Vision and ocular health have a pervasive and deep-rooted impact on many aspects of life, health, sustainable development, and the economy.1 Despite this, ∼2.2 billion people suffer from ocular diseases, including blindness,2 and this number is expected to increase significantly as the global population ages.3 In 1991, ocular imaging was transformed by the advent of optical coherence tomography (OCT).4 For the first time, micron-scale resolution of ocular tissues became possible, offering nearly histological detail in the living human eye. This has had a profound influence on the clinical understanding of ocular diseases such as diabetic retinopathy, glaucoma, and macular degeneration.5^,6 Subsequently, researchers have developed strategies to visualize dynamic scattering within retinal blood vessels using OCT imaging as a way to perform ocular angiography without exogenous vascular contrast agents.7 However, the imaging of human iris vasculature in vivo remains a persistent challenge, resulting in limited knowledge of the relationship between iris vasculature and the pathogenesis of ocular diseases.8

The iris stroma contains a rich vascular plexus that maintains anterior chamber homeostasis by delivering nutrients to avascular ocular tissues (e.g., the cornea, lens, and trabecular meshwork) and supplying oxygen to the aqueous humor. Pathological iris vasculature formations can be either the cause or result of systemic vascular disorders,5 including diabetes and Alzheimer’s disease, as well as ocular diseases such as iris neovascularization, cataract, glaucoma, uveitis of various etiologies, and tumors.9^,10 In clinical settings, fluorescein angiography and indocyanine green angiography are often used to visualize the iris vessels. However, intravascular exogenous contrast agents can sometimes lead to adverse symptoms or immune hypersensitivity reactions such as nausea, vomiting, dyspnea, syncope, and anaphylaxis.11 Although primarily used for retinal vascular imaging,5^,12 optical coherence tomography angiography (OCTA) is now also implemented for mapping blood flow in the normal iris,13^,14 studying abnormal iris vessel formations,15^–19 and assessing treatment outcomes.20^,21 OCTA imaging is a dye-free method, thereby eliminating the risk of dye-related side effects.22^,23 OCTA imaging is also renowned for its volumetric and cross-sectional perspectives, which can potentially enhance the localization of lesions and provide more precise measurements of vascular features.24^–26 In addition, the near-infrared wavelengths used for OCTA imaging enable better iris tissue penetration, especially for highly pigmented irises.14^,24^,27

Despite recent advances, OCTA still lacks adequate imaging contrast for iris vascular analysis. Key limitations—radial-field-dependent contrast (particularly blind spots), a small field of view, and long data acquisition time—hinder its use in both scientific research and clinical applications.28 The core problem stems from the standard OCTA raster scan protocol, which uses repeated fast scans in one direction while stepping slowly in the perpendicular direction. Crucially, vessel visibility depends heavily on their orientation relative to the fast-scan direction. This strategy works best when a vessel runs perpendicular to the fast-scan direction. However, as this angle decreases, the imaging becomes increasingly blurred, and vessels can become completely undetectable when aligned parallel to the fast scan.14

Circular scanning29^–31 offers a perpendicular fast scanning direction to the radially aligned iris vasculature, making it an ideal alternative for iris OCTA imaging. However, its implementation in OCTA is not straightforward. OCTA imaging relies on motion contrast, with blood flow causing dynamic OCT signals through light reflection or scattering, in contrast to the more stable signals from surrounding tissues.12^,32 The motion contrast is mainly determined by flow flux (or speed) and interscan time (time lag between repeated B-scans at the same location). Shorter interscan times are more effective for capturing fast-moving blood, whereas longer intervals are better for slower blood flow but increase the risk of motion-related distortions.33^–38 Our study indicates that an interscan time of 5 to 15μs yields clear iris vessel images in OCTA.14 However, varied trajectory lengths from center to periphery in circular scanning can result in significantly varied interscan times at the same scan speed, leading to substantial inconsistencies in OCTA imaging contrast, which is a major limitation of this method.

In this work, we introduce a circular interleaving scan OCTA solution designed to revolutionize iris vascular imaging. We employed a circular fast-scan mode to provide quasiperpendicular intersections with the blood flow in every angular direction, and we used different times of the interleaving scanning in the central, middle, or peripheral iris to ensure consistent imaging contrast across all concentric scanning paths. We then performed a coordinate transformation to convert the projected triangular blood flow map into a rectangular format, which is pivotal for eye-motion compensation, vessel alignment, and imaging contrast enhancement. Subsequently, we converted this data from Cartesian to polar coordinates, culminating in the reconstruction and visualization of the iris vasculature. This method provides 360 deg high-contrast imaging of the iris vasculature without the need for hardware changes. It eliminates angular-dependent blind spots, significantly boosting the reliability of vascular quantification, thereby showing great promise for future clinical applications.

