Acoustic Radiation Force Optical Coherence Elastography of the Crystalline Lens: Safety

Abstract

Purpose: To assess the safety of acoustic radiation force optical coherence elastography in the crystalline lens in situ.

Methods: Acoustic radiation force (ARF) produced by an immersion single-element ultrasound transducer (nominal frequency = 3.5 MHz) was characterized using a needle hydrophone and used for optical coherence elastography (OCE) of the crystalline lens. Preamplified signals at 50, 100, 250, 500, 750, 1000, and 1250 mV peak amplitude were tested on ex vivo porcine eyes (n = 21). Three-dimensional optical coherence tomography (OCT) and confocal microscopy images were acquired before and after ARF exposure to each signal amplitude to determine damage.

Results: The acoustic intensity of the ultrasound transducer at 100-mV preamplified peak amplitude input demonstrated a signal-to-noise ratio high enough for tracking elastic wave propagation in the lens and spatial-peak pulse-average (SPPA) intensity of 24.1 W/cm² and mechanical index (MI) of 0.46. The SPPA intensity was lower than the U.S. Food and Drug Administration (FDA) safety limit (28 W/cm²), but the MI was twice the safety limit (0.23). OCT structural and confocal microscopy images showed damage only at levels exceeding 1150 W/cm² and 3.2 for SPPA intensity and MI, respectively.

Conclusions: OCT and confocal microscopy showed that, even when the intensity exceeded FDA recommendations (>100 mV), no noticeable damage was observed. Although a further reduction in acoustic intensity is necessary to meet FDA safety limits, ARF-based elastography shows promise for safe clinical translation in quantitatively characterizing lenticular biomechanical properties.

Translational Relevance: This work assessed the safety standards for acoustic radiation force to be used in human lens elastography according to the FDA safety limits.

Introduction

The crystalline lens is a biconvex transparent multilayered structure with a complex lens fiber cell architecture within an outer encapsulating membrane.¹ The lens fiber cells contain densely packed crystalline proteins that contribute to its transparency, transmitting and focusing light on the retina.² The ability of the lens to accurately focus light from different distances on the retina relies on reflexive dynamic contractions of the ciliary muscles (a mechanism referred to as accommodation) which reshapes the lens and adjusts its refractive power.³ The lens experiences mechanical and geometrical modifications throughout the accommodation, and its ability to accommodate degrades during aging, leading to the loss of near visual function, a condition that is termed presbyopia.¹ As the lens ages, it also becomes stiffer, which has been postulated as the primary source of age-related accommodation loss.⁴ 

Therefore, lens biomechanical changes associated with aging^5,6 can help us understand the mechanism of presbyopia, particularly the role of lens stiffness. Being able to measure the lens biomechanical properties would also help assess the efficacy of potential treatment options for presbyopia based on decreasing lens stiffness.⁷ Several elastography methods have been used to determine crystalline lens biomechanical properties. Methods such as ultrasound elastography,⁶ Brillouin microscopy,^8,9 and optical coherence elastography (OCE)¹⁰ have revealed differences in the lens biomechanical properties in a diversity of animal models and under different physiological conditions. OCE¹¹ is well suited for imaging the mechanical properties of the lens due to its noninvasive nature and high displacement sensitivity (nanometer scale¹²) because it relies on optical coherence tomography (OCT).¹³ However, the lens location inside the eye globe represents a challenge in generating lens deformation. One commonly utilized excitation modality for the lens is acoustic radiation force (ARF),^5,14–16 which can provide pinpointed tissue stimulation of the lens through the cornea and aqueous humor. ARF has been used to determine the biomechanical properties of other inner ocular structures, such as the iris,¹⁷ retina,¹⁸ and even whole eye globe.¹⁵ Thus, ARF is a popular tissue stimulation method for determining the biomechanical properties of ocular tissues¹⁹; however, the safety of ARF excitation on the lens has not been well characterized. The U.S. Food and Drug Administration (FDA) recommends the use of acoustic field emissions lower than 28 W/cm² spatial-peak pulse-average intensity (I_SPPA) and a mechanical index (MI) of 0.23²⁰ for ophthalmic ultrasound exposure. Ultrasound exposure levels have been demonstrated to be safe in human subjects using exposure levels close to the recommended levels during imaging but not elastography.²¹ However, high-intensity focused ARF can cause potential physical damage to delicate ocular tissue, which limits the translation of this approach for elastography in ophthalmology practice.²² 

Clinical investigations of OCE are rapidly expanding,²³ and ARF may be well suited for clinical assessments of lens viscoelasticity. This study aimed to assess the safety of ARF as a mechanical excitation method for OCE in the crystalline lens and provide insights into the potential hazards of high-intensity ARF with regard to the ocular lens. 

