In Vivo Quantification of Anterior and Posterior Chamber Volumes in Mice: Implications for Aqueous Humor Dynamics

Abstract

Purpose: Aqueous humor inflow rate, a key parameter influencing aqueous humor dynamics, is typically measured by fluorophotometry. Analyzing fluorophotometric data depends, inter alia, on the volume of aqueous humor in the anterior chamber but not the posterior chamber. Previous fluorophotometric studies of the aqueous inflow rate in mice have assumed the ratio of anterior:posterior volumes in mice to be similar to those in humans. Our goal was to measure anterior and posterior chamber volumes in mice to facilitate better estimates of aqueous inflow rates.

Methods: We used standard near-infrared (NIR) optical coherence tomography (OCT) and robotic visible-light OCT (vis-OCT) to visualize, reconstruct, and quantify the volumes of the anterior and posterior chambers of the mouse eye in vivo. We used histology and micro-computed tomography (CT) scans to validate relevant landmarks from ex vivo tissues and facilitate in vivo measurement.

Results: Posterior chamber volume is 1.1 times the anterior chamber volume in BALB/cAnNCrl mice, that is, the anterior chamber constitutes about 47% of the total aqueous humor volume, which is very dissimilar to the situation in humans. Anterior chamber volumes in 2-month-old BALB/cAnNCrl and C57BL6/J mice were 1.55 ± 0.36 µL (n = 10) and 2.05 ± 0.25 µL (n = 10), respectively. This implies that previous studies likely overestimated the aqueous inflow rate by approximately twofold.

Conclusions: It is necessary to reassess previously reported estimates of aqueous inflow rates and, thus, aqueous humor dynamics in the mouse. For example, we now estimate that only 0% to 15% of aqueous humor drains via the pressure-independent (unconventional) route, similar to that seen in humans and monkeys.

Aqueous humor dynamics (AHDs) determine intraocular pressure (IOP) and are thus important in understanding ocular physiology and pathophysiology and drug delivery in the anterior segment.^1–4 Mice are widely used to study AHDs, where they show important similarities to humans. Despite the small size of the mouse eye, high-quality measurements are available for a number of important AHD parameters, such as murine IOP and outflow facility.^3,5–8 However, a key parameter that has received less attention in the mouse eye is the aqueous production rate. The most recent measurements by Toris et al.⁹ used a customized fluorophotometer to measure inflow in CD-1 mice; other authors have used similar tracer dilution methods.^8,10,11 

Tracer dilution methods, including fluorophotometry, are based on measuring the rate of loss of the tracer signal in the cornea and anterior chamber and relating this quantity to the aqueous inflow rate. This procedure requires accurate knowledge of the corneal and anterior chamber volumes. In the human eye, nearly 80% of the total aqueous humor volume is in the anterior chamber,¹² which has also been assumed to be the case in the mouse eye. However, this assumption may be incorrect considering the anatomic differences between the mouse and human eye (e.g. the murine eye has a relatively much larger lens). For example, tracer dilution studies in the mouse eye have assumed that the anterior chamber volume is approximately equal to the total aqueous volume determined by aspiration of all aqueous humor (5.1–7.2 µL).^10,11,13 If erroneous, this estimate leads to significant errors in determining mouse aqueous inflow rate. Incomplete knowledge of chamber volumes has other implications; for example, when conducting preclinical studies of agents delivered into the mouse eye intracamerally,^14–16 dosing and dilution effects can be misestimated. 

Due to the small size of the mouse eye, quantifying anterior and posterior chamber volumes is not trivial. For example, in vivo imaging methods, such as magnetic resonance imaging (MRI) and ultrasound, have insufficient resolution in mice. Ex vivo studies of chamber volume are not ideal because tissue handling can deform the globe and lead to incorrect estimates of volumes. To overcome these shortcomings, we adopted an approach based on optical coherence tomography (OCT), which was informed by postmortem studies. OCT has been used to measure the anterior chamber depth but not to reconstruct the 3D anatomy of the anterior and posterior chambers.^17–23 Further, clear boundaries for the posterior chamber cannot be well resolved within a single optical field of view with the OCT beam orientated along the optical axis of the eye, making posterior chamber volume measurement even more challenging. An ideal solution is to obtain a volumetric OCT image of the entire anterior and posterior chambers in vivo and measure chamber volumes from the reconstructed volumetric image. Toward this end, we used a robotic visible-light OCT (vis-OCT) system. Vis-OCT has a higher axial resolution than conventional OCT using near-infrared (NIR) light, with an axial resolution approximately 1.3 microns in tissue,²⁴ allowing us to reconstruct the anterior segment with high resolution in vivo. We validated the accuracy of volume measurements and overall volumetric reconstruction with a 3D-printed phantom and then segmented robotic vis-OCT images, with landmarks validated by other imaging modalities, to obtain chamber volumes in living mice. 

Materials and Methods

Animal Handling

All procedures were approved by the relevant institutions’ Institutional Animal Care and Use Committee and conformed to the Association for Research in Vision and Ophthalmology Statement on Animal Research. All mice used in this study were wild type. 

At Northwestern University, vis-OCT imaging was carried out in ten 2-month-old adult BALB/cAnNCrl albino mice (Charles River Laboratories, Skokie, IL, USA; 5 male mice and 5 female mice). Using albino mice eliminated pigment-induced scattering, providing the best possible visualization of structures posterior to the iris. Body weights of BALB/cAnNCrl mice were not measured. For strain comparison, we also imaged the anterior chambers of ten 2-month-old C57BL/6J (JAX stock number 000664; Jackson Laboratory, Bar Harbor, ME, USA; 5 male mice and 5 female mice) and eight 7-month-old C57BL/6J mice (4 male mice and 4 female mice). Mean body weights for 2-month-old C57BL/6J mice were 22.3 g (male mice) and 19.6 g (female mice). Mean body weights for 7-month-old C57BL/6J mice were 35.5 g (male mice) and 28.0 g (female mice). All mice were housed under 12-hour:12-hour light:dark cycles within the Center for Comparative Medicine at Northwestern University. 

