earliest exposure of choroidal vessels was frequently detected. The remaining area of the image is labeled R2.

Four grades (grade 0, grade 1, grade 2, and grade 3) were used to represent the condition of the choroidal vessels exposed in these three regions according to the sequence of the exposure region as the wavelength increased (Fig. 3). Grade 0 represents no exposure of choroidal vessels. Grade 1 represents choroidal vessels exposed in R1. Grade 2 represents choroidal vessels exposed in R2. Grade 3 represents choroidal vessels exposed in R3. 

Figure 3

 

Exposure grading method using MSI. (A) Grade 0: no exposure of choroidal vessels. (B) Grade 1: choroidal vessels exposed in R1. (C) Grade 2: choroidal vessels exposed in R2. (D) Grade 3: choroidal vessels exposed in R3. The white arrow represents exposure of choroidal vessels.

Measurement of Test–Retest Reliability

The sample size calculation was based on the exposure grade at 590 nm. A sample size of 28 subjects achieved 90% power to detect a true κ value of 0.85 in a test of H0: κ equal to 0.40 versus H1: κ < > 0.40 when there were four categories with frequencies equal to 0.35, 0.51, 0.05, and 0.09 (the distribution of sample subjects at a 590 nm wavelength). This power calculation was based on a significance level of 0.05. 

We tested the exposure grade of 34 eyes for test–retest reliability. All 34 eyes involved were retested within 1 year after the initial check and were required to satisfy the condition that no eye diseases occurred and no obvious changes of refractive status/visual acuity were detected. The retest followed the same protocol as the initial test but was performed by a different technician. 

Statistical Analysis

SPSS software (version 16.0 for Windows; SPSS, Inc., Chicago, IL, USA) was used to perform all statistical analyses. The Mann-Whitney U test was used to assess the difference in the grade variables among the different AL/age groups. Spearman rank correlation analysis was used to analyze the correlation between the exposure grades and AL/age. One-way ANOVA was used to assess the difference in the AL/age variables among different grade groups at 590 nm. The McNemar-Bowker test was used to examine the reliability between two measurements for each wavelength. The degree of reliability was described with κ values. P values less than 0.05 were considered to be statistically significant. 

Results

Characteristics of the Subjects

Two hundred subjects were recruited in this study. After screening, 122 subjects were included. The 122 subjects had a median age of 36 years (range, 20–77 years) and a median AL of 24.90 mm (range, 22.23–31.41 mm). Of the 122 subjects, 40 males (32.79%) and 82 females (67.21%) were analyzed. 

Characteristics of Choroidal Vessels Exposure in MSI

Exposure of choroidal vessels first took place around the optic disc, particularly the inferior area of the optic disc. With the increase in the wavelength, choroidal vessels were gradually exposed in the temporal regions of the macula, and they finally emerged within the macular region. The wavelength at which choroidal vessels were initially exposed is shown in Figure 4. Among the 11 cases whose choroidal vasculature was exposed at wavelengths shorter than 590 nm, 55% had an AL longer than 28.00 mm, and 45% were older than 60 years old. All subjects whose choroidal vessels were first exposed at a wavelength longer than 620 nm were younger than 60 years old and had an AL less than 25.00 mm. 

Figure 4

 

The wavelength at which choroidal vessels were initially exposed. (A) The number of patients in whom the choroidal vessels were initially exposed at each wavelength. (B) Choroidal vessels initially exposed at 590 nm. (C) Choroidal vessels initially exposed before 590 nm. (D) Choroidal vessels initially exposed after 620 nm. The white arrow represents the initial exposure of choroidal vessels.

Analysis of the Relationship Between the AL and Exposure Grade

To eliminate the confounding factor of age, subjects between 20- and 40-years old were selected for analysis of the relationship between the AL and exposure grade, and there was no correlation between age and AL in these subjects (r = 0.163, P = 0.172). As shown in Figure 5 and Table 1, the AL had a noticeable positive correlation with the exposure grade of the choroidal vasculature at each wavelength. Significant differences in the exposure grades were detected between either two of the three AL groups (≤25.00, 25.01–30.00, >30.00 mm) within the wavelength range of 590 to 690 nm or between the less than or equal to 25.00 mm and greater than 25.00 mm groups within the wavelength of 740 to 850 nm (Fig. 6; Table 2). 