Methods

Acoustic Intensity Characterization

A needle hydrophone (NH0200; Precision Acoustics, Dorset, UK) with a 0.2-mm sensor diameter and a sensitivity of ∼72.5 mV/MPa at 3.5 MHz was used to measure the acoustic pressure of a focused immersion ultrasound transducer (V382-SU; Olympus, Tokyo, Japan) that had a focal length of ∼19 mm and central frequency of 3.5 MHz. A function generator (DG4162; Rigol Technologies, Suzhou, China) generated a rectangular pulse of 0.5-ms duration, amplified by a radiofrequency power amplifier (1040L; Electronics & Innovation, Rochester, NY) with a gain of 55 dB. The amplified signal then drove the immersion transducer which generated a focused ultrasound pressure field. The hydrophone detected the pressure field, producing an electrical signal (voltage) that was subsequently received by a digital oscilloscope (DS4000; Rigol Technologies, China) (Figs. 1A, 1C). The hydrophone was aligned such that the peak pressure was detected to ensure that the hydrophone measurements were at the focus of the transducer. The preamplified driving signal was produced with a variable peak amplitude voltage from 25 mV to 500 mV with a step size of 25 mV. The reported values from the oscilloscope were recorded to calculate acoustic pressure (P) as \(P = \frac{{Voltag{e_{hydrophone\ measured}}}}{{Sensitivit{y_{hydrophone}}}}\) and intensity (I) as \(I = \frac{{{P^2}}}{{\rho c}}\), where ρ is the density of water (1000 kg/m³) and c is the speed of sound in water (1540 m/s). The I_SPPA was calculated at the focal point of the transducer as an average intensity for one ultrasound pulse \({I_{SPPA}} = \frac{1}{T}\mathop \smallint \limits_0^T \frac{{{P^2}}}{{\rho c}}dt\),²⁴ where T is the total duration of the acoustic radiation force pulse. The mechanical index was calculated as \(MI = \frac{{Pr}}{{\sqrt f }}\), where Pr is the peak negative pressure and f is the central frequency of the transducer (3.5 MHz).²⁵ 

Figure 1.

 

ARF characterization. (A) Schematic representing characterization of the acoustic field emission of the immersion transducer using a needle hydrophone. (B) Diagram of the crystalline lens exposure to ARF during OCE measurements. (C) The voltage profile delivered by the ultrasound immersion transducer measured by hydrophone at 100-mV peak amplitude preamplified input.

In Situ Experiments

Porcine eyes (n = 21) enucleated less than 24 hours prior to the ARF-OCE measurements were utilized in the whole eye globe configuration. External tissues, such as eyelids and muscles, were removed while preserving all other ocular tissues, including the cornea. The experiments were conducted using a previously described OCE system (Fig. 1A).¹⁴ For mechanical excitation, the immersion ultrasound transducer was attached to the OCT system scan head, and the ultrasound push was focused on the lens through the cornea, as shown in Figure 1B. The driving signal to the ultrasound transducer was the same as in the acoustic intensity characterization described earlier. The crystalline lenses were exposed to 50, 100, 250, 500, 750, 1000, and 1250 mV preamplified peak amplitudes. In each case, the lens was exposed to a train of pulses during OCE imaging. OCE imaging was performed with M–B-mode imaging over M = 251 measurements with N = 1000 A-lines in one direction (13.04 mm on the x-axis). The camera rate was set to 25 kHz during all measurements. Before and after the ARF exposure, three-dimensional (3D) OCT images were acquired for structural assessment of the lens integrity. To analyze the potential damage in the lens surface, the x,y plane maximum intensity projection (MIP) was analyzed using the ImageJ²⁶ Analyze Particles tool (National Institutes of Health, Bethesda, MD), where the image intensity was normalized using a maximum entropy autothreshold that estimated the percentage of the white area by default. The white area percentages before and after ARF-OCE exposure were calculated for each case.²⁷ For 100, 1000, and 1250 mV, the crystalline lenses were dissected and imaged using confocal reflectance microscopy (HRT3; Heidelberg Engineering, Heidelberg, Germany) to assess their inner structure and estimate the physical damage. 

Results

The acoustic field produced by the ultrasound immersion transducer was characterized using the needle hydrophone at a range of different input voltages that were used to calculate the pressure, intensity (I_SPPA), and MI, as shown in the Table and Figure 2. The preamplified peak amplitude of 100 mV was within the limit recommended by the FDA in terms of I_SPPA (28 W/cm²). For the MI, however, the preamplified peak amplitude of 50 mV was within the FDA limit (0.23). 

Table.

 

Acoustic Field Characteristics of the Immersion Ultrasound Transducer

 

Figure 2.

