Diopter Measurement of Human Eye Based on Dual-Focus Swept Source Optical Coherence Tomography

Abstract

Refractive error affects people’s vision and may be the cause of a variety of fundus diseases. Early detection of refractive error and its location are very important to protect vision and prevent the occurrence of pathological symptoms. In this paper, a dual-focus SS-OCT system was developed to obtain whole-eye imaging and eight biometry parameters, which include corneal center thickness, lens center thickness, anterior chamber depth, vitreous chamber depth, corneal anterior and posterior surface curvature, lens anterior and posterior surface curvature. The diopter of each refractive component and the whole eye can be calculated from these parameters. Healthy subjects were measured under different accommodative stimuli using the proposed system and compared with the refractometer. The results show that there is a high consistency between the two in the overall diopter. The advantage of our system is that it cannot only measure the overall refractive deviation, it can also find the location of the deviation, which will be critical for the prevention and treatment of refractive system diseases in humans.

1. Introduction

Refractive error is characterized by the inability of external objects to focus on the retina after bending through the refractive system of the eye in the unadjusted state. It is mainly divided into myopia, hyperopia and astigmatism [1]. Refractive error not only affects people’s vision; more importantly, serious refractive error can also induce a variety of fundus diseases, such as posterior scleral staphyloma, choroidal atrophy, retinal degeneration, macular hole, retinal detachment, and choroidal neovascularization (CNV) [2,3,4]. The eye’s refractive condition depends on corneal curvature (CC), anterior chamber depth (ACD), axial length (AL), and lens refractive power, as well as the coordination between them. Any abnormality of these parameters will lead to refractive error [5,6]. Therefore, the early detection of human refractive status (overall and various parts) is of great significance to protect vision and prevent the development of pathological symptoms.

At present, the most commonly used methods of refractive examination in clinical practice are subjective optometry, shadow optometry, computer automatic optometry and photographic optometry. These optometric methods can only obtain the overall refractive value of the human eye and cannot indicate the location of refractive error. Optical coherence tomography (OCT) offers non-contact and non-invasive imaging with high resolution, making it a valuable tool for objective and accurate image references for ophthalmic disease screening. Owing to these advantages, OCT has become a research hotspot nowadays [7,8,9]. After Izatt et al. applied time domain OCT (TD-OCT) technology to anterior segment imaging [10], OCT technology gradually became an ideal tool for measuring and analyzing the structure of the anterior segment and has been successfully applied to detect abnormalities in the eye, such as cornea, iris, atrial horn and lens, especially the examination and postoperative follow-up for cataract and glaucoma [11,12]. OCT is also widely used in basic research, such as the adjustment of the lens and its related mechanism of presbyopia and myopia [13,14,15].

Fourier-domain OCT (FD-OCT) represented a significant advancement over TD-OCT by enhancing both sensitivity and imaging speed, thus enabling real-time, high-quality visualization of anterior segment structures [16,17]. It represents the latest generation of commercial ophthalmic OCT technology, and several systems have already been developed and extensively adopted in clinical ophthalmology. FD-OCT includes spectral domain OCT (SD-OCT) and swept source OCT (SS-OCT). Wojtkowski and colleagues successfully performed in vivo retinal imaging using SD-OCT [18]. Currently, many commercial SD-OCT systems (Cirrus OCT and REVO NX) have reported excellent repeatability and reproducibility for corneal and epithelial thickness maps, supporting use in refractive screening and postoperative assessment [19,20]. However, given that the distance from the cornea to the anterior lens surface is approximately 4 mm and the depth of the entire anterior segment exceeds 10 mm, the typical imaging range of SD-OCT (3–4 mm) is insufficient for capturing the whole anterior segment. To improve this problem, researchers proposed methods to eliminate mirrors [21,22,23]. Benefiting from the advantages of SS-OCT in imaging depth and sensitivity, ANTERION, CASIA2, and IOLMaster 700 have been developed and applied in clinical ophthalmic, providing highly repeatable measurements of ophthalmic biometric parameters. And many studies confirm excellent repeatability and reasonable agreement with other devices, which have been employed in the diagnosis and treatment of numerous ophthalmic conditions [24,25,26]. SSOCT has also been used to achieve whole-eye imaging recently. Liu et al. demonstrated imaging results of an entire rodent eye using a 1050 nm wavelength SS-OCT system [27]. Grulkowski et al. applied the advantages of vertical-cavity surface-emitting lasers (VCSELs) to develop an ultra-fast human eye retina, a full anterior segment and a full eye three-mode SS-OCT system. Axis parameters were measured, and the measurement results were compared with clinical optical instruments and A-mode ultrasound [28]. Zhong et al. performed a comparative analysis of axial length measurements obtained from SS-OCT and IOL Master, demonstrating a high degree of consistency between the two devices [29]. However, in pursuit of full-eye imaging depth, a traditional single-focus SS-OCT system greatly sacrifices spatial resolution (for example, the system reported by Grulkowski et al. has a lateral resolution of 73 μm and an axial resolution of 12.4 μm). Insufficient resolution seriously affects the detection accuracy of each interface of the human eye.