Figure 5

 

Correlation between the AL and exposure grade at each wavelength (scatterplot). For subjects between 20- and 40-years old, the AL had a positive correlation with the exposure grade of the choroidal vasculature at each wavelength. Spots in each panel represent the AL of each patient with different exposure grades.

Table 1

 

Correlation Between AL/Age and Exposure Grade at Each Wavelength

 

Figure 6

 

Exposure grades for different AL groups at each wavelength (the median). Subjects between 20 and 40 years old were divided into three groups according to the AL (≤25.00, 25.01–30.00, >30.00 mm). Significant differences in the exposure grades were detected between either two of the three AL groups within the wavelength range of 590 to 690 nm or between the ≤25.00 mm and >25.00 mm groups within the wavelength range of 740 to 850 nm.

Table 2

 

Differences in the Exposure Grade Among Different AL Groups

 

Analysis of the Relationship Between the Age and Exposure Grade

Subjects with an AL shorter than 25.00 mm were selected to analyze the relationship between the age and exposure grade because there was no correlation between age and AL in these subjects (r = −0.268, P = 0.060). As shown in Figure 7 and Table 1, age was positively correlated with the exposure grade of the choroidal vasculature at each wavelength. Significant differences in the exposure grades were detected between the younger than 60 years old and the 60 years and older groups within 590 to 810 nm (Fig. 8; Table 3). 

Figure 7

 

Correlation between the age and exposure grade at each wavelength (scatterplot). For subjects with an AL shorter than 25.00 mm, age was positively correlated with the exposure grade of the choroidal vasculature at each wavelength. Spots in each panel represent the age of each patient with different exposure grades.

Figure 8

 

Exposure grades for different age groups at each wavelength (the median). Subjects with an AL shorter than 25.00 mm were divided into three groups in accordance with age (60–79, 40–59, 20–39 years). Significant differences in the exposure grades were detected between the less than 60- and greater than 60-year groups within 590 to 810 nm. No significant difference in the exposure grades was detected between the 20- to 39- and 40- to 59-year groups.

Table 3

 

Differences in the Exposure Grade Among Different Age Groups

 

Analysis of AL/Age in Different Grade Groups at 590 nm

At 590 nm, there were significant differences in the AL (χ² = 66.125, P < 0.001)/age (χ² = 11.274, P < 0.001) among the four exposure grades. Because there was no significant difference in the AL (P = 0.477)/age (P = 0.856) between the grades 1 and 2 groups, these two grades were merged into group 2. Significant differences in AL/age were detected between the two grade groups at 590 nm (Table 4). Figure 9 provides reference values of AL/age based on different exposure grade groups of healthy subjects. 

Table 4

 

Axial Length/Age in Different Grade Groups at 590 nm

 

Figure 9

 

Axial length (A) and age (B) in the different grade groups at 590 nm. At 590 nm, there were significant differences in the AL (χ² = 66.125, P < 0.001)/age (χ² = 11.274, P < 0.001) among the four exposure grades. There was no significant difference in the AL (P = 0.477)/age (P = 0.856) between the grade 1 and grade 2 groups; these two grades were merged into group 2. Significant differences in the AL/age were detected between the two grade groups at 590 nm. **P < 0.001; *P < 0.05.

Analysis of the Relationship Between Sex and AL, Age, and Exposure Grade

There was no significant relationship between sex and the following three variables: the AL (Z = −0.078, P = 0.393), age (Z = 0.160, P = 0.078), and exposure grade at different wavelengths (P > 0.05). 

The Results of Test–Retest Reliability

There was no significant difference between the exposure grades of the two measurements at 590 nm (P = 0.317), and the κ value was 0.954. The differences between two measurements were all not significant at other wavelengths; the κ values were all equal or surpassed 0.900 except at the 620 nm wavelength (Table 5). 

Table 5

 

The Results of Test–Retest Reliability (N = 34)

 

Discussion

This study describes the exposure features of choroidal vessels shown in MSI images detected at different wavelengths and presents an innovative grading method according to the exposure region, which can be regarded as a semiquantitative indicator for the transmission function of the retina. The measurement of exposure regions of choroidal vessels at different wavelengths was originally conducted in healthy subjects. We also propose possible physiological factors that may be correlated with retinal transmission function, such as AL and age in healthy subjects. 

