2.2.1.

Fresh porcine vocal folds

We acquired fresh porcine airway specimens from a local slaughterhouse and isolated the larynx in a room temperature saline bath within 2 h after sacrificing the animal. Porcine vocal folds have layered LP very similar in structure to human vocal folds. Although some preliminary studies showed the presence of a prominent subepithelial collagen layer in the SLP, there is no agreement whether the superior or inferior LP of the porcine is primarily responsible for phonation.⁴² Therefore, we decided to use both the superior and inferior porcine vocal folds for our experiments to test whether differences in their microanatomy correlate in meaningful ways with differences observed in laser ablation and imaging.

After excision, we placed each vocal fold in saline and covered it with a glass cover slip (0.17-mm thickness) to flatten the surface. In a clinical application, the glass cover slip would be akin to having the window of a microsurgery probe in contact with the sample, and thus helping to maintain a constant depth of ablation. Short-lived cellular autofluorescence of fresh tissue samples guided us to identify the surface of the sample during imaging.

For follow-up histology, we placed the superior and inferior vocal fold samples in 10 mL of 10% formalin (SF98-4, Fisher Scientific, Pittsburgh, PA) and stored them at 4°C for at least 48 h prior to paraffin embedding, sectioning, and staining (TherapeUTex, Austin, Texas). The dehydration process was performed with ethanol and Citrisolv. A combination Masson-Trichrome and Verhoeff staining was used to differentiate elastin and collagen.⁴³

2.2.2.

Hamster cheek pouch mucosa as a scar model

We used adult male Golden Syrian hamsters (Charles River Labs, Wilmington, Massachusetts) of 100 to 120 g body weight as an animal model, in which scar can be created within an easily accessible mucosal surface. The goal in these experiments was to observe how the structural differences between normal and scarred buccal mucosae affected ablation results. To scar the cheek pouches, we first anesthetized the animals by injecting a mixture of ketamine ($200\mathrm{mg}/\mathrm{kg}$) and xylazine ($8\mathrm{mg}/\mathrm{kg}$) intraperitoneally. We then cauterized 5 to 10-mm-diameter circular areas of one cheek pouch with an electrocautery unit (Conmed Saber, Utica, NY, 2400). The hamsters were euthanized after a survival period of 1 month using 0.5 ml intraperitoneal euthasol. The scarred and contralateral normal cheek pouches were removed and mounted on rubber test-tube stoppers with fine needles to hold the cheek pouch mucosa flat. After rinsing with saline, the tissue was frozen in isopentane cooled in liquid nitrogen. The cheek pouch tissue was prepared at Massachusetts General Hospital Voice Laboratory (MGH) in Boston and shipped on dry ice to the Ben-Yakar Laboratory at the University of Texas at Austin for bench-top testing. After delivery, the cheek pouches were stored at $-80°\mathrm{C}$. For each experiment, we thawed the cheek pouches in saline solution and covered them with a glass cover slip to flatten their epithelial surface.

For the histological analysis of the healthy and scarred tissues, we placed them in 10 mL of 10% formalin (SF98-4, Fisher Scientific, Pittsburgh, PA) and stored them at 4°C prior to paraffin embedding, sectioning, and staining (TherapeUTex, Austin, Texas). For cheek pouches, an additional dehydration step with ethanol and xylene was added due to the abundant fat content in this tissue. Masson-Trichrome staining was used to show the orientation of the collagen fibers.⁴³

3.2.1.

Ablation versus pulse energy

We previously showed that ultrashort pulses with submicrojoule energies can create voids below the epithelium layer using 0.75 NA, $20\times$ air objective.¹⁴ Here, we studied how the size of these voids varies with the pulse energy and how scarring affects the ablation properties.