At Columbia University, C57BL/6J mice (JAX stock number 000664, n = 4) were obtained from Jackson Laboratory (Bar Harbor, ME, USA), whereas the DBA/2J mice were wild type mice from a DBA/2JSj substrain that we had separated from the Jackson Laboratory's DBA/2J strain in 2019 (and so are essentially the same genetically, n = 4). Animals were housed under a 14-hour:10-hour alternating light:dark cycle at 21°C, and fed with a 6% fat diet within the Institute of Comparative Medicine animal facility. The mice were euthanized, and their eyes were collected for ex vivo micro-computed tomography (CT) imaging, as described in detail in the Micro-CT Image Acquisition. 

At Duke University, the mice were handled in accordance with an approved protocol (A226-21-11). C57BL/6J mice were purchased from the Jackson Laboratory (JAX stock number 000664, 2 female mice), and CD-1 mice were purchased from Charles River (Charleston, SC; stock number 22, 1 female mouse), bred/housed in clear cages and kept in housing rooms at 21°C on a 12-hour:12-hour light:dark cycle in the Duke animal facility. C57BL/6J mice were used for ex vivo histology, and CD-1 mice were used for conventional OCT imaging. 

Histology and Conventional Near-Infrared OCT Imaging

To conduct ex vivo histological analysis, 3-month-old C57BL/6J mice (n =2) were anesthetized using isoflurane at Duke University. Once they reached a deep plane of anesthesia, the animals were decapitated, and the eyes were carefully enucleated and immersion fixed in 4% PFA + 1% glutaraldehyde at 4°C. The posterior sclera and part of the retina were carefully dissected, after which the remainder of the globe was processed for embedding in Epon using standard approaches. The block was then trimmed, oriented, and sectioned until the sectioning plane reached the approximate center of the eye. Sections were then collected, stained with 1% methylene blue, and examined by light microscopy (Axioplan2; Carl Zeiss MicroImaging, Thornwood, NY, USA). 

To conduct conventional OCT imaging, one 7-week-old CD-1 female mouse was anesthetized with ketamine (100 mg/kg)/xylazine (10 mg/kg) and secured in a custom-made platform. Lubricant eye gel (GenTeal) was applied to the eye to keep the cornea hydrated. Eyes were imaged with an Envisu R2200 high-resolution spectral domain (SD)-OCT system (Bioptigen Inc., Research Triangle Park, NC, USA). The mouse was positioned until the OCT probe faced the center of the cornea, and cross-sectional images spanning from the nasal to temple sides of the globe were recorded. The mouse was euthanized after imaging by decapitalization while it was under anesthesia. 

Micro-CT Image Acquisition

All micro-CT imaging experiments were conducted at Columbia University. Eyes were collected within 15 minutes of euthanasia by cervical dislocation and immersion fixed overnight in 3% paraformaldehyde plus 1% glutaraldehyde in phosphate-buffered saline at 4°C. Whole eyes were stained either with eosin-Y, as previously described,²⁵ and phosphotungstic acid (PTA), or with PTA alone, as described previously,²⁶ and then dehydrated with hexamethyldisilazane. Eyes were scanned in a Bruker SkyScan 2214 multiscale-CT system (Micro Photonics Inc., Allentown, PA, USA) utilizing a tungsten x-ray source at 53 keV and an 11 Mpixel CCD detector. This setup has an achievable voxel size of 120 nm and maximum spatial resolution of 500 nm. A 360-degree scan was acquired with rotation steps of 0.15 degrees and 6 frame averages. Projection images were reconstructed with Bruker NRecon software. Three-dimensional virtual sections for the figures presented were produced with Bruker's CTvox software. Opacity and luminosity were adjusted for each image to show the target structures as appropriate. 

Vis-OCT Image Acquisition

All vis-OCT imaging experiments were conducted at Northwestern University. Before imaging, mice were anesthetized with a ketamine/xylazine cocktail (ketamine = 11.45 mg/mL and xylazine = 1.7 mg/mL, in saline) delivered via intraperitoneal injection (10 mL/kg). During imaging, we maintained the mouse's body temperature using a heating lamp and applied artificial tears to prevent corneal dehydration. 

To obtain high-resolution in vivo anterior segment images, we used an experimental robotic vis-OCT system,²⁷ technical details are given in the Supplementary Figure S1. Due to the location of the posterior chamber within the anterior segment and the angle-dependent backscattering of the lens and borders of the posterior chamber, the boundaries of the posterior chamber are best visualized when the incident OCT beam is normal to the limbus. However, in this configuration, OCT does not have sufficient imaging depth to capture the entire anterior segment within a single volumetric acquisition. To reconstruct the entire anterior segment, we captured 8 volumes (Fig. 1c), each separated by 45 degrees, around the eye positioned by a 6-degree-of-freedom robotic arm (Mecademic, Montreal, Canada; Meca500 Robot Arm). Each scan consisted of 512 B-scans, and each B-scan consisted of 512 A-lines, together acquiring a volume with a lateral area of 2.04 × 2.04 mm² and a depth of 1.56 mm in air. Adjacent volumetric scans had an overlap of approximately 40%. We acquired each volume using a temporal speckle averaging scan pattern, where each B-scan was repeated twice per volume, and we acquired two repeated volumes at each of the eight positions.²⁸ We also processed the interferograms using the optical microangiography algorithm to generate visible-light OCT angiography (vis-OCTA) images.²⁹ 

Figure 1.