We developed a spectral domain OCT system for iris structure and vasculature imaging, as reported previously.14 This system employs a (1290±40)nm light source for deeper iris stromal penetration, incorporates a linear-wavenumber (k) spectrometer (76 kHz line rate) for improved detection sensitivity, and uses a custom iris scan objective lens optimized for focusing across the entire iris. The maximum light power used was 1.8 mW, the lateral resolution was ∼11.5μm, the axial resolution was ∼15.0μm, and the maximum imaging depth was 6.94 mm. The OCT detection sensitivity [20 log(Amplitude)] decreased from 99.7 to 74.1 dB, and the 6 dB fall-off range was ∼0 to 3 mm (calculated in air). Further details of the OCT/OCTA system are provided in Fig. 6, and the corresponding text in the Appendix.

Compared with a raster scan with a duty cycle of 80% to 90%,12 a circular scan can provide nearly 100% duty cycle (see Fig. 7 in the Appendix) that improves imaging efficiency. For the circular scan pattern, the sinusoidal waveforms that control X and Y scanners can be expressed as {Xk=Ak·cos[ωktk+φk(ωk)]Yk=Ak·sin[ωktk+φk(ωk)],(1)where Xk and Yk are the driving voltages; k is the circle number, k=1,2,…,M (M is the maximum circle number); Ak is the magnitude of the voltage; ωk is the angular frequency (rad/s); tk is the time sequence; and φk(ωk) is the compensation phase value.

To maintain a constant increment in the scanned radius (ΔR), the driving magnitude can be set as Ak=kA1. To maintain a constant sampling interval (ΔL) in the circular direction for circles with varying radii, the circular scan must be conducted at a constant linear velocity.39 In this case, ωk can be expressed as ωk=2πfANk,(2)where fA is the A-scan rate; Nk is the A-scan number in the circle k; and nk is the sampling dot sequence in the circle k, nk=1,2,…,Nk. The time sequence tk can be expressed as tk=nk/fA. To maintain a constant ΔL, Nk=kN1.

Due to mechanical inertia, continuous variations in driving frequency, phase, and magnitude at different concentric trails [Eqs. (1) and (2)] can result in phase delays between the control and execution signals, resulting in nonuniform sampling and image distortion. By utilizing the compensation phase φk(ωk), we refined the scan positioning and enhanced image quality according to the following expression: {φk(ωk)=arctan(2ξωkωn1−ωk2ωn2),(0≤ωkωn≤1)φk(ωk)=0−arctan(2ξωkωnωk2ωn2−1).(ωkωn>1),(3)Here, we used the calibrated values for the galvo mirrors: an undamped angular natural frequency (ωn) of 4956rad/s and a damping coefficient (ξ) of 0.63.40 The imaging performance with and without distortion correction is demonstrated in Fig. 8 in the Appendix.

The conventional circular scan method exhibits radius-dependent variations in interscan time (Δt), leading to inconsistent OCTA detection sensitivity across the iris. Specifically, peripheral regions experience longer Δt that enhances slow-flow detection but increases motion artifacts, whereas central regions have shorter Δt, favoring fast-flow visualization. To overcome this limitation, we implemented a radial field-dependent interleaving strategy that maintains Δt within 5–15μs at all radii. This optimization ensures uniform motion contrast while preserving consistent spatiotemporal sampling across the iris vasculature. Figure 1 illustrates the circular interleaving scan method, featuring 270 concentric circles with diameters spanning 48.1μm to 13.0 mm and sampling points per circle increasing progressively from 12 to 3240 A-lines. The scanning area was divided into the slingshot (1X), double-shot (2X), and triple-shot (3X) interleaved OCTA imaging zones, spanning circles 1 to 90, 91 to 180, and 181 to 270, respectively, as illustrated in Fig. 1(a). In the 1X zone, a standard OCTA scanning protocol was implemented using three-repeat B-scans. Meanwhile, in the 2X and 3X zones, the scans were initially undersampled at the 2X or 3X ratios, with each B-scan repeated three times at the specified undersampling rate. Next, interleaving scans at 2X or 3X were conducted, along with additional repeated B-scans. This process allowed for the connection of interleaving, undersampled, and repetitive B-scans, resulting in high-resolution cross-sectional imaging for iris structure and blood flow signals.