We found that AL and age were the two main factors that were positively correlated with the exposure grade of the choroidal vasculature (Figs. 5, 7; Table 1). The exposure grade in different AL/age groups was significantly different at each wavelength. Specifically, subjects with an AL that was longer than 25.00 mm had significantly larger exposure regions of the choroidal vessels than those with an AL of 22.00 to 25.00 mm at each wavelength. In addition, from 590 to 690 nm, different exposure regions could be measured between subjects with ALs longer and shorter than 30.00 mm (Fig. 6; Table 2). Subjects older than 60 years had significantly larger exposure regions of the choroidal vessels than younger subjects at each wavelength (Fig. 8; Table 3). 

This research also showed that the initial exposure of the choroidal vessels occurred at 590 nm in as many as 56% of participants. After synthesizing the correlation between the exposure grade and axial length as well as between the exposure grade and age, we found that the correlation noted above was most significant at 590 nm (P < 0.001) compared with other wavelengths (Table 1). The grades at this wavelength were also significantly different among AL/age groups (P < 0.05; Tables 2, 3). We also further analyzed the values of AL/age in different grade groups at 590 nm and observed significant differences in the AL/age among the four exposure grades (Table 4). With regard to the special statistical features of the exposure grade at 590 nm, we propose that the exposure grading results at this wavelength can be used as an indicator to evaluate retinal transmission function, which would improve the convenience for analyzing MSI images in clinical application. To provide references for future studies, referential values of AL/age were calculated based on different exposure grade groups of healthy subjects (Fig. 9). Figures 10 and 11 showed examples of different exposure grades in subjects with distinct AL/age at 590 nm. 

Figure 10

 

Representative images of the different exposure grades in subjects with different AL at 590 nm. (A–C) Multispectral imaging images at 590 nm from three subjects who belonged to the same age group (20–40 years old) but different AL groups. The exposure grades of choroidal vessels were different among these subjects. Details of these subjects are as follows: (A) grade 0, AL 23.96 mm; age 29 years. (B) Grade 1, AL 26.92 mm; age 37 years. (C) Grade 3, AL 31.41 mm; age 39 years.

Figure 11

 

Representative images of different exposure grades in subjects of different ages at 590 nm. (A–C) Multispectral imaging images at 590 nm from three subjects who belonged to the same AL group (below 25 mm) but different age groups. The exposure grades of choroidal vessels were different among these subjects. Details of these subjects are as follows: (A) grade 0, age 23 years; AL 24.72 mm. (B) Grade 2, age 50 years; AL 24.55 mm. (C) Grade 3, age 77 years; AL 23.49 mm.

The statistical results shown above raise the following question: why are there different exposure grades of choroidal vessels in variant AL/age subjects, or, in other words, different levels of retinal transmission function? The transmission function of the tissue depended on the DLA. Within the wavelength range of 550 to 850 nm, ocular fundus tissues that absorbed light were blood (hemoglobin), macular pigment, and melanin.¹⁷ Any morphologic or functional change in the above factors could affect the DLA as well as the transmission function of the ocular fundus. Recent studies attempted to investigate ocular fundus diseases using MSI.¹⁴ Balaskas et al.¹³ demonstrated that eyes with neovascular AMD had lower levels of DLA than eyes with early age-related maculopathy. The authors posited that AMD reduced the tissue oxygen saturation in retinal vessels, which was thought to be an essential factor influencing the DLA.¹³ Denniss et al.¹⁵ observed the spectral absorption properties of the neuroretinal rim using MSI and speculated that the DLA was related to the oxygen saturation in vessels. 