 

Experimental setup of vis-OCT imaging. (a) Schematic of robotic vis-OCT. Light from an NKT Photonics laser is filtered by a dichroic mirror (DM), spectral shaping filter (SSF), and bandpass filter (BPF). The output light is coupled into a collimator (CL) and split by a 90:10 fiber coupler (FC). The reference arm includes a polarization controller (PC) and dispersion compensation (DC). Light in the sample arm is scanned by a galvanometer scanning mirror (SM) before being focused by a 25-mm scan lens (SL). The interference signal is split by a 50:50 FC into 2 spectrometers. (b) Schematic cross-section of the mouse eye with the anterior chamber shaded in green and the posterior chamber in blue. (c) Eight vis-OCT volumes, with scan planes perpendicular to the incident vis-OCT beam shaded in blue, are acquired around the eye.

Vis-OCT Volume Fusion

To combine the eight volumes, we developed an algorithm based on point cloud registration methods commonly used in LIDAR.^30,31 Briefly, we represented each volume as a point cloud and registered the point clouds together. When two sets of point clouds had sufficient overlap, we found the transformation matrix that minimized the distance between the overlapping point clouds²⁷ (Fig. 2). 

Figure 2.

 

Overview of anterior segment reconstruction. (a) The outer surface of the globe for each sub-volume is extracted and (b) landmark points are identified along the top surface of each sub-volume. The locations of several landmark points are shown by the red dots, whose spatial locations correspond to the red dots in a. (c) The landmark points are registered between adjacent volumes and used to align them. (d) After alignment, all sub-volumes are mapped into a common spatial reference frame.

To transform each sub-volume into a common spatial reference frame, we represented the outer surface of the eye for each volume as a point cloud. To determine the spatial position of the outer surface, we applied a threshold to each vis-OCT volume, kept the largest connected binarized object, and found the outer surface of the remaining binarized object (see Fig. 2a). For vis-OCT volumes acquired adjacent to each other, we identified common reference points (landmarks) in the overlapping regions of the point clouds. Specifically, from vis-OCTA images, we manually identified 15 blood vessel branch points as landmarks (Fig. 2b). We then used the M-estimator sample consensus (MSAC) algorithm to obtain an initial estimate of the rigid transformation matrix that would match the spatial position of the landmark points between two adjacent volumes in three dimensions³² (see Fig. 2c). We aligned all volumes within a common coordinate system and obtained the pixel-wise intensity of that combined volume (see Fig. 2d), as described in the Supplementary Material. 

Anterior and Posterior Chamber Segmentation and Volumetric Rendering

We defined the anterior chamber as the space bounded by the cornea, anterior iris, and anterior lens and the posterior chamber as the space between the lens, posterior iris, and anterior hyaloid membrane³³ (Fig. 1b). To generate volumetric representations of the anterior and posterior chambers, we used the Segment Anything Model (SAM; Meta AI, New York, NY, USA³⁴) to segment individual B-scan images and combine the segmented results to form a volumetric representation of the chambers. SAM is a general segmentation model that allows zero-shot segmentation of various images.³⁴ With appropriate fine-tuning steps, SAM has previously been shown to be compatible with medical images, including applications in CT, MRI, and OCT.³⁵ 

To segment the anterior and posterior chambers, we first obtained an initial mask of the chambers using SAM. During this step, we manually marked points every 50 B-scans within and outside the chambers, as SAM's zero-shot segmentation requires users to identify inlier and outlier regions with point inputs. We interpolated the position of the inlier and outlier points with a third-order polynomial to approximate their locations across the B-scans and fed these inputs to the model checkpoint based on the Vision Transformer-Huge (ViT-H) image encoder.³⁶ Following initial segmentation, we manually checked and fine-tuned the segmentations using the artificial intelligence (AI) segmentation website Biodock.³⁷ With fine-tuning, the output neural network generated high-quality segmentations of the chambers without user input. 

Figure 3 illustrates the process of reconstructing the posterior and anterior chambers. Each vis-OCT volume consists of 512 B-scans, from which we segmented the posterior chamber (the blue-shaded region in Fig. 3a) for each B-scan. Then, we merged the segmented posterior chamber for each of the eight vis-OCT volumes (Fig. 3b). Finally, we applied the rigid transformation matrices obtained during montaging to map the posterior chamber for each vis-OCT volume into a common reference frame to reconstruct the entire volume, as shown in Figure 3c. To reconstruct the anterior chamber, we used the fully reconstructed montaged anterior segment volume generated using the methodology described in the Vis-OCT Volume Fusion (Fig. 3d). Next, we used the trained SAM to segment the anterior chamber in each digital cross-sectional image of the montaged volume along the x-z plane, as highlighted by the green shaded region in Figure 3e. Finally, we merged the segmented anterior chamber in each digital cross-sectional image to form the circumlimbal volume of the anterior chamber (Fig. 3f). We measured the volumes of each chamber in each eye after segmentation based on the vis-OCT voxel size and the number of voxels. We determined the voxel size in the axial direction based on the parameters of the spectrometers³⁸ and the lateral direction by calibrating to a gridded sample with known grid sizes (R1L3S3P; Thorlabs, Newton, NJ, USA). The voxel size in the axial direction was 1.53 µm in air, corresponding to 1.15 µm in aqueous, assuming a refractive index of 1.335.³⁹ The voxel size in the lateral direction was 3.98 µm for the vis-OCT volumes. 

Figure 3.