Figure 1(b) demonstrates the sinusoidal driving waveforms for both scanners and the data acquisition signal from the line scan camera. Repeated sinusoidal waveforms are used in each circular scan path for repeated B-scans, whereas phase shifts in control signals are necessary for interleaving scans. An extra A-scan time of 13.16μs was added as a stabilization time to facilitate the transition between circles or the initiation of interleaving scans. The duty cycle in this example was 99.95%. Figure 1(c) shows the required B-scan time (interscan time, Δt) for a single circle, ranging from 0 to 14.2μs (1X zone), 7.2 to 14.2μs (2X zone), and 9.5 to 14.2μs (3X zone). Figure 1(d) illustrates the angular velocities, which decrease sharply from the center to the periphery of the iris. The scanner driving frequencies (fscan=ωn/2π) also experienced a considerable decline from 6333.3 to 70.4 Hz. Figure 1(e) illustrates the linear scan velocities (V) required to maintain a consistent spatial sampling resolution: 0.96m/s (1X zone), 1.92m/s (2X zone), and 2.87m/s (3X zone).

To assess the performance of the circular interleaving scan method, we compared its imaging capabilities with those of raster scans. The raster scans featured fast-scan directions in either the horizontal or vertical orientation, whereas the circular scans were configured with either a larger field of view (FOV) or a higher sampling density. Prior to OCTA acquisition, rapid horizontal and vertical scout scans established the scanning center, ensuring consistent centration for both circular and raster protocols. Table 1 provides a comprehensive summary of these scanning protocols. For each protocol, we conducted three repeated measurements for OCTA signal acquisition, executing over 1 million A-scans. All protocols complied with laser safety standards (ANSI Z136.1) at a 1.8 mW maximum power and a 17.3 s scan duration.

As shown in Table 1, circular scans offer multiple advantages for iris vascular imaging. First, they achieved a 99.95% duty cycle (versus 82.0% for raster scans), enabling denser spatial sampling within similar acquisition time. Second, circular scan OCTA efficiently captured the round iris tissue: unlike raster scans, which gather excessive redundant information from imaging field corners (27.3% of the circular area), circular scans focused on the iris to optimize data collection. Moreover, circular scans provide direction-optimized sampling for iris vasculature. Although raster scans use different sampling rates in fast/slow directions, circular scans offer two distinct options: circumferential and radial. To enhance the iris blood flow detection signal, we set the circumferential sampling density to approximately twice that of the radial direction during circular scanning.

Two healthy human subjects participated in this study. Subject 1 (Asian, male, 38 years old) underwent iris vascular imaging for both eyes. We used a drop of pilocarpine nitrate 30 min before OCTA imaging to constrict the pupils to ∼1.2mm in diameter. Subject 2 (Asian, male, 24 years old) underwent OCT iris structure imaging in his right eye without any eye drops. This study was approved by the Institutional Review Board of Foshan University and adhered to the tenets of the Declaration of Helsinki.

To expand on the benefits of circular scan OCTA, we developed a coordinate-transformation-based image processing strategy to mitigate the effect of eye motion, realign disconnected vessels, and enhance imaging contrast. The accompanying imaging example shown in Fig. 2 illustrates the results of a large-FOV circular scan performed on the constricted iris of subject 1’s left eye. Figure 2(a) illustrates the vascular signal extraction procedure. The top row shows the raw B-frames, which were acquired with 2X interleaved and repetitive scans. These interleaved B-frames were then combined interpolatively to create a high-resolution B-frame (second row); subsequently, the repetitive high-resolution B-frames were used to extract blood signals using an optical micro-angiography algorithm.41 The reconstructed high-resolution B-frame blood signals are shown in the third row. We removed the anterior boundaries of the iris pigmented epithelial layer and projected the blood signals from each B-scan using the means of the 20 brightest signals in depth, resulting in a triangular mapping of blood flow (e.g., 12 pixels at the top, 3240 pixels at the bottom, and 270 pixels in height). We stretched the height of the triangular map for better visualization, as shown in Fig. 2(b). Subsequently, the triangular map was directly transformed into a circular form from Cartesian to polar coordinates, as shown in Fig. 2(c). The horizontal bright lines in Fig. 2(b) and the concentric bright circles in Fig. 2(c) represent eye motion artifacts.