To address this limitation, multiple-channel [30,31] and multiple-focus OCT [32,33] have been studied. Zhou et al. proposed a dual-channel dual-focus SD-OCT system to achieve full anterior segment imaging and used the system to image and measure structural changes under different accommodative states [34]. Fan et al. developed a dual-band, dual-focus SD-OCT system for simultaneous high-resolution imaging of both the whole anterior segment and retina [35]. However, the system uses two different scanning modes for the two eye structures, making the system extremely complex. Moreover, the resulting image does not cover the full eye. In conclusion, the current ophthalmic OCT systems all have certain defects in whole-eye imaging. It is necessary to design and implement a better OCT system that can accurately measure the biometry parameters of the human eye.

In this paper, a dual-channel dual-focus SS-OCT system was developed and was applied to whole-eye imaging. This system enables high-resolution imaging of the full eye depth by focusing one beam on the anterior segment and the other beam on the fundus. After performing a series of image processing, including image segmentation, surface fitting, etc., on the whole-eye images, eight biometry parameters can be obtained, include corneal center thickness (CCT), corneal anterior surface curvature (CASC), corneal posterior surface curvature (CPSC), anterior chamber depth (ACD), lens center thickness (LCT), lens anterior surface curvature (LASC), lens posterior surface curvature (LPSC) and vitreous chamber depth (VCD). Based on these measurements, a mathematical eye model was constructed, and the refractive power of each refractive component and the whole eye was calculated using the thick-lens formula. The accuracy of the proposed system was first validated with a model eye. Then, the diopters of healthy subjects under accommodative stimuli were measured and compared with results from a computer optometry instrument.

2. Methods

2.1. System

There is a mutually restrictive relationship between the effective imaging depth and spatial resolution in OCT. To address this issue, a dual-focus SS-OCT system was proposed. This system employs a swept-source (MEMS-VCSEL, Thorlabs Inc., Newton, NJ, USA), providing a 1060 nm center wavelength, 100 nm bandwidth, and line scan speed of 200 k/s. The actual longitudinal resolution of the system can reach ~5 μm. A B-scan image contained 300 A-lines, and each A-line contained 7000 points. Figure 1 presents the schematic illustration of a proposed dual-focus SS-OCT system.