 

Posterior chamber and anterior chamber reconstruction processes. (a) B-scans were segmented using SAM, with the posterior chamber shaded in blue. (b) After segmenting all B-scans, we generated a volumetric representation of the posterior chamber for each volume. A transformation matrix mapped each volume into a common coordinate system and the union of (c) segmented volumes was taken to be the posterior chamber. (d) The transformation matrices were also used to map the OCT structural data into a common reference frame, where (e) the anterior segment of each cross-section was segmented using SAM. (f) The union of the segmented B-scans was used to generate a volumetric representation of the anterior chamber.

After volume reconstruction, we measured pupil size from the vis-OCT volume by calculating the distance between the edges of the lens from the cross-sectional B-scan where the lens was the largest. We ensured that we chose the cross-section where the lens was largest by measuring the distance between lens edges for multiple B-scans (B-scan spacing = 139.4 µm) and selecting the cross-section where this distance was maximum. 

Posterior Chamber Volume Correction

Due to its position within the eye and its tenuous structure, the anterior hyaloid membrane – needed to define the posterior boundary of the posterior chamber – could not reliably be visualized by vis-OCT, even in BALB/cAnNCrl mice. Thus, the posterior boundary of the posterior chamber segmentation generated by SAM was a curve connecting the lens with the ciliary body. However, the ciliary body lies anterior to the hyaloid membrane (Supplementary Fig. S2), so the anterior hyaloid membrane location generated by SAM was incorrect and led to an underestimation of the posterior chamber volume. To correct this underestimation, we approximated the posterior boundary of the posterior chamber (anterior hyaloid membrane) by the equator of the eye,⁴⁰ based on our micro-CT images. As described in detail in the Supplemental Materials, we thus approximated the anterior hyaloid membrane location by the plane that passed through the center of the lens, and that was normal to the optical axis of the eye (Fig. 4). 

Figure 4.

 

Workflow for identifying the posterior boundary of posterior chamber. (a) Chamber segmentations were used to obtain the posterior boundary of the anterior chamber (green) and the interior boundary of the posterior chamber (blue). (b) These boundaries were taken to coincide with the anterior surface of the lens. (c) An ellipsoid (red) was fit to the lens’ upper boundaries. The center of the ellipsoid and the optical axis of the eye were used to (d) generate a plane at the equator of the eye, approximating the location of the anterior hyaloid membrane. (e) The posterior border of the reconstructed posterior chamber in the montaged B-scans is (f) updated with the estimated position of the anterior hyaloid membrane.

Volume Measurement Validation

We validated our vis-OCT volume measurement by designing a cavity to mimic the posterior chamber of the mouse eye (Fig. 5a). To create the cavity, we 3D-printed a phantom consisting of a hemisphere with a cavity inside. The designed cavity consisted of the unfilled space within the hemispheric phantom. For clarity, we defined the printed phantom as the positive phantom and the cavity within the positive phantom as the negative phantom, with the negative phantom shown in Figure 5a. We designed the positive phantom using SolidWorks and 3D-printed the positive phantom using our homemade micro continuous liquid interface production (µCLIP) system⁴¹ with a photocurable clear resin, which was made by mixing 98.85 wt.% Poly(ethylene glycol) diacrylate (PEGDA; Sigma-Aldrich Inc., St. Louis, MO, USA) as a low-viscosity monomer, 1 wt.% Phenylbis (2,4,6-trimethylbenzoyl), phosphine oxide (Irgacure 819; Sigma-Aldrich Inc., St. Louis, MO, USA) as photoinitiator, and 0.15 wt.% Avobenzone (Tokyo Chemical Industry Co., Tokyo, Japan) as a UV absorber. After printing, we washed the positive phantom with isopropyl alcohol to remove any remaining resin and post-cured it under UV light. Finally, we filled positive contrast resin into the 3D-printed hollow positive phantom structure and placed the positive phantom under UV light to fully cure the contrast resin. The negative phantom consists of the space filled by the positive contrast resin within the positive phantom. The positive contrast resin consisted of 93.8 wt.% 2-Hydroxyethyl methacrylate (HEMA; Sigma-Aldrich Inc., St. Louis, MO, USA), 3 wt.% Ethylene glycol dimethacrylate (EGDEA; Sigma-Aldrich Inc., St. Louis, MO, USA), 2.2 wt.% Irgacure 819, and 1% Intralipid (Sigma Aldrich, St. Louis, MO, USA). We visualized the geometry of the positive phantom by scanning electron microscopy (SEM) and found the volume of the negative phantom by filling the cavity with de-ionized water and measuring the mass of the structure before and after water filling. We slowly added de-ionized water to the positive phantom until it was slightly overfilled. Then, we removed any excess water with a cleanroom wipe so that the water surface was at the same height as the edge of the positive phantom. If the water surface and the top of the positive phantom were not visually flush or an air bubble was observed, we cleaned and dried the positive phantom and repeated the de-ionized water filling. To conduct SEM imaging, we deposited a thin layer of 10-nm Au/Pd onto the positive phantom with sputter coating (Denton Vacuum, Moorestown, NJ, USA) and acquired images using an EPIC SEM FEI Quanta 650 (FEI, Hillsboro, OR, USA). We measured the negative phantom volume from eight vis-OCT volumes using the methodology described in the Anterior and Posterior Chamber Segmentation and Volumetric Rendering. 

Figure 5.

 

Validation of reconstruction volume accuracy. (a) Design of the cavity within our phantom. (b) Reconstructed cavity volume using vis-OCT after montaging. (c) SEM image of the 3D-printed phantom. (d) The cavity volumes obtained by water weighing and from the OCT reconstruction agreed to within one percent, with error bars representing the 95% confidence intervals (*P < 0.05; **P < 0.01, ***P < 0.001, ****P < 0.0001).