Notably, realigning the disconnected vessels in the triangular blood flow map is challenging due to the fluctuating numbers of pixels in different lines; this task becomes even more demanding in the circular map due to the necessity of image rotation in different circles, which is challenging and often impractical. To address this, we transformed the triangular map into a rectangular map, where the bright lines (eye motion noises) remained horizontal, but the blood signals became mostly vertical and parallel, as shown in Fig. 2(d). This facilitated the horizontal registration of the disconnected vessels to enhance blood flow continuity. We next realigned the vessels from the top to the bottom as the scans were less affected by eye motion at the top due to a shorter scanning time. We applied a threshold of mean ±0.2SD to eliminate motion-induced bright lines by clipping pixel brightness to these limits. Subsequently, we enhanced the contrast of each line by rescaling its values to mean ±2SD. The processed rectangular map is shown in Fig. 2(e), where the continuity and visibility of the vessels were significantly improved by bright line removal, vessel realignment, and imaging contrast enhancement. We transformed the image back into the triangular map [Fig. 2(f)], and then converted it into a circular image presented in a polar diagram. We applied a median filter to remove noise while preserving the geometry of the iris vasculatures [Fig. 2(g)]. In comparison, the insets in Fig. 2(g) show lower noise levels, enhanced contrast, and better flow connectivity than those in Fig. 2(c).

To evaluate and compare the imaging quality and measurement repeatability of the iris vasculature, we performed OCTA imaging using the large-FOV circular interleaving scan, horizontal raster scan, and vertical raster scan protocols (see Table 1 for measurement parameters). Seven repeated scans were acquired per protocol within one hour. Blood vessels in the OCTA images were then manually segmented by an expert with 2 years of experience. Results are presented in Fig. 3. The iris is divided into a pupillary zone and a ciliary zone by the collarette—a zigzag circle approximated as an ideal circle. We partitioned the iris into eight regions, categorized by these two zones and four directions: superior (S), inferior (I), temple (T), and nasal (N). We measured the vessel density by computing the ratio of pixels marked as blood signals to the total pixels in each region, and we counted the vessel numbers only if the vessel length was at least one-third of the radial length of that region. The quantification results are shown in Figs. 3(d) and 3(e). In comparison, the iris vessels in the pupillary zones (1–4) exhibited higher vessel densities but lower vessel counts than those in the ciliary zones (5–8).

As evidenced by Fig. 3, circular interleaving scan OCTA outperformed raster scan OCTA in terms of enhanced imaging contrast, elevated vessel densities, and increased vessel numbers for iris vascular characterization. The raster scans have angular-dependent imaging contrast, making it difficult to detect iris vessels that align with the fast-scan direction. In general, the horizontal regions (zones: 1, 2, 5, and 6) exhibited lower vessel densities and vessel numbers in the horizontal scan compared with the vertical scan, with reductions of 33.6%±17.6% and 29.1%±19.7%, respectively; the vertical regions (zones 3, 4, 7, and 8) showed lower vessel densities and vessel numbers in the vertical scan compared with the horizontal scan, with reductions of 27.8%±7.7% and 17.9%±14.1%, respectively. As a result, the circular interleaving scan achieved 360 deg high-contrast iris vascular imaging by eliminating angular-dependent blind spots, with a significant enhancement of 39.0%±11.2% in vessel density measurement and 25.2%±15.7% in vessel number quantification compared with the raster scan measurements. In addition, circular scan OCTA efficiently captured the round-shaped iris tissue without gathering excessive redundant information from the surrounding corners of the imaging field, which is a common issue with the raster scan method.

Inspired by Figs. 2(d) and 2(e), we found that using coordinate transformation in circular OCTA imaging introduces a distinctive feature that allows for the analysis of iris vessel distribution in the angular dimension. We then employed the large-FOV and high-density circular scan protocols (Table 1) to visualize the iris vasculature in the right eye of subject 1 within 30 min. We quantified vessel density and vessel numbers in the rectangular blood flow maps at every 30 deg angular span. The imaging results are displayed in Figs. 4(a) and 4(b), showcasing examples of OCTA imaging and corresponding marked vasculature networks in both circular and rectangular coordinates. For each scan protocol, we performed seven repetitive measurements. The additional six measurements, with marked blood vessels overlaid on the OCTA imaging, are also provided in Figs. 4(a) and 4(b).