 

(A, B) Profiles of I_SPPA (A) and MI (B) as functions of the input intensity (mV). Dashed red line indicates FDA safety thresholds for I_SPPA and MI.

The different input voltages were also analyzed with OCE to determine its capacity to detect wave propagation at the lens surface. At 0.3 mm from the excitation (focal) point in the crystalline lens surface (Fig. 3A), preamplified peak amplitude voltage inputs from 50 mV to 250 mV with a step size of 50 mV were analyzed to image and characterize wave propagation (Figs. 3B, 3C). The peak signal-to-noise ratio (SNR) was calculated comparing each input intensity (Fig. 3D) as SNR = 20log(V_peak/STD_noise), where V_peak is the maximum amplitude of the particle velocity for a given excitation and STD_noise is the standard deviation of the noise after the excitation (Fig. 3C). 

Figure 3.

 

Assessment of wave propagation with different preamplified peak amplitudes (from 50 mV to 250 mV with a step size of 50 mV). (A) Localization (yellow arrow) of the ARF focal point to produce mechanical excitation at the crystalline lens surface. (B) Wave propagation snapshots for each input voltage at a time ∼4 ms after excitation. (C) Analysis of the axial particle velocity stationary temporal profiles at 0.3 mm from the ARF excitation for each of the input voltages. (D) SNRs calculated for each input voltage at 0.3 mm from the ARF excitation zone.

As shown in Figure 4, which shows en face MIPs of the 3D-OCT data at 50, 100, 250, 500, and 750 mV, there was no apparent damage to the crystalline lens structure caused by ARF (lenses with minimal damages due to storage were still considered); however, using 1000 mV and 1250 mV, there was clear damage of the lens surface represented by the appearance of black dots after ARF exposure. The presence of these potentially damaged areas in the crystalline lens increased the white area ratio before and after exposure to ARF in the crystalline lens (Fig. 4H). 

Figure 4.

 

(A–G) 3D-OCT MIPs of the crystalline lens before and after ARF exposure to 50 mV (A), 100 mV (B), 250 mV (C), 500 mV (D), 750 mV (E), 1000 mV (F), and 1250 mV (G) preamplified peak voltage signals. (H) Representative B-mode image of ARF damage after 1000-mV exposure. (I) Ratios of the white area percentage before and after ARF excitation for each case. Scale bar: 1 mm.

The damage observed at 1000 mV and 1250 mV was further assessed using confocal reflectance microscopy (Fig. 5), where the damage was evident in the inner structure of the lens. After 1250-mV ARF exposure, there was significant disruption of the lens fiber alignment; in contrast, at 1000 mV, the fibers were still aligned but there was still a notable disruption in the lens integrity. 

Figure 5.

 

(A, B) Typical confocal microscopy images depict the inner crystalline structure in the anterior Y-suture before ARF exposure (A) and after 100-mV exposure (B). (C, D) Structural crystalline damage in the anterior Y-suture after 1000-mV exposure (C) and after 1250-mV exposure (D). Scale bar: 50 µm.

Discussion

OCE has widely utilized ARF for ophthalmological research. It has been used to characterize a several intraocular tissues, including the iris,¹⁷ lens,⁵ and cornea.²⁸ Although there has been a recent introduction of air-coupled ARF tissue stimulation,^19,28 we focused our analysis on water-coupled ARF because air-coupled ARF deposits significantly less energy into the tissue and cannot specifically target internal tissues such as the crystalline lens. Nevertheless, ARF-based OCE has been used in vivo without presenting any obvious damage to the tissues.^18,29,30 This is because of the low intensities required to produce mechanical excitation in the tissue that is detectable by OCE, as shown in Figure 3 (nanometer to micrometer scale).³¹ However, no explicit analysis of ARF safety has been reported, specifically in the crystalline lens. 