Statistical Analysis

All statistical computations were carried out using GraphPad Prism 10.1.0 (Boston MA). We compared the phantom volumes measured using water weight versus OCT with an unpaired t-test. Similarly, we compared anterior and posterior chamber volumes, anterior chamber volume, and pupil size between the BALB/cAnNCrl and C57BL/6J mice, and anterior chamber volume and pupil size between the C57BL/6J mice of different ages using an unpaired t-test. Finally, we used Pearson correlation to assess the relationship between pupil size and anterior chamber volume. All numerical data are presented as mean ± standard deviation (SD). The error bars on all plots represent 95% confidence intervals. We used P < 0.05 as the threshold for statistical significance. 

Results

Histology, Conventional OCT Imaging, and Micro-CT Imaging

Although significant distortion was evident in histologic images (see Supplementary Fig. S1), a morphometric analysis based on measuring anterior and posterior chamber boundaries and rotating the images through 180 degrees to compute the corresponding volumes suggested that anterior and posterior chamber volumes were approximately equal. To investigate the situation in vivo, we thus carried out conventional NIR OCT imaging and a similar morphometric analysis, rotating the OCT images through 180 degrees and estimating anterior and posterior chamber volumes, again finding that < 50% of total aqueous humor volume resided within the anterior chamber (data not shown). 

The above results were suggestive but not definitive due to tissue deformation occurring during histologic processing and poor visualization of the posterior chamber structures by conventional OCT imaging. We, therefore, undertook micro-CT imaging of postmortem eyes designed to more clearly identify the location of posterior chamber structures, particularly the anterior hyaloid membrane. We observed (see Supplementary Fig. S2) that the anterior hyaloid membrane, which we took as the posterior margin of the posterior chamber, was approximately located at the equator of the eye. Importantly, we observed that the position of the anterior hyaloid membrane was similar in both 13-month-old DBA/2J and 1.5-month-old C57BL/6J mice (see Supplementary Fig. S2), suggesting that the equator was an appropriate landmark for the anterior hyaloid membrane. 

Algorithm Validation Using a Phantom

Before using very high spatial resolution vis-OCT imaging to quantify the anterior and posterior chamber volumes in vivo, we assessed the accuracy of our vis-OCT volume measurement and reconstruction algorithm by imaging a 3D-printed phantom containing a cavity mimicking the posterior chamber (Fig. 5b). The outer edge of the printed phantom had a diameter of 3 mm, approximately the diameter of the mouse eye. Because 3D printing is subject to errors when printing features with submillimeter dimensions, we acquired an SEM image of the phantom to validate its structure and dimensions (Fig. 5c). We found that the features of the SEM image matched those of our reconstructed vis-OCT image. We then measured the cavity volume by determining the mass of water required to fill the phantom cavity, obtaining 2.99 ± 0.06 µL (n = 7 technical replicates). The volume determined by vis-OCT imaging was 2.96 ± 0.12 µL (n = 6 technical replicates), which was within 1% of the volume determined by the water-filling approach (Fig. 5d). This difference was not statistically significant, and we conclude that our vis-OCT-based approach accurately determined the volume of a cavity in a phantom of similar size to the anterior and posterior chambers. 

In Vivo Anterior and Posterior Chamber Volume Measurements

We reconstructed the entire anterior and posterior chambers of 2-month-old BALB/cAnNCrl albino mice (n = 10; Fig. 6a). As expected, the anterior chamber formed a continuous volume anterior to the iris, whereas the posterior chamber formed a continuous volume posterior to the iris. The anterior chamber resembled a spherical cap below the cornea, and the posterior chamber resembled the upper half of a torus. When viewing the volumetric reconstruction from the posterior view (Fig. 6b), we found that the outer radius of the posterior chamber was larger than the anterior chamber. Figure 6c shows a cross-sectional view of both chambers. 

Figure 6.

 

Reconstruction of the anterior and posterior chambers. (a) Isometric view of the reconstructed anterior segment with anterior chamber (AC) in green and posterior chamber (PC) in blue. (b) Posterior view of reconstruction with the AC and PC shaded in green and blue respectively. (c) Montaged B-scan of the anterior segment with the anterior chamber overlayed in green and the posterior chamber in blue. (d) Comparison of the anterior chamber volume and posterior chamber volume in BALB/cAnNCrl mice reveals that the volumes of aqueous humor in the anterior and posterior chambers are comparable. (e) Comparison of anterior chamber volumes in 2-month-old BALB/cAnNCrl albino mice and 2-month-old C57BL/6J mice reveals that the anterior chamber is larger in the C57BL/6J mice (*P < 0.05; **P < 0.01, ***P < 0.001, ****P < 0.0001).

In 2-month-old BALB/cAnNCrl mice, the measured anterior chamber volume was 1.53 ± 0.34 µL and the posterior chamber volume was 1.72 ± 0.39 µL (Fig. 6d). The total (anterior plus posterior chamber) volume was 3.25 ± 0.49 µL. As an approximate indicator of overall eye size, we also measured the distance between the apexes of the iridocorneal angle on opposite sides of the eye to be 2.73 ± 0.09 mm. We measured the pupil diameter as 2.05 ± 0.12 mm and found no statistically significant correlation coefficient between pupil diameter and anterior chamber volume (r = 0.33, P = 0.35). 

We found that the posterior chamber had a greater volume than the anterior chamber in 7 of the 10 mouse eyes. Overall, the ratio of the anterior chamber volume to the posterior chamber volume ranged from 0.50 to 1.36, with an average of 0.93 ± 0.28. We found that the anterior chamber constitutes 33% to 58% of the total aqueous humor volume (anterior + posterior chamber volumes), with an average of 47 ± 8%. 