First, we verified how well nonlinear images correlated with the extent of the ablated voids by comparing them with histology images of H&E-stained tissue. Figure 7 presents this comparison for two different pulse energies including a near-threshold energy. Overall, there was a good correlation between the TPAF and the histology images in terms of the extent and morphology of voids for each pulse energy. The void ablated with the near-threshold energy of 250 nJ appeared incomplete, and the one ablated with the higher energy of 750 nJ was more complete and uniform. Consistent with our previous study,¹⁴ histology images showed slightly smaller voids mainly in their heights because of three potential reasons: (1) contraction during the histology preparation, (2) focal-point shift in the TPAF images due to the deflection of the beam at the tissue-void interface, as a result of the index of refraction mismatch between tissue and gas, and (3) shrinkage of voids during the period between TPAF imaging and tissue fixation. The heights of the voids were 40% to 50% smaller in the histology images, as a result of the reasons mentioned above.

Fig. 7

Comparison of cross-sectional TPAF images with histological analysis of voids in an inferior porcine vocal fold. The voids were ablated using (a) 250 nJ and (b) 750 nJ pulse energies at a depth of 90 μm beneath the tissue surface. Pre-ablation TPAF, post-ablation TPAF, and H&E-stained histology images are presented in the first, second, and third rows, respectively. Arrows in the TPAF images indicate the plane targeted for ablation. Arrows in the histology images indicate the location of the identified voids, corresponding well with the TPAF images. All scale bars are 100 μm.

We then studied ablation properties of healthy and scarred tissue samples. Figure 8 illustrates representative TPAF images of porcine superior and inferior vocal folds 1 min after ablation with different incident pulse energies. The ablation depth was 90 μm, where collagen fibers covered the whole FOV, and the plane of focus was completely below the epithelium within the SLP. We chose the ablation size ($250\times 250{\mu \text{m}}^2$) smaller than the FOV for imaging ($430\times 430{\mu \text{m}}^2$) to visualize the entire region of interest before and after ablation. The black areas with no signal presumably represent voids that are filled up with a temporary bubble containing gaseous products of the vaporized ablated tissue. At low-pulse energies from 250 up to 500 nJ, the ablation was partial and could not cover the whole targeted area, and the bubbles appeared small and scattered. Pulse energies of 750 nJ and above provided a more uniform and complete ablation. Inferior porcine vocal folds repeatedly required slightly less energy to obtain complete ablation.

Fig. 8

The effect of incident laser pulse energy on the subsurface ablation properties of porcine vocal folds. The TPAF images show the characteristics of subepithelial voids (bubbles) 1 min after ablation with different pulse energies (as measured at the surface) in superior (a) and inferior (b) vocal folds at the ablation depth of 90 μm. Scale bar indicates 100 μm.

Figure 9 shows representative SHG images of control and scarred hamster cheek pouch samples 1 min after ablation, performed at a depth of 60 μm, where the SHG signal covered most of the FOV. While laser pulse energies of 150 nJ were just enough to initiate partially ablated spots in the control samples, they resulted in more uniformly ablated regions in scarred tissue. Scarred tissue repeatedly required less energy to obtain complete ablation.

Fig. 9

The effect of incident laser pulse energy on the subsurface ablation properties of hamster cheek pouches. SHG signal images of subepithelial voids (bubbles) for different incident energies 1 min after ablation in control (a) and scarred (b) hamster cheek pouches at the ablation depth of 60 μm. Scale bar indicates 100 μm.

3.2.2.

Sample-to-sample repeatability of tissue ablation

To understand how the ablation characteristics of a specific tissue type vary from sample-to-sample, we studied five sample sets of fresh porcine vocal folds (five samples for superior and five samples for inferior) and three sample sets of hamster cheek pouches (three samples for control and three samples for scarred) at three different locations per pulse energy per sample. We measured the initial area of the bubble at the targeted plane immediately after ablation for each incident pulse energy varying between 50 and 1500 nJ. Figure 10 represents the plots of these measurements as normalized to the size of the targeted ablation area of $250\times 250{\mu \text{m}}^2=0.0625{\mathrm{mm}}^2$. A unit area is considered to indicate a complete ablation. As a general trend, the size of the bubbles gradually increased with the laser pulse energy and covered the whole targeted ablation area at higher energies.