Figures 4(c)–4(f) show the quantification of iris vessel distribution in the angular dimension. Interestingly, we observed slightly lower vessel densities (13.6% lower in the pupillary zone and 16.5% lower in the ciliary zone) in the high-density circular scan measurements than in the large-FOV circular scan measurements. However, the vessel numbers were slightly higher (9.5% higher in the pupillary zone and 7.1% higher in the ciliary zone) in the high-density circular scan measurements. This discrepancy is attributed to the fact that the high-density scan provides sharper imaging of the iris vessels, resulting in smaller vessel diameters than those observed in the larger FOV circular scan measurements. These insights highlight the potential variations in imaging and quantitative analysis results that can arise when using different OCTA scanning protocols. The standardization of OCTA scanning and quantification methods will facilitate further studies aimed at distinguishing between healthy and diseased irises, providing more precise, robust, and consistent quantitative OCTA metrics for the characterization of iris vasculature.

In addition to the angiographic imaging, we also compared the OCT structural imaging of the iris in subject 2’s right eye using both raster scan [Figs. 5(a) and 5(b)] and circular scan methods [Figs. 5(c) and 5(d)]. With a duty cycle close to 100%, the circular scan strategy covered a larger scan field (70.8mm2) than raster scan imaging (duty cycle: 82%, scan field: 64mm2) while maintaining similar data acquisition times (∼3s) and spatial resolutions (∼18μm).

Notably, the iris crypts, which are openings located on either side of the collarette, are clearly visible in the en face images from both raster scan and circular scan OCT images in Figs. 5(b) and 5(d). These crypts allow the aqueous humor in the eye to flow into the stroma and deeper layers of the iris. In cross-sectional imaging, superficial crypts [indicated by the red arrows in Figs. 5(b) and 5(d)] are also clearly visible in both raster and circular scan OCT images. Critically, because the crypts exhibit a circular alignment, the circular scan provides perpendicular sectioning relative to the crypt orientation. This offers a distinct advantage over the raster scan, excelling in the visualization and characterization of crypts within the iris stroma (highlighted by green arrows). This demonstrates the superior capability of the circular scan method in providing comprehensive and detailed imaging of the iris structure, making it a valuable tool for clinical and research applications.

In this study, we have presented an advanced approach to iris vascular imaging using a circular interleaving scan OCTA method. We have demonstrated that the circular scan pattern is the most effective approach for visualizing the radially oriented iris vascular network in OCTA, providing 360 deg high-contrast imaging of the iris vasculature. By combining circular scanning with a radial field-dependent, repetitive, and interleaving scanning strategy (1X/2X/3X zones, Fig. 1), we can maximize motion contrast and maintain consistent spatiotemporal resolutions across a large scan field for the entire iris. We also developed a coordinate-transformation-based image processing strategy, specifically tailored for circular scan iris vascular imaging, to effectively reduce eye motion, realign vessels, and enhance the imaging contrast (Fig. 2). Compared with the raster scan OCTA method, the circular interleaving scan OCTA can provide superior visualization and quantification of the iris vasculature (Fig. 3) and improved data acquisition efficiency, all without the need for hardware upgrades. This innovation represents a significant advancement in iris vascular imaging, opening the door to a better understanding of ocular diseases and holding great promise for prospective clinical applications.

Circular scan OCT further enhances cross-sectional visualization of iris structural features, particularly crypts, due to its perpendicular sectioning relative to crypt orientation (Fig. 5). These crypts, located at both superficial and deeper tissue layers, contribute significantly to fluid exchange across and through the iris. It is postulated that irises with higher crypt grading exhibit increased compressibility and greater volume changes during physiological pupil dilation and constriction,42 which is a known mechanism for reducing the risk of acute angle-closure glaucoma attack.43 Therefore, the distinct cross-sectional perspective of iris features provided by circular scan OCT imaging could pave the way for innovative diagnostic and therapeutic strategies in ophthalmology, especially in the prevention and management of glaucoma.