The infrared light from the swept-source is separated into two beams by coupler C1. These two beams form two sets of systems (SS-OCT-1, shown by the blue solid line and SS-OCT-2, shown by the red solid line), which can be transmitted coaxially but focused at different depths. Among them, SS-OCT-1 is used for anterior segment imaging. The sample light is reflected by the splitter to the X-Y galvanometer mirror, passes through the focusing objective lens L4 and the dichroiscope (DS), and finally focuses on the anterior surface of the lens. SS-OCT-2 is used for fundus imaging. The sample light passes through two lenses (L3, L4) into parallel light and then enters the pupil. In this process, it passes through the splitter and merges with the sample light of SS-OCT-1 to achieve coaxial transmission. Finally, it is reflected to the human eye by the dichroic mirror (DS) and focused near the retina with the help of the cornea and lens. The incident power of each sample arm is approximately 2 mW, complying with international standards (ANSI Z136.1/IEC 60825-1) [36,37]. The backscattered signal reflected by the human eye enters the corresponding fiber coupler and interferes with the corresponding reference light. The OCT interferograms were detected using two high-speed dual-balanced photodetectors (PDB480C-AC, 1.6 GHz bandwidth, Thorlabs Inc., Newton, NJ, USA) and a super data sampling digitizer card (National Instruments PXIe-5162, 5 GHz sampling rate, National Instruments, Austin, TX, USA). A high-speed oscilloscope card with two channels was used for data acquisition. The sensitivity of each channel in the system is approximately 92 dB. Finally, the two images are stitched together to obtain a full-eye image. To ensure the stitched image reflects the true axial length, we first calibrated the system using a model eye before performing measurements on real human eyes. Accommodative stimuli from 0 D to 6 D were generated using a 1.5-mm-high letter ‘E’ as a visual target mounted on a fixation board. The target was secured to a sliding rail aligned parallel to the subject’s visual axis. Human eyes gazing at a distant target through DS enables the ciliary muscle to relax to the greatest extent and avoid the interference caused by lens adjustment. The focal length of lens L3 and L4 is 100 mm, which can achieve a lateral resolution of 32 μm.

2.2. Refraction Correction, Boundary Detection and Curvature Calculation

Since light is refracted when it enters the eye, a certain deformation occurs between the OCT image and the real human eye structure [38]. In order to obtain the real biometry parameters of human eye, it is necessary to perform refraction correction processing. The average refractive index of air and human eyes is about 1 and 1.33, respectively. The refraction is primarily generated at the cornea–air curved interface. We take the anterior corneal surface as a starting point and map the pixels in the original image I to the image B that can reflect the real tissue structure.

The anterior corneal surface boundary was extracted first. The previously proposed “improved Canny operator” method was used to extract the boundaries. In brief, a gradient template was first used to enhance the image boundary, followed by “non-maximum suppression” along the A-line direction to identify boundary peaks. Most of the peak points were found near the corneal edge and served as seed points for boundary searching. Then, the double threshold method was employed to refine and connect the corneal boundary. We fit the boundary with a polynomial. The anterior corneal surface can be described by the curve equation f(j), where j represents the horizontal ordinate of the image.

In OCT images, the incident light coincides with the coordinate axis z. The incident angle at point j is determined by the surface normal direction with respect to the z-axis and can be equivalently expressed as the angle formed between the tangent line and the x-axis (i.e., the slope). The incident and refracted angles are:

$$A_{in}\left(j\right)=360\times f\prime \left(j\right)/2\pi , j=1,2,\dots ,n$$

(1)

$$A_{out}\left(j\right)=\arcsin \left[\sin \left(A_{in}\left(j\right)\right)/n_{21}\right]$$

(2)

where f′(j) is the slope of the curve at point j, and A_in(j) and A_out(j) are the incident and refracted angles, respectively. n₂₁ represents the ratio of the refractive indices between medium 2 and medium 1. Since the refractive index of air is close to 1, n₂₁ can be regarded as the refractive index of human ocular medium. The coordinate position of each pixel after reconstruction is:

$$y_B\left(\mathrm{z},j\right)=f\left(j\right)+\cos \left[A_{in}\left(j\right)-A_{out}\left(j\right)\right]\cdot \left[z-f\left(j\right)\right]/n_{21}$$

(3)

$$x_B\left(\mathrm{z},j\right)=j-\tan \left[A_{in}\left(j\right)-A_{out}\left(j\right)\right]\cdot \left[y_B\left(z,j\right)-f\left(j\right)\right]$$

(4)

Some pixels in the reconstructed image will overlap or be missing. For overlapping situations, we do average processing, and for vacancies, we do interpolation processing. After refraction correction, the posterior corneal surface, anterior and posterior lens surfaces were extracted using the same method.