Fig. 10

Normalized area of the initial bubbles in the ablated voids at different incident pulse energies 1 min after ablation. Each data point represents an average of (a) 15 measurements in five fresh porcine vocal folds and (b) nine measurements in three hamster cheek pouches. The error bars represent the standard deviation of the mean of the samples.

Figure 10(a) indicates that inferior vocal folds generally required less energy than the superior to create the same size of bubbles. The area of the initial bubble matched the targeted area as the incident pulse energy approached to 1.0 and 1.5 μJ in the inferior and superior vocal folds, respectively, and exceeded the targeted area beyond those energy levels. The difference in the initial bubble area was also distinctly close to the threshold energy. For example, ablation at 250 nJ resulted in minimally observable bubbles in the superior folds and in slightly larger bubbles, covering more than 40% of the targeted area, in the inferior. The ratios of the pulse energies for complete (100% of the area) and partial (40% of the area) ablations were approximately 4 for both types of vocal folds.

Figure 10(b) compares the bubble area results for healthy and scarred hamster cheek pouch samples. As shown in the previous section, the minimum energy required for ablation in this tissue was lower than for porcine vocal folds. Interestingly, however, the scarred tissue required less energy than the healthy tissue to initiate ablation at 60 μm depth. Close to threshold, where 40% of the area is ablated, ablation initiated in the scarred tissue with incident energies as low as 50 nJ, at which no bubble appeared in the healthy tissue. The initial bubble area approached the targeted area between 600 and 700 nJ pulse energy, while it was consistently larger in scarred samples for the same energy levels.

The standard deviation of each data presented in Figure 10 shows noteworthy differences in the areas of the initial bubbles. Nonetheless, we could extract general trends of each tissue type, as the error bars of each tissue type did not significantly overlap with its counterpart (inferior versus superior and control versus scarred). To summarize, these results indicated that it was easier to ablate scar tissue than healthy tissue, and it was easier to ablate inferior vocal fold than the superior vocal fold. In the next section, we will discuss the hypotheses explaining these differences in ablation thresholds.

3.2.3.

Ablation threshold and tissue scattering properties

The differences in the incident energy thresholds for different tissue types might be a result of one or both of the following factors: (1) the scattering length of the tissue that determines the portion of the incident energy reaching the focal volume and (2) the sensitivity of the specific tissue to ablation, which we refer to as tissue-specific ablation threshold—TSAT fluence ($F_{\mathrm{th}}$). Independent measurement of these two parameters is not straightforward, as they depend on each other. While it is easy to measure the incident pulse energy at the surface that is required to initiate ablation at a given depth, the calculation of the energy reaching the actual ablation depth requires knowledge of the extinction properties of the tissue. Furthermore, the extinction properties might vary with depth as the light passes through different layers, e.g., from the epithelium into the LP. We therefore developed a method to measure ablation threshold fluence and extinction properties of tissue simultaneously in a single set of experiments.

In our approach, we assumed that scattering dominated the extinction properties of tissue in the NIR wavelengths and is different in the epithelium and LP but constant within each layer. This assumption is valid, since the absorption coefficients are expected to be 3 orders of magnitude smaller than the scattering coefficients as observed in skin epidermal measurements.^48,49 We also assumed that the TSAT was independent of depth within the LP, and thus should remain constant in the targeted regions of subepithelial ablations. Finally, we assumed that Beer’s law could be used to estimate the attenuation of light reaching the focal volume, since the nonlinear ultrafast ablation processes are mostly governed by the ballistic photons reaching the focal plane.

In the light of these assumptions, we measured the minimum incident energy required to initiate ablation at three ablation depths, $z_{\mathrm{ab}}$, and used the Beer’s law to describe the exponential beam attenuation with depth for each case:

Here, the pulse energy reaching the focal plane at the ablation depth decays exponentially with the effective scattering length, ${\ell }_{s,\mathrm{eff}}$:

We could then estimate the three unknown parameters: TSAT fluence at the focal plane, $F_{\mathrm{th}}$, epithelium scattering length, ${\ell }_{s,\mathrm{ep}}$, and LP scattering length, ${\ell }_{s,\mathrm{LP}}$ by solving Eq. (1) at three depths. The radius of the laser beam at the focal plane was $w=0.7\mu \text{m}$, and the local epithelium thickness, $z_{\mathrm{ep}}$, was measured via the nonlinear images for each ablation data point.