To confirm that the accuracy of chamber volumes was not impacted by the volumetric montaging algorithm, we measured anterior chamber volume using both the montaged volume consisting of eight volumes and from a single-volume vis-OCT acquisition that captured the entire anterior chamber. Unfortunately, we could not capture the posterior chamber within a single volume, so we only compared the anterior chamber volume measurements. We found an anterior chamber volume of 1.53 ± 0.34 µL from the montaged reconstructions and 1.55 ± 0.36 µL (n = 10) from the single-volume acquisitions, with no statistical difference between the 2 methods. As compared to the single-volume acquisition, the multi-volume reconstructed volumes had greater volume for six eyes and smaller volume for four eyes. We conclude that it is unlikely the reconstruction scheme biased the measured volume in a specific direction. 

We also imaged the anterior segments in 2-month-old and 7-month-old C57BL/6J mice to investigate chamber volumes in another strain and the effects of age. In 2-month-old C57BL/6J mice, we measured an anterior chamber volume of 2.05 ± 0.25 µL (n = 10), which was, on average, 32% larger than the anterior chamber volume measured in 2-month-old BALB/cAnNCrl animals (Fig. 6e). In 7-month-old C57BL/6J mice, we measured an anterior chamber volume of 2.41 ± 0.29 µL (n = 8), which was, on average, 18% larger than the anterior chamber volume measured in 2-month-old C57BL/6J animals (see Fig. 6e). (Note that because the highly pigmented iris of C57BL/6J mice did not allow sufficient light penetration to the posterior chamber, we could not image the posterior chamber in these mice.) For the 2-month-old C57BL/6J mice, we found that the distance between the apexes of the iridocorneal angle on opposite sides of the eye was 2.89 ± 0.10 mm, which was 6% larger than in the 2-month-old BALB/cAnNCrl mice. If we assume that anterior chamber volume scales with linear dimensions cubed, the 6% difference in linear dimension would correspond to a 19% greater anterior chamber volume in 2-month-old C57BL/6J mice versus 2-month-old BABL/cAnNCrl mice, which is comparable to, although slightly smaller than, the measured 32% difference. We also found that the pupil diameter was 1.96 ± 0.22 mm, which was not significantly different from the BALB/cAnNCrl mice. There was no statistically significant correlation between pupil diameter and anterior chamber volume in the 2-month-old C57BL/6J mice (r = −0.32, P = 0.37). 

We found the distance between the apexes of the iridocorneal angle on opposite sides of the eye for 7-month-old C57BL/6J mice as 3.17 ± 0.12 mm, which was 10% larger than in the 2-month-old C57BL/6J mice. If we assume that anterior chamber volume scales with linear dimensions cubed, the 10% difference in linear dimension would correspond to a 32% greater anterior chamber volume in 7-month-old C57BL/6J mice versus 2-month-old C57BL/6J mice, which is comparable to, although slightly larger than, the measured 18% difference. This suggests that anterior chamber volume scales approximately as the cube of eye size within a strain, as expected. We found that the pupil diameter was 1.77 ± 0.36 mm in 7-month-old C57BL/6J mice, not significantly different from the 2-month-old C57BL/6J mice. However, we did find a statistically significant correlation coefficient between pupil diameter and anterior chamber volume (r = 0.83, P = 0.01). 

Discussion and Conclusions

In this work, we used robotic vis-OCT imaging to obtain volumetric representations of the anterior and posterior chambers of mice in vivo. Two potential applications of this information include a better understanding of aqueous humor dynamics and optimizing intracameral injections in mice. For example, knowledge of inflow rate calculation often depends on anterior chamber volume and is required when using the modified Goldmann's equation to estimate AHD parameters such as unconventional aqueous drainage rate.⁴² Further, several existing models of aqueous humor fluid dynamics, such as the movement of aqueous humor through the iris-lens channel, require knowledge of specific volumes of the chambers.⁴³ 

A key finding of this work is that only approximately 47% of the total aqueous humor volume is contained within the anterior chamber in BALB/cAnNCrl mice. This is generally consistent with findings based on histology in C57BL/6J mice and conventional NIR-OCT done on CD-1 mice, which found that approximately half of the aqueous humor resides in the mouse posterior chamber. Overall, this suggests that a posterior chamber of comparable size to the anterior chamber is not an isolated finding observed in vis-OCT imaging of BALB/cAnNCrl mice. 

To preliminarily explore the effect of mouse strain, we also imaged the anterior chamber of 2-month-old C57BL/6J mice, obtaining 2.05 ± 0.25 µL compared to 1.55 ± 0.36 µL in the BALB/cAnNCrl animals, perhaps suggesting that chamber volumes may be larger in C57BL/6J mice; however, further evaluation under varied experimental conditions is warranted. It is also noteworthy that albinism, present in the BALB/cAnNCrl mice we used in this study, is known to affect IOP and anterior segment development.⁴⁴ Thus, further studies will be required to evaluate how age, sex, and strain affect anterior chamber volume. Future studies should also consider variables such as type of anesthesia and hydration status. 

We also examined the effect of age on anterior chamber volume, finding that the anterior chamber volume in 7-month-old C57BL/6J mice was 18% larger than that in 2-month-old C57BL/6J mice. This observation is consistent with the fact that globe size is known to increase with age in mice,^45,46 with a particularly significant increase over the first 6 months. Using data from Li et al.,⁴⁵ we estimate that the globe diameter increases by 8% in C57BL/6J mice between 2 and 7 months of age, which implies a 26% increase in globe volume if we assume isotropic growth. This is similar to, although slightly larger, the 18% difference that we observed between 2-month-old C57BL/6J mice and 7-month-old C57BL/6J mice. 