Quartic polynomial function was adopted to fit extracted boundaries:

$$\widehat{f}\left(j\right)=a_0+a_{1}j+a_2{j}^2+a_3{j}^3+a_4{j}^4$$

(5)

where a₀, a₁, a₂, and a₄ are the unknown coefficients, and $\widehat{f}\left(j\right)$ is the fitted curve. The least squares method was employed to determine these coefficients by minimizing the sum of squared residuals between the measured data and the fitted values. By solving the matrix formed by its partial derivatives, the optimal polynomial coefficients are obtained, and the fitted polynomial provides the best approximation of the experimental data in the least squares sense.

The curvature radius is mainly used to describe the degree of curvature change at a certain place on the curve. In clinical applications of ophthalmology, the accurate measurement of the curvature radius (RAC, RPC, RAL and RPL) is critical for the refractive power calculation and the eye accommodation exploration. The curvature radius R of the curve at point j can be expressed as:

$$R={\left(1+f\prime {\left(j\right)}^2\right)}^{3/2}/\left|f″\left(j\right)\right|$$

(6)

where f′′(j) represents the second derivative of the curve at point j.

2.3. Calculation of Diopter

The human eye can be regarded as a coaxial spherical optical system formed by different media in geometrical optical point of view. Under paraxial conditions, the single spherical refractive power is F = (n₁ − n₀)/r. Where n₁ and n₀ represent the refractive indices of the incident and refractive media, and r represents the curvature radius of the surface. In the refractive power calculation of the double spherical refraction system, it can be expressed by the thick lens diopter formula:

$$F=F1+F2-F1\cdot F2\cdot \left(d/n\right)$$

(7)

where n represents the refractive index of the lens, d represents lens thickness, and F1 and F2 represent the refractive power of the lens anterior and posterior surface, respectively.

Considering cornea as a thick lens system, its overall refractive power is not only related to the anterior surface refractive power and the posterior surface refractive power but also the corneal thickness. Then, the overall corneal optical power is expressed:

$$F_C=\left(n_C-n_0\right)/r_{C1}+\left(n_H-n_C\right)/r_{C2}-\left[\left(n_C-n_0\right)/r_{C1}\right]\cdot \left[\left(n_H-n_C\right)/r_{C2}\right]\cdot \left(d_C/n_C\right)$$

(8)

where (n_C − n₀)/r_C1 represents the anterior corneal surface optical power, r_C1 represents the anterior corneal surface radius of curvature, n_C represents the corneal refractive index, and n₀ = 1 represents the air refractive index. (n_H − n_C)/r_C2 represents the posterior corneal surface refractive power, r_C2 represents the posterior corneal surface radius of curvature, and n_H is the aqueous humor refractive index. d_C is the central corneal thickness and the distance between the vertices of the anterior and posterior corneal surfaces. F_C is the total corneal refractive power.

Similarly, the overall refractive power of the lens can be expressed as follows.

$$F_L=\left(n_L-n_H\right)/r_{L1}+\left(n_L-n_V\right)/r_{L2}-\left[\left(n_L-n_H\right)/r_{L1}\right]\cdot \left[\left(n_L-n_V\right)/r_{L2}\right]\cdot \left(d_L/n_L\right)$$

(9)

where (n_L − n_H)/r_L1 represents the optical power of the anterior lens surface refractive power, r_L1 represents the anterior lens surface radius of curvature, n_L is the lens refractive index, (n_L − n_V)/r_L2 is the posterior lens surface refractive power, r_L2 is the posterior lens surface radius of curvature, n_V is the vitreous refractive index. d_L is the central lens thickness and F_L is the total lens optical power.

The whole-eye optical power results from the combined effects of corneal and lens optical systems, and the formula is as follows.

$$F=F_C+F_L-F_C\cdot F_L\cdot \left(d/n\right)$$

(10)

where n represents the aqueous humor refractive index and d represents anterior chamber depth. F_C and F_L are the refractive power of the cornea and lens, respectively. F is the total refractive power of the whole eye.