To find the three unknowns in our experiments, we performed ablations at three different depths (72, 90, and 108 μm for porcine vocal folds and 60, 78, and 96 μm for hamster cheek pouches). At each depth, we measured the local $z_{\mathrm{ep}}$ and found $E_{\mathrm{th},\text{surface}}$, which is defined as the energy required to create an initial bubble covering 40% of the targeted area. For each tissue type, we used two samples, and for each depth, we collected data at three different locations to obtain an average value. The results are tabulated in Table 1.

Table 1

Summary of tissue optical and mechanical properties and ablation thresholds. Epithelium thickness (zep), scattering lengths of epithelium (ℓs,ep) and superficial lamina propria (SLP) (ℓs,LP), the resultant effective scattering length at the ablation depth (ℓs,eff), and tissue-specific ablation threshold—TSAT fluence (Fth) for each tissue type for 776 nm light and 1/e2 focused beam spot diameter of 1.4 μm. The ablation depth for vocal folds was 90 μm and for cheek pouches was 60 μm. The last four columns summarize the observed trends for the content of elastin and collagen fibers of each tissue type with their interpreted stiffness and tensile strength.

These data shed light on the ablation threshold results observed in the previous section. We found that cheek pouches to be more scattering than porcine vocal folds (${\ell }_{s,\mathrm{eff}}=25\mu \text{m}$ versus 33 μm), but they had $\sim 3.5\text{times}$ lower ablation threshold fluence than the porcine vocal folds, an average $F_{\mathrm{th}}$ of $0.5\mathrm{J}/{\mathrm{cm}}^2$ for cheek pouches versus $1.75\mathrm{J}/{\mathrm{cm}}^2$ for vocal folds. While the scattering lengths were different, the normalized depths presented in Figs. 8–10 were nearly constant ($z_{\mathrm{ab}}/{\ell }_{s,\mathrm{eff}}=2.4$ and 2.7 for cheek pouches and vocal folds, respectively) considering the different ablation depths. We could therefore conclude that the difference in the ablation threshold fluences was the main factor for the lower incident energies, $E_{\mathrm{th},\text{surface}}$, required to ablate hamster cheek pouches (50 to 150 nJ) as compared with those required for porcine vocal folds (250 to 500 nJ) to initiate ablation.

The data in Table 1 also assist in interpreting the lower incident energies required for scarred cheek pouches. In comparing normal and scarred hamster cheek pouch mucosae, surprisingly we found that the scarred samples showed similar scattering properties as normal tissue. Thus, the main reason for lower incident ablation energy threshold in scarred mucosa observed in the previous section could also be related to the TSAT fluence that was about 30% lower than that for normal tissue.

Considering the ablation and scattering properties measured in this set of experiments, we plotted the maximum ablation depths that are achievable for a given incident pulse energy for each tissue type in Fig. 11. For the maximum pulse energy of 1.5 μJ available in our system, the maximum ablation depth is approximately $5{\ell }_{s,\mathrm{eff}}$ in cheek pouches and $3.5{\ell }_{s,\mathrm{eff}}$ in vocal folds for creating partial ablation (where initial bubbles cover 40% of the targeted area). Since the ablation threshold is higher for a more complete ablation, the maximum ablation depth will further decrease. For example, for the ablation of 60% to 80% of the target area, maximum ablation depth may be roughly $4{\ell }_{s,\mathrm{eff}}$ for cheek pouches and $3{\ell }_{s,\mathrm{eff}}$ for vocal folds, as shown in dashed lines in Fig. 11. In principle, one could ablate deeper with the availability of larger pulse energies. However, as the pulse energy increases, other nonlinear phenomena, such as self-focusing, might initiate and cause out-of-focus damage in the form of filamentations.⁵⁰ We can estimate that self-focusing in scarred cheek pouches might occur when ablating at depths of $\sim 6$ to $7{\ell }_{s,\mathrm{eff}}$ with $\sim 10\text{-}\mu \text{J}$ pulse energies using 3-ps NIR laser pulses.