It is important to compare our measured volumes with previous reports. We are not aware of any papers describing direct measurements of anterior chamber volume in mice other than a passing comment by Avila et al.,⁸ who stated “we have estimated the anterior chamber volume to be approximately 2 µL, calculated as the volume of revolution from the projection of a plastic-embedded tissue section of a formalin-fixed mouse eye.” This is remarkably close to our best estimates (see below). When removing all aqueous humor from the eye, John et al.¹³ measured a total aqueous volume of 5.8 µL in C57BL/6J mice (n = 12, mean ± SEM) and 5.1 ± 0.4 µL in C3HeB/FeJ mice (n = 9), with all mice being 8 to 12 weeks old. Using a similar aspiration method, Zhang et al.¹¹ and Aihara et al.¹⁰ measured total aqueous volumes of 7 to 7.2 µL in CD-1 mice (age = 4–6 weeks) and of 5.9 µL in NIH white Swiss mice (8–12 weeks of age), respectively. 

Attributing approximately half of the total aqueous volume to the anterior chamber, consistent with our data, the above studies would imply anterior chamber volumes of 2.5 to 3.6 µL, which is larger than our direct optical measurements of 1.55 to 2.41 µL. Some of this difference may be due to strain and age effects; further, one cannot exclude the possibility of inadvertent collection of secondary aqueous during aspiration, despite careful efforts to avoid such effects.^47,48 Finally, we note that the Zhang et al. data are larger than the other direct measurements and is perhaps somewhat of an outlier, especially considering that the mice in that study were only 5 to 6 weeks of age. Thus, we are inclined to consider total aqueous volumes of 3 to 6 µL as reasonable, depending on age, with corresponding bounds on anterior chamber volume between 1.55 and 2.8 µL, that is, from our lowest directly measured volume in 2-month-old BALB/cAnNCrl mice to 47% of the presumed upper bound of total aqueous humor volume of 6 µL. 

One important consideration in interpreting our data is the possibility of visible light-induced pupillary constriction, which could alter anterior chamber volume. In myopic humans, the anterior chamber depth has been positively correlated with pupil diameter,⁴⁹ although no such data exist in mice that we are aware of. Thus, we calculated the Pearson correlation between pupil diameter and anterior chamber volume for each experimental group, finding no statistical significance for both 2-month-old BALB/cAnNCrl mice (r = 0.33, P = 0.35) and C57BL/6J mice (r = -0.32, P = 0.37); however, this correlation was statistically significant for the 7-month-old C57BL/6J mice (r = 0.83, P = 0.01). Consequently, vis-OCT-induced pupillary constriction may affect measured volumes in this later cohort. However, for the 2-month-old BALB/cAnNCrl mice, we measured a pupil diameter of 2.05 ± 0.12 mm, which is closer to the mouse's dilated pupil size than to the constricted size.⁵⁰ This suggests that the fast scanning of the vis-OCT beam does not fully constrict the mouse pupil. For the 2-month-old C57BL/6J mice, we measured a pupil diameter of 1.96 ± 0.22 mm, which was not statistically different from the BALB/cAnNCrl mice. For the 7-month-old C57BL/6J mice, we measured a pupil diameter of 1.77 ± 0.36 mm, which was not statistically different from the 2-month-old C57BL/6J mice. Thus, it is unlikely that differences in pupillary size fully explain differences in measured anterior chamber volumes between these two strains of mice and C57BL/6J mice of different ages. Nonetheless, future studies should control for, or at least measure, pupillary diameter when determining anterior and posterior chamber volumes. 

Another important consideration in interpreting our data (and others) is that anesthesia affects ocular physiology in several important ways. First, anesthesia is known to affect aqueous inflow rate in both monkeys^51,52 and mice⁹ in an anesthesia and time-dependent manner (see below). Second, anesthesia also affects IOP, which in turn affects ocular volume (and thus anterior chamber volume) through an ocular compliance effect. The literature in this area is somewhat contradictory; we here focus only on IOP measurements in mice where awake IOPs measured by TonoLab rebound tonometry were compared to IOPs under ketamine/xylazine anesthesia, since this anesthetic regimen was used during our vis-OCT imaging. Even with this focus, reported IOP changes due to anesthesia are discordant, ranging from a 2.7 millimeters of mercury (mm Hg) drop at 10 minutes in BALB/cAnNCrl mice⁵³ to a 6.4 to 7.8 mm Hg increase in C57BL/6J mice.⁵⁴ We were unable to obtain reliable IOP measurements during the vis-OCT imaging process, but we here argue that, in any case, anesthesia-induced IOP changes in anterior chamber volume were likely very small, as follows. Sherwood et al. measured mean ocular compliance in control eyes of 11-week-old C57BL/6J mice to be 43 to 49 nL/mm Hg at a reference IOP of 13 mm Hg.⁵⁵ This means that an IOP change of 5 mm Hg would change total ocular volume by 215 to 245 nL, which is less than 2% of the total volume of the mouse eye. Even in the unlikely scenario that all the volume change of the eye occurred in the anterior chamber, these IOP-associated volume changes would only be of order 10% of our estimated anterior chamber volumes. Thus, this effect is judged to be small and can be safely ignored. 

Implications for Aqueous Humor Dynamics

As noted above, determination of aqueous inflow rate depends on accurate knowledge of anterior chamber volume. Here, we reanalyze a recent paper⁹ on this topic in light of our finding that anterior chamber volume is significantly less than total aqueous volume. We specifically consider the work of Toris and colleagues, who used a custom fluorophotometer to estimate an aqueous inflow rate, Q, according to a standard equation for human eyes:  

\begin{eqnarray*} Q = \frac{{d\left( {\ln \left( {{C_c}} \right)} \right)}}{{dt}}\ \left[ {{V_a} + \frac{{{C_c}}}{{{C_a}}}{V_c}} \right] \end{eqnarray*}

where C_c and C_a are measured concentrations of fluorescein in the cornea and anterior chamber, respectively, and V_c and V_a are the volumes the cornea and anterior chamber, respectively. We have selected this paper not because the measurements were poorly done; quite the converse – the work represents the use of custom technology to carefully determine inflow rates in the mouse eye. We note in passing that there are a number of assumptions underlying the above equation, some of which may be less valid in the mouse eye versus the human eye, for example, neglect of tracer diffusion into the posterior chamber. Here, we will not concern ourselves with these complex topics, and simply investigate the effects of different anterior chamber volumes. 