OCT measurements represent optical path, not geometric length. Therefore, the calculation of axial length requires dividing the measured optical path of each refractive medium (corneal thickness, anterior depth, lens thickness and vitreous thickness) by the respective refractive index and adding them:

$$AL=OP{L}_C/n_C+OP{L}_H/n_H+OP{L}_L/n_L+OP{L}_V/n_V$$

(11)

where AL represents axial length, and OPL_C, OPL_H, OPL_L and OPL_V are the optical path of each medium (corneal thickness, anterior depth, lens thickness and vitreous thickness). n_C, n_H, n_L and n_V are the corresponding refractive indices.

The spherical equivalent refraction (SER) can be expressed as follows.

$$SER=n_{mean}/AL-F$$

(12)

where n_mean represents the mean refractive index of the human eye. AL represents the axial length and F is the eye’s total refractive power. The refractive indices of different ocular media are listed in

Table 1. Refractive index of each medium in human eye.

Medium	Refractive Index
Cornea	1.387
Aqueous humor	1.342
Lens	1.415
Vitreous	1.341
Average	1.3549

.

3. Results

To verify the accuracy of the system, we first used a model eye for imaging (OEMI-7; Ocular Instruments, Inc. Bellevue, DC, USA; Figure 2), with the results presented in Figure 2. The model eye, composed of polymethyl methacrylate (refractive index: 1.48), contained cornea, crystalline lens, and retina, which were filled with saline solution (refractive index: 1.33). The visual axis lengths of the model eye were measured by the proposed system and a commercial measuring instrument LenStar 900 (Haag Streit AG, Koeniz, Switzerland). The comparison results are shown in

Table 2. Comparison of biometry parameters measured by dual-focus SS-OCT and LenStar 900. CCT: corneal center thickness, ACD: anterior chamber depth, LCT: lens center thickness, VCD: vitreous chamber depth, AL: axial length. Unit (mm).

	CCT	ACD	LCT	VCD	AL
Dual-focus SS-OCT	0.51	3.14	3.82	19.62	27.09
Lenstar 900	0.54	3.12	3.99	19.58	27.23

. The results demonstrate that the interfacial distances measured in the model eye by the dual-focus SS-OCT system show excellent agreement with those obtained by Lenstar 900, validating the high accuracy of our proposed system.

Next, we measured the left eye of five subjects. All subjects are the authors of this article. During the experiments, we collected complete 3D OCT data and selected the B-scan images closest to the eye’s optical center for refractive power measurements. For each accommodative stimulus, 50 ocular images were acquired. The 20 most similar images were averaged to generate a new image, which was used for ocular parameter calculation. Figure 3a shows one of the full-eye images obtained by the dual-focus SS-OCT system. In order to perform refraction correction on the whole eye image, we first extracted the boundary of the front surfaces of the cornea. A gradient template ([1, 1, −1, −1] and [−1, −1, 1, 1]) was applied to enhance the ocular surface boundary (Figure 3b). Although the anterior corneal surface is clearly visible, due to the presence of eyelashes, some interference information will appear above it. A method of removing outliers was used to eliminate isolated interference signals, as shown in Figure 3c. Then, the “improved Canny operator” method was applied to the detected boundaries for precise identification and extraction of the contours from Figure 3c. Finally, quartic polynomial fitting is performed on discrete coordinates using the least squares method, and the fitted curve was marked with a solid red line. The final result is shown in Figure 3d.

Refraction correction was performed using the method described in the Methods section, and each pixel in the original human eye image was mapped to its true anatomical position. Figure 4 shows the results after refraction correction. The posterior corneal surface and both anterior and posterior lens surfaces can be extracted. The result is shown in Figure 5. Finally, the obtained curve was used to calculate the curvature of each interface at the position of the visual axis.

Table 3. Eight biometry parameters of five subjects under different accommodative stimulus. CCT: corneal center thickness, LCT: lens center thickness, ACD: anterior chamber depth, VCD: vitreous chamber depth, CASC: corneal anterior surface curvature, CPSC: corneal posterior surface curvature, LASC: lens anterior surface curvature, LPSC: lens posterior surface curvature. Unit (mm).