We have noticed that eye motion is the primary factor contributing to the observed variations in OCTA imaging during the repetitive measurement of iris vasculature (Figs. 3 and 4). Unfortunately, effective solutions for anterior eye motion correction, whether through active tracking/stabilization techniques or passive imaging registration/compensation methods, remain elusive. In circular scan OCTA imaging, eye motion can be decomposed into axial and lateral eye motions, with lateral eye motion further categorized into radial and circumferential components. Compensating for axial eye motion is relatively straightforward using image processing methods (e.g., cross-correlation). Herein, we successfully addressed circumferential eye motion in the coordinate-transformed rectangular map using basic thresholding/realigning approaches (Fig. 2). For challenging cases (e.g., severe eye motion or irregular vasculature in disease), more sophisticated algorithms (e.g., feature-based non-rigid registration) may be required to further enhance vessel continuity. However, correction of radial eye motion remains unachieved. In the current study, because subjects were free of ocular vascular diseases and used pilocarpine nitrate for pupil constriction prior to OCTA imaging, residual radial motion was not obvious (Figs. 3 and 4). Nevertheless, in future clinical applications, uncorrected radial motion may severely affect imaging quality—particularly among unmedicated patients with large pupils, individuals with pathological conditions inducing iris tremors, and elderly subjects with impaired fixation. Thus, correcting residual eye motion is crucial in clinical settings. Although eye-tracking systems, such as scanning laser ophthalmoscopy,44 have proven beneficial for tracking and stabilizing the eye’s position during data acquisition in clinical retinal OCT/OCTA systems, their successful implementation in anterior-segment OCT/OCTA systems has yet to be realized. Therefore, developing an eye-tracking system for the anterior-segment OCT/OCTA system may represent the optimal solution to eliminate motion artifacts from iris vasculature images.

The distinct pattern of the iris’s outward radiating vascular network can be converted into a vertically parallel blood flow map in the angular dimension through coordinate transformation, as depicted in Figs. 2(d) and 2(e). This facilitates quantitative analysis of the iris vascular distribution across the angular dimension from 0 to 360 deg (Fig. 4). Leveraging the orthogonal relationship between blood flow signals and motion noises, we can further enhance the signal-to-noise ratio by employing image processing methods such as direction-dependent filters. Recently, deep learning has emerged as a valuable tool in OCT/OCTA imaging, facilitating tasks such as retinal layer segmentation, disease classification, and image quality enhancement.45 Leveraging the power of deep learning, we can also enhance the image quality of iris vascular imaging and automatically annotate blood vessels, thus eliminating the need for manual intervention, as depicted in Figs. 3 and 4.

Although this study validates the methodology in healthy irises under optimized conditions (pharmacological pupil constriction), future work will explore its capabilities in irises with varied pigmentation, across age groups, and affected by pathologies such as diabetes, neovascularization, uveitis, tumors, and iris coloboma. It is important to note that although iris vessels in normal areas are expected to show quasiparallel alignment in the rectangular map and radial alignment in the circular map, vessels in lesion areas may have irregular distributions, variations in blood connectivity, and differences in vessel density. For such cases, combining circular scanning with adaptive protocols (e.g., regional high-density scans) could enhance diagnostic capability. These distinct characteristics can also be effectively identified and differentiated using deep-learning-based methods. Our goal is to create a valuable clinical tool for the diagnosis and quantitative assessment of iris diseases by harnessing the potential of the circular scan OCTA imaging technique to capture high-resolution iris vascular imaging. This tool will serve to complement the information provided by retinal OCT/OCTA imaging, enabling more comprehensive diagnoses and potentially revealing insights while fostering innovative advancements in the field of vision science and ophthalmology.

The home-built spectral domain iris OCT/OCTA system [Fig. 6(a)] uses a (1290±40)nm light source for deep tissue penetration, a linear-wavenumber (k) spectrometer for enhanced sensitivity, and a custom iris scan objective lens optimized for full-iris focusing.14 The spectrometer employs a grating-prism dispersion group to linearly disperse interference signals in wavenumber space.46 An InGaAs line scan camera (GL2048L-10A-ENC-STD-210, Sensors Unlimited, Princeton, New Jersey, United States) records these signals at 76 kHz. A frame grabber (PCIE-1433, National Instruments, Austin, Texas, United States) transfers data to a computer for real-time fast Fourier transform processing, generating depth profiles (A-scans). System sensitivity and axial resolution (0 to 6.94 mm depth in air) are characterized in Figs. 6(b) and 6(c).