Our study demonstrated that choroidal vessels were exposed in the macular region at relatively longer wavelengths; in other words, the DLA of the macular region is much higher than in other parts of the retina. We speculated that the mechanism might have a relationship with the special morphologic features of the macular region. Although the macular region lacked retinal vessels at the fovea, choroidal vessels in this area were prominently thicker than in other regions.^18,19 Because the exposed choroidal vessels in MSI images were mainly large choroidal vessels, thicker medium choroidal vessels, and choriocapillaris layer in the macular region may have offset the influence of a lack of retinal vessels in this area on the DLA. Additionally, retinal pigment epithelial (RPE) cells in the macular region had distinct features compared with those in other retinal regions. Retinal pigment epithelial cells were intensely distributed and contained more melanin in the macular area.²⁰ Therefore, a larger amount of melanin may have been a critical factor leading to a high DLA in the macular region, which is represented as a delayed exposure of the choroidal vessels compared with other regions. Previous studies demonstrated that RPE melanin showed a trend of decreasing content with aging,²⁰ whereas in our research, the choroidal vascular exposure region increased with aging. Consequently, we speculated that our grading results of choroidal vascular exposure regions using MSI may reflect the variation of melanin in RPE. This grading method may provide greater opportunities to detect diseases related to RPE melanin. However, this speculation requires more clinical study and need further discussion in the future. 

In our research, we performed MSI examinations on patients with a few types of fundus diseases and evaluated the results using our grading method. Figures 12 and 13 show MSI images of two cases. These examinations indicated the potential value of our grading method in clinical application. 

Figure 12

 

Multispectral imaging images and color fundus photograph for a patient with autoimmune retinopathy in the left eye. This was a patient with autoimmune retinopathy in the left eye (A2–F2) and a healthy right eye (A1–F1). The patient was younger than 60-years old, and the axial lengths of both eyes were less than 25 mm. (A–E) Multispectral imaging images at the wavelength of 580 to 690 nm. These images show that different exposure regions of choroidal vessels between the left eye and the right eye began to emerge at the wavelength of 590 nm. In addition, after grading the choroidal vascular exposure regions using the method proposed in our study, the grade variance between the two eyes was most significant at 590 nm (OD, grade 2; OS, grade 1). In contrast, there were no obvious differences in exposure regions between the two eyes examined by color fundus photograph (F1, F2).

Figure 13

 

Multispectral imaging images for a patient with retinitis pigmentosa and a healthy subject (right eye). (A1–A8) Multispectral imaging images at 550 to 760 nm of a healthy subject who was 30-years old with an axial length of 24.17 mm. (B1–B8) Multispectral imaging images at the same wavelength of a patient with retinitis pigmentosa who was 34-years old with an axial length of 23.87 mm. This patient had a greater exposure grade than the healthy subject at the same wavelength. The images also show that choroidal vessels were first exposed in the macular region, which followed a different exposure sequence from the “around the optic disc-macular temporal region-macular region” sequence in healthy subjects.

Based on our data, we graded the choroidal vascular exposure level at different wavelengths using MSI according to the exposure regions, which can be regarded as a semiquantitative method to evaluate the transmission function of the retina. For the same patient, the analysis results from different time points and concluded by different technicians were consistent, which proved this grading method to be highly reliable. The advantage of this grading method is that it is easy to learn and apply and can be quickly used in clinical application to obtain a rapid evaluation result. However, the limitation of our study is that it describes only a semiquantitative method to evaluate MSI images. We propose that computer image processing can be used to enable fully quantitative analyses in future research with spectral correction of crystalline lens and pseudophakic based on previous studies.^21,22 

In conclusion, this study used exposure of the choroidal vessels as an indicator of the retinal transmission function and presented a grading method according to exposure regions to perform semiquantitative analysis on healthy subjects using MSI. We found that choroidal vascular exposure grades were positively correlated with AL and age. Additionally, we provide normal reference values of grades in different AL/age groups for future clinical application. We also demonstrated that the exposure grades at the wavelength of 590 nm could be potentially useful in distinguishing normal and abnormal retinal transmission functions. However, MSI is still in its initial stage of clinical use. Many contrast studies using this technique and research on the mechanisms of different transmission features are needed. 

Acknowledgments

The authors thank Xin Ying, Ye Tao, and Xiulan Xu, technicians for measuring IOL Master and MSI in the present study. 

Supported by the National Basic Research Program of China (973 Program, 2011CB510200; Beijing, China) and the Beijing Municipal Science & Technology Commission (No. Z131107002213127, Beijing, China). 

Disclosure: S. Li, None; L. Huang, None; Y. Bai, None; Y. Cheng, None; J. Tian, None; S. Wang, None; Y. Sun, None; K. Wang, None; F. Wang, None; Q. Zhang, None; Q. Meng, None; Y. Qi, None; Y. Yu, None; X. Li, None