Accommodative Stimulus	CCT	LCT	ACD	VCD	CASC	CPSC	LASC	LPSC
Subject 1
0D	0.65	4.00	3.56	17.07	7.82	6.64	10.94	6.21
1D	0.66	4.02	3.52	17.08	7.82	6.65	10.13	6.01
2D	0.65	4.05	3.49	17.09	7.83	6.65	9.50	5.72
3D	0.67	4.11	3.47	17.03	7.83	6.64	9.38	5.64
4D	0.64	4.16	3.42	17.06	7.84	6.65	9.10	5.53
5D	0.65	4.18	3.36	17.09	7.84	6.67	8.69	5.28
6D	0.66	4.22	3.35	17.05	7.84	6.66	8.54	5.18
Subject 2
0D	0.62	4.07	3.23	16.96	7.75	6.62	11.35	6.54
1D	0.63	4.10	3.21	16.94	7.77	6.62	10.73	6.26
2D	0.63	4.13	3.18	16.95	7.78	6.62	9.82	6.03
3D	0.61	4.15	3.14	16.98	7.78	6.63	9.10	5.88
4D	0.61	4.19	3.11	16.98	7.78	6.62	8.53	5.54
5D	0.61	4.23	3.08	16.95	7.80	6.64	8.26	5.32
6D	0.61	4.26	3.07	16.95	7.79	6.64	8.01	5.23
Subject 3
0D	0.56	4.39	3.46	17.26	7.86	6.51	12.32	6.12
1D	0.55	4.43	3.36	17.33	7.86	6.52	11.73	6.05
2D	0.56	4.45	3.27	17.39	7.86	6.52	11.24	6.01
3D	0.58	4.47	3.26	17.37	7.87	6.52	10.28	5.96
4D	0.58	4.49	3.19	17.42	7.87	6.52	9.75	5.65
5D	0.57	4.54	3.16	17.40	7.87	6.52	9.04	5.57
6D	0.57	4.57	3.13	17.40	7.88	6.64	8.46	5.22
Subject 4
0D	0.53	3.80	3.33	16.71	7.63	6.29	11.24	6.67
1D	0.53	3.84	3.24	16.76	7.63	6.30	10.55	6.65
2D	0.53	3.88	3.22	16.73	7.64	6.30	10.02	6.63
3D	0.52	3.95	3.10	16.79	7.64	6.30	9.10	6.60
4D	0.52	4.02	3.09	16.74	7.64	6.31	8.18	6.55
5D	0.51	4.06	3.07	16.72	7.66	6.32	7.57	6.47
6D	0.51	4.07	3.06	16.73	7.65	6.32	7.14	6.41
Subject 5
0D	0.60	3.93	3.31	14.90	7.70	6.69	10.59	6.11
1D	0.59	3.96	3.28	14.90	7.71	6.27	10.28	6.01
2D	0.60	3.98	3.27	14.89	7.71	6.10	9.94	5.93
3D	0.59	4.04	3.19	14.91	7.72	6.14	9.32	5.77
4D	0.59	4.09	3.16	14.89	7.72	6.05	8.72	5.61
5D	0.58	4.17	3.11	14.87	7.72	5.99	7.99	5.50
6D	0.58	4.23	3.08	14.84	7.72	6.35	7.36	5.38

shows the eight biometry parameters of five subjects under different accommodative stimulus. The eight biometry parameters were put into the diopter calculation formula, the corneal refractive power, lens refractive power and total diopter were calculated, respectively (

Table 4. Refractive power of each refractive medium and overall diopter of the subjects under different accommodative stimulus.