In this study, we focused on I_SPPA and MI as safety characteristics for ARF-based elastography of the lens. We demonstrated that temporal stationary particle velocity profiles with high SNRs, suitable for estimating the velocity of the elastic wave, can be achieved within the safe I_SPPA limit of 28 W/cm², specifically at 24.1 W/cm² for a 100-mV preamplified peak amplitude, as shown in the Table. The SNR of the particle velocity profile at 100 mV preamplified peak amplitude is well above 20 dB, which is the minimum threshold necessary for accurately determining group velocity, as demonstrated in our previous study.³² However, at this intensity level, the MI value was twice the permissible limit for ophthalmological applications, as it reached 0.46. To achieve an MI less than 0.23, a 50-mV preamplified peak amplitude is required where the SNR is too low for reliable elastography measurements. Therefore, additional efforts are necessary to reduce the mechanical index further and satisfy FDA safety requirements for diagnostic imaging. One potential method to improve the SNR is to use harmonic narrowband low-intensity acoustic pulses. This approach allows noise to be filtered out using a frequency filter, enabling successful OCE imaging at lower excitation energies. Furthermore, using a common path interferometer³³ and/or two-beam³⁴ OCE methods can reduce noise caused by environmental factors or sample motion, potentially enhancing the displacement sensitivity of the system at lower excitation energies. For in vivo measurements, bulk motion artifacts associated with eye movement can be further minimized using various postprocessing techniques that involve structural and phase images of the lens, as discussed in previous studies.^35,36 In this study, we did not apply any de-rating of acoustic pressure even though the acoustic energy propagates through the cornea and aqueous humor to reach the lens. For humans, the corneal thickness is approximately 500 µm, and, with a standard attenuation coefficient of 0.3 dB/MHz/cm, the pressure attenuation is insignificant (<1%). The attenuation in the aqueous humor is even more insignificant as it is predominantly water. 

In addition to I_SPPA and MI, the FDA safety standard also includes spatial-peak temporal-average intensity (I_SPTA) and thermal index (TI), representing possible temperature changes.³⁷ The I_SPTA, which is the time average of the instantaneous intensity during the ultrasound exposure, is defined as I_SPTA = DC · I_SPPA, where DC is the duty cycle. We did not consider and did not optimize I_SPTA in this study because the duty cycle can be easily controlled. For example, a lower pulse-repetition frequency could be utilized, which would only result in a longer experimental time. In this study, each M-mode image was 42 ms, where 1000 A-lines were obtained at the 25-kHz camera rate and 2-ms delay to settle the scan mirror. Considering 0.5-ms pulse duration, the duty cycle was approximately 0.012, and I_SPTA for 100-mV preamplified amplitude was about 287 mW/cm². Although this value significantly exceeds the FDA safety standards (50 mW/cm² for Track 3 devices²⁵), the scope of this paper is the estimation of the safe limits for I_SPPA, and we used a high-duty cycle to decrease the experimental time. In clinical use, however, control of I_SPTA is critical. High I_SPTA is connected to temperature increases in tissue and high TI values. Hoerig et al.²¹ developed an ultrasound approach capable of detecting structural modifications in the inner sclera on myopia in vivo human eyes, where a value of 50 mW/cm² I_SPTA was considered the maximum safety threshold. Higher I_SPTA levels (>1000 W/cm²), such as the ones used in oncology, could cause thermal cavitation and potentially damage tissue.²⁴ 

Ophthalmic exposure to high-intensity ultrasound could result in partial loss of vision by damaging the lens. High-intensity focused ultrasound (HIFU) has been reported to cause iris atrophy, lens opacity (peripheral cataract), and corneal opacity^37,38 when used in the eyelid. The cosmetic HIFU intensity is ∼8 times higher than that allowed by ophthalmic recommendations.³⁹ As shown in Figure 4, high intensities (1000-mV and 1250-mV preamplified amplitudes) can disrupt the lens surface and structure, even affecting its internal fibers (Fig. 5D). At the same time, lower intensities do not introduce noticeable immediate damage to the lens ex vivo, as seen in Figure 4. 

A significant limitation of this work is that the experiments were conducted ex vivo, which may introduce additional effects on the lens due to ARF excitation outside of its natural in vivo environment. In vivo studies utilizing an animal model with long-term imaging will be the next step of our work, and these studies are currently underway. 

Conclusions

We determined the short-term tissue response for ARF excitation in the crystalline lens ex vivo. Further measurements were conducted on porcine lenses in situ within the whole eye globe configuration, using energy levels that were both fractions and multiples of the established safety standards. ARF-based OCE imaging at an I_SPPA level of 24.1 W/cm² and an MI of 0.46 provided an acceptable SNR to capture wave propagation in the lens. Although the intensity was within the FDA safety limit, the mechanical index exceeded this limit by a factor of two. No acute structural damage to the lens was observed until ARF stimulation reached an intensity of 1150 W/cm² and an MI of 3.2. These results indicate that ARF-OCE has the potential for safe clinical translation for the quantitative characterization of lenticular biomechanical properties. 

Acknowledgments

Supported in part by grants from the National Eye Institute, National Institutes of Health (R01EY030063 and core grant P30EY07551). 

Disclosure: C. Zevallos-Delgado, None; T.T. Mekonnen, None; C. Duvvuri, None; L. Rohman, None; J. Schumacher, None; M. Singh, ElastEye (F); S.R. Aglyamov, None; M.D. Twa, None; J.-M. Parel, None; G. Scarcelli, None; F. Manns, None; K.V. Larin, ElastEye (F)