If we denote the true value of the anterior chamber by \(V_a^*\), with the corresponding true value of the aqueous outflow rate being denoted by Q*, then we can write  

\begin{eqnarray*} \frac{{{Q^*}}}{Q} = \frac{{\left[ {V_a^* + \frac{{{C_c}}}{{{C_a}}}{V_c}} \right]}}{{\left[ {{V_a} + \frac{{{C_c}}}{{{C_a}}}{V_c}} \right]}} \end{eqnarray*}

which can be interpreted as a correction factor for reported values of Q based on an incorrect anterior chamber volume, V_a. 

Toris et al.⁹ took corneal volume to be V_c =  0.5 µL and anterior chamber volume to be V_a = 5.9 µL. We digitized Figure 5b of the Toris paper and determined that the mean value of the ratio \(\frac{{{C_c}}}{{{C_a}}}\) was 4.0. Using this value, we then substituted our range of anterior chamber volumes in the above equation to determine that \(\frac{{{Q^*}}}{Q}\) lies in the range of 0.45 to 0.61. Stated differently, the over-estimation in the Toris et al. paper in the reported aqueous flow rate is somewhere between 39% and 55%, which is substantial. Toris et al. reported an aqueous flow rate of 90 ± 70 nL/min in female CD-1 mice greater than 6 months of age and weighing between 35 and 45 g under ketamine/xylazine anesthesia. Using the above correction factor, we would instead estimate a corrected mean aqueous inflow rate of Q* = 40 − 55 nLl/min. The expected aqueous inflow rate in younger (smaller) mice would be less than the above value. Other studies of inflow rate that make similar assumptions about anterior chamber volume will suffer from the same inaccuracies.^3,8,10,11 

It is important to note that Toris et al. also reported a large effect of anesthesia on aqueous inflow rate in the mouse, with the estimated inflow rate under 2,2,2-tribromoethanol anesthesia being more than 2-fold greater than that estimated under ketamine/xylazine, reinforcing the point that careful consideration of the anesthesia regimen is indicated when studying aqueous humor dynamics.⁹ In what follows, we will consider the case of ketamine/xylazine anesthesia, because this is a commonly used regimen in mice (although different groups use different doses) and because our vis-OCT measurements were obtained on mice under this regimen. 

Goldmann's equation relates inflow rate to other AHD parameters, and may be written as  

\begin{eqnarray*} Q - {Q_u} = C\left( {IOP - EVP} \right), \end{eqnarray*}

where EVP is episcleral venous pressure; C is pressure-dependent (conventional) outflow facility; Q is aqueous inflow rate; and Q_u is pressure-independent outflow rate, sometimes called unconventional outflow. Sherwood et al. measured facility in 66 postmortem eyes of C57BL/6J mice (10- to 14-week-old male mice), determining a geometric mean population value of 5.89 nL/min/mm Hg at an IOP-EVP difference of 8 mm Hg, which corresponds to a conventional outflow rate of 47 nL/min. Using this conventional outflow rate with our adjusted inflow rates above indicates that 0% to 15% of aqueous humor is predicted to exit the eye via the pressure-independent (unconventional) route, similar to that seen in humans and monkeys.⁵⁶ Using values for inflow that are not corrected for anterior chamber volume causes this estimate for unconventional outflow to jump to 48%. Regrettably, the above calculations have drawn on data from different strains and ages of mice, and carrying out careful measurements of inflow rate, anterior chamber volume, and aqueous outflow facility in mice of the same age, sex, and strain may help us better understand the role of unconventional outflow in mice, which has been controversial in the past.⁵⁶ 

Limitations

A drawback of our study is that the anterior hyaloid membrane was not well visualized. To address this, we approximated the hyaloid membrane location by the equator of the globe, an assumption we validated with micro-CT imaging. A second limitation is that we could only visualize the posterior chamber in non-pigmented mice, even when using our advanced robotic vis-OCT imaging approach, although we compared estimates from vis-OCT images to standard histology of pigmented mice. Finally, in this study, we considered only a few mouse strains and limited ages. Future work should investigate more strains and also a wider range of ages. In fact, chamber volumes could be tracked longitudinally in an in vivo setting to assess how specific treatments or procedures impact ocular development and growth. 

Acknowledgments

The authors thank Darryl Overby (Imperial College London) for insightful comments on the manuscript and Ying Hao (Duke Eye Center Core Facility) for histological studies. This work was supported in part by National Institutes of Health grants R01EY029121, U01EY033001, R01EY033813, R01EY034740, R01EY034353, R01EY030124, R01EY032062, R01EY032507, F30EY034033, R01EY031710, and R44EY026466, P3EY019007, and P30EY005722, Illinois Society for the Prevention of Blindness, the Christina Enroth-Cugell and David Cugell Fellowship for Visual Neuroscience and Biomedical Engineering, unrestricted departmental funding (Columbia) from Research to Prevent Blindness, NY First Empire Fund, and the Georgia Research Alliance (CRE). 

Disclosure: D. Kim, None; R. Fang, None; P. Zhang, None; Z. Yan, None; C. Sun, Opticent Inc. (F) which, however, did not support this work; G. Li, None; C. Montgomery, None; S.W.M. John, None; W.D. Stamer, None; H.F. Zhang, Opticent Inc. (F); C.R. Ethier, None