As shown in Fig. 6(d), the scan objective was designed to focus light on the iris plane using average anterior-segment parameters (from 100 left eyes14 measured with clinical SD-OCT; VeliteC3000, Weiren Meditech, Foshan, Guangdong, China). Here, we further calibrated its imaging performance (spot radius) and lateral/axial distortions using a dataset of pre-operative right eyes (from 134 Asian subjects47 measured with corneal topography; Pentacam, Oculus). Figure 6(e) shows the root-mean-square (RMS) spot radius across the scan field (±5.3mm). The theoretical diffraction-limited Airy disc radius (0.61λ/NA) ranged from 11.02 to 12.5μm. As the measured mean RMS spot radius (2.4 to 14.5μm) was typically smaller than the Airy disc, the system achieved diffraction-limited lateral performance. Figures 6(f) and 6(g) characterize lateral and axial distortions: lateral distortion arises from field-dependent scan angle/length relationships, whereas axial distortion stems from optical path length variations among scan beams. Individual anterior-segment anatomy (especially corneal structure) causes personalized distortion variance. Maximum distortions were (0.27±0.09)mm (lateral) and (−0.50±0.08)mm (axial) at ±5.3mm positions. Calibration using mean values reduced residual distortions to ±0.09mm (lateral) and ±0.08mm (axial). Finally, Fig. 6(h) demonstrates a representative corner-to-corner iris image (10-frame average), clearly showing anatomical features.

In OCT systems, galvanometer scanners are typically used to create two-dimensional scan patterns such as raster, radial, Lissajous,48 spiral,39 and circular scans, along with hybrid patterns such as the spiral–circular ammonite scan.38Figure 7 compares three scanning patterns relevant to this study. Raster scan [Fig. 7(a)] is the most common OCT scan pattern where the fast-axis (X) scanner uses a sawtooth waveform, whereas the slow-axis (Y) scanner uses stepwise motion [Fig. 7(b)]. A square wave indicates the data acquisition period—synchronized with the linear segment of the sawtooth’s forward scan while excluding nonlinear and retrace phases, yielding a duty cycle of 80% to 90% in OCT imaging.12

Spiral scan method [Figs. 7(c) and 7(d)] enables 100% duty cycle efficiency through continuous center-to-periphery motion driven by sinusoidal X/Y waveforms with π/2 phase difference and increasing amplitudes. This approach has been implemented in OCT structural imaging39; however, it remains unsuitable for OCTA imaging due to its continuously expanding scan path, which prevents acquisition of repetitive B-scans within the short time required for OCTA motion contrast.

The circular scan method [Fig. 7(e) and 7(f)] can provide repeated scans for OCTA imaging. However, abrupt transitions between circles cause sudden changes in driving waveforms [Fig. 7(f)], requiring extra time either to accommodate the circle transition or to stabilize the scanners. The duty cycle of the circular scan can be calculated as Duty Cycle=∑k=1MNk×i/fA∑k=1MNk×i/fA+∑k=1M−1Tk×100%,(4)where i represents the repeated measurements and Tk is the additional time between the k’th and (k+1)’th circles, which can be set as Tk=(jk−1)/fA, jk≥1. We typically set jk=2 to stabilize the scanner with one A-line time per transition that can be removed from the imaging sequence [Fig. 7(f)].

Figures 8(a) and 8(b) compare the OCT images of a 1 mm grid target before and after distortion correction using the active phase compensation [Eq. (3)]. The 13 mm scan field comprised 270 circles with 12,000 sampling points each (k=1,…,270). After applying the frequency-dependent phase compensation, distortion was significantly reduced—particularly in the central region.

Acknowledgment. This work was supported by the National Key Research and Development Program of China (Grant No. 2021YFF0502900), the National Natural Science Foundation of China (Grant Nos. 62575066 and 62027824), the Guangdong Basic and Applied Basic Research Foundation (Grant No. 2024A1515011344), the Innovation and Entrepreneurship Teams Project of Guangdong Pearl River Talents Program (Grant No. 2019ZT08Y105), the Guangdong-Hong Kong-Macao Intelligent Micro-Nano Optoelectronic Technology Joint Laboratory (Grant No. 2020B1212030010), and the National Institutes of Health/National Eye Institute (NIH/NEI) (Grant Nos. P30EY07551, R01EY022362, and R01EY022362).

Biographies of the authors are not available.