Accommodative Stimulus	Corneal Refractive Power	Lens Refractive Power	SER
Subject 1
0D	42.195	21.406	−1.586
1D	42.206	22.456	−2.341
2D	42.142	23.701	−3.123
3D	42.137	24.013	−3.267
4D	42.078	24.572	−3.648
5D	42.098	25.711	−4.466
6D	42.091	26.053	−4.656
Subject 2
0D	42.606	20.442	−0.779
1D	42.483	21.436	−1.300
2D	42.420	22.657	−2.110
3D	42.425	23.673	−2.914
4D	42.426	25.144	−3.899
5D	42.310	26.076	−4.413
6D	42.372	26.651	−4.867
Subject 3
0D	41.812	20.730	−1.471
1D	41.819	21.222	−1.930
2D	41.822	21.618	−2.312
3D	41.765	22.416	−2.848
4D	41.765	23.618	−3.708
5D	41.763	24.484	−4.374
6D	41.811	26.049	−5.478
Subject 4
0D	43.033	20.602	−1.517
1D	43.050	21.163	−2.096
2D	43.004	21.664	−2.511
3D	43.016	22.660	−3.502
4D	42.970	23.892	−4.579
5D	42.893	24.960	−5.481
6D	42.931	25.813	−6.298
Subject 5
0D	42.955	22.317	−0.862
1D	42.555	22.831	−0.967
2D	42.336	23.331	−1.219
3D	42.302	24.368	−2.177
4D	42.238	25.503	−3.163
5D	42.172	26.804	−4.311
6D	42.517	28.150	−5.875

SER: Spherical equivalent refraction. Unit (D).

).

Figure 6 shows the comparison of equivalent spherical refraction obtained from the Dual-focus SS-OCT system and the computer refractometer (Grand Seiko). It was found that the results from Subject 2 and Subject 5 were highly consistent between the two groups. For Subject 1, Subject 3, and Subject 4, there was an approximately 1D difference between the results and the computer refractometer when the accommodative stimulus was small. However, when the accommodative stimulus was larger, the results from both groups tended to converge.

4. Discussion

Timely understanding of the degree and location of refractive errors is of great significance for protecting vision and preventing fundus complications. Existing refractometers can only provide the overall equivalent spherical refraction and cannot localize the source of refractive errors. Some scholars have proposed using ophthalmic detection devices to collect ocular biometry and construct a mathematical model to calculate refractive power. A mathematical model of a human eye is established based on its structure and optical properties. It can describe the refractive state, optical characteristics, and imaging features. Gullstrand proposed a mathematical model for the human eye, which was continuously improved by Le Grand and others, forming the widely used Gullstrand-Le Grand human eye model [39,40,41,42]. This model includes curvature radii of the anterior and posterior corneal surfaces, central corneal thickness, curvature radii of the anterior and posterior lens surfaces, central lens thickness, anterior chamber depth, axial length, and refractive indices of ocular components. However, existing techniques for measuring human eye biological parameters, such as A-scan ultrasound, slit-lamp microscopy, Placido disc, Scheimpflug imaging, etc., can only obtain a limited set of biological parameters of the human eye and cannot independently complete the calculation of refractive power [43,44,45,46,47]. If all of the above parameters could be obtained through a single scan, the screening process would be greatly simplified, saving time and reducing the patient’s need for coordination.

In this paper, we propose a concept that uses SS-OCT with full-eye imaging capability to collect human eye parameters, construct a human eye mathematical model, and combine it with the Gaussian thick lens formula for refractive power calculation, to compute overall human eye refractive power and internal structure refractive power. This can serve as a diagnostic basis for refractive errors, enabling the timely detection of refractive abnormalities and allowing for appropriate intervention. Compared to SD-OCT, SS-OCT has been demonstrated distinct advantages, such as faster A-line rates, increased light intensity, improved spectral resolution, and nearly negligible system sensitivity loss. These advantages allow SS-OCT to greatly improve imaging range and penetration depth. However, previous single-focus SS-OCT systems generally improved imaging depth by sacrificing resolution. To ensure both resolution and imaging depth, we designed a dual-channel dual-focus SS-OCT system and employed in full-eye imaging and refractive power calculation. This system can focus on both the anterior segment and the fundus, achieving high-resolution imaging of the full eye’s depth. The system has a line scan speed of 200 k/s and can complete a full-eye 3D data acquisition of size 300 × 300 ×7000 in less than 1 s. The fast detection speed is advantageous for clinical translation.

Our system can quickly determine whether vision issues are caused by abnormal corneal curvature, dysfunctional lens accommodation, or axial length changes. This diagnostic capability provides critical guidance for clinicians to develop personalized treatment plans, such as recommending orthokeratology for corneal abnormalities, vision therapy for lens accommodation disorders, and intensive myopia control for excessive axial elongation. For pediatric refraction assessments, our objective measurement method eliminates reliance on subjective cooperation required in traditional refraction tests, making it particularly suitable for screening infants with low compliance.

From the results of the refractive power calculation, our calculations can well reflect the trend of changes in the eye’s refractive power with varying stimuli. When comparing the results with those measured by the computer refractometer Grand Seiko, we found that for some subjects, there was an approximate 1D difference between the results and the computer refractometer when the accommodative stimulus was small. However, when the accommodative stimulus was larger, the results from both groups tended to converge. This phenomenon primarily results from the combined effects of environmental interference and inherent system limitations. During near stimulation (4–6D), when the eye is in a high accommodative state with pronounced ciliary muscle contraction and stable, substantial lens deformation, our dual-focus SS-OCT system accurately captures accommodative signals, showing strong agreement with autorefractor measurements. However, under distant stimulation (1–3D), when accommodative demand is low and the system is in a relaxed state, subjects become more susceptible to interference from stray light in the experimental environment, such as reflections from dichroic mirrors or shadows from optical components—factors that do not affect traditional autorefractors due to their different measurement principles. Additionally, our system demonstrates reduced sensitivity in detecting subtle lens deformations during low accommodation, where limitations in low-order aberration resolution further amplify measurement errors. In future research, we plan to address these limitations through hardware and software optimizations, including enhancing sensitivity in the primary optical path, implementing deep learning algorithms, and improving device shielding.

The current system has certain limitations. First, approximately 50% optical power loss occurs in backscattered light due to the beam-splitting design, which affects the imaging signal-to-noise ratio to some extent. Second, although the shared scanning path design for both channels simplifies the system configuration, it prevents simultaneous optimization of both focal points for imaging. Another limitation of our system is its lateral resolution of only 32 μm. This specification results from the fundamental trade-off between lateral resolution and depth of focus in optical design. The 32 μm lateral resolution corresponds to a 380 μm depth of focus, which proves marginally adequate for both anterior segment and retinal imaging pathways.

Fortunately, benefiting from the dual-focus strategy, the system in this paper is still able to distinguish between the inner and outer retinal layers that conventional single-focus OCT systems cannot achieve. In many previous OCT-based studies, the total eye’s axial length is defined as the distance from the anterior corneal surface to the retinal pigment epithelium (RPE). The main reason for this is the insufficient resolution of the fundus, which prevents the system from distinguishing the inner retinal layers. However, photoreceptor cells, including rods and cones, are primarily located in the inner retinal layer. These cells convert light into neural signals. Light enters the pupil and focuses on the inner layer of the retina, allowing people to see the scene. Therefore, the distance from the anterior corneal surface to the inner retinal layer represents a more accurate definition of ocular axial length.

The greatest advantage of this work is that it enables not only measurement of the overall refractive power but also precise measurement of each optical medium, thus providing a convenient and effective method for diagnosing refractive error-related diseases. A comprehensive analysis of the obtained parameters can help identify the causes of refractive errors, contributing to the development of vision correction plans. It is a convenient means to gain the geometrical changes and the relative contribution of different ocular structures during accommodation, thereby offering deeper insights into the mechanism of the human eye’s accommodation.

5. Conclusions

This paper presents a refractive power measurement method that differs from traditional refractometry techniques, using dual-focus SS-OCT to obtain full-eye images, accurately measuring human eye parameters, and providing a comprehensive analysis of these parameters to assess the degree and cause of refractive errors. Both the model eye and human eye experiments have demonstrated the high accuracy of the system presented in this paper. The work in this paper plays an important role in preventing vision deterioration or fundus damage caused by refractive errors and investigating the accommodation of the eye.