Fig. 11

Minimum pulse energy at the surface to initiate ablation at different depths, and the resultant maximum ablation depth dependence on pulse energy. Three experimental data points for each tissue type and the resultant fits according to Eq. (1) considering the calculated ablation and tissue properties are tabulated in Table 1. Solid lines represent estimates for the ablation of 40% of the targeted area, and dashed lines represent estimates for a more complete ablation of 60% to 80%. Vertical black-dashed line indicates the maximum pulse energy of 1.5 μJ that can be delivered to tissue surface in our setup.

This ablation depth limitation might be overcame by introducing novel techniques, such as temporal focusing, to reduce the out-of-focus peak intensities.^51,52 In simultaneous spatial and temporal focusing, a dispersive element, such as a grating, is used to spatially separate the spectral components of the ultrafast laser pulse.⁵³ The resulting pulse is temporally as well as spatially dispersed, resulting in a dramatically reduced peak intensity. Focusing by the objective lens serves to spatially confine the spectrally dispersed beam back to a diffraction-limited spot, and thus recombining the spectral pulse components and restoring the ultrafast pulse duration at the focus. The reduced peak power of the pulse during propagation reduces undesirable nonlinear effects, such as self-focusing and filamentation, thereby improving the maximum ablation depth and axial confinement.

3.2.4.

Bubble dynamics in ablated voids

In our previous studies, we have found that the initial bubbles formed right after ablation, shrank to a thin layer within 45 min. Figure 12 shows representative time-lapse images of the side view of such a bubble in porcine vocal fold, where the tissue was illuminated from the left side. The images show a gradually shrinking bubble, as the vaporized tissue diffused out. As the bubble shrinks, it leaves behind a thin layer of ablated void, which is indicated as a thin layer of dark region in the last nonlinear image.

Fig. 12

Time-lapse side view TPAF images of a bubble in a subepithelial void that was created with 750 nJ pulse energy in inferior vocal fold. Multiple arrows on the left side show the direction of the laser beam. The arrow on top shows the axial location of the targeted ablation plane. Scale bar indicates 100 μm.

These bubbles could also be detected with the naked eye after the ablation and used as indicators of successful void creation and presumably for guiding the injection needle. We therefore studied the lifetime of these bubbles for a range of energies from which bubbles covering more than 80% of the targeted area can be created and how they vary between two representative samples. Specifically, we followed volumes of bubbles in two sets of samples for each tissue type, whose initial bubble areas were presented in Fig. 10.

Figure 13 illustrates the dynamics of bubble volumes with time. We measured the bubble volume by counting the number of dark pixels in three-dimensional nonlinear images. To correct for the artifact of tissue/gas index of refraction mismatch at the boundaries of the void, the axial dimension of the bubble was downsized by a factor of 1.2, assuming that the bubble had an index of refraction similar to that of air. We note that this correction simply reverses the original focal shift correction made for the air lens when focused within the tissue.

Fig. 13

Bubble lifetimes for different pulse energies for two different samples for each tissue. (a, b) Superior and inferior porcine vocal folds. (c, d) Control and scarred hamster cheek pouches. Solid lines represent sample 1, and dashed lines represent sample 2.

The bubble lifetimes were generally consistent within a sample, but varied between the two samples. For example, in the first porcine samples, the bubble lifetimes in both superior and inferior vocal folds were between 80 and 100 min. In the second sample, their lifetime dropped to 30 and 40 min, highlighting sample-to-sample variability. In a given sample, the bubble lifetime did not change significantly with incident pulse energy, even if their initial volume did.

In the hamster cheek pouches, the lifetime of bubbles in both control and scarred tissues were shorter than porcine samples, while the biggest drop was observed for the scarred tissue. Although the initial volumes of bubbles were, in general, larger in scarred cheek pouches than in the control, they lasted a shorter time on the average. As will be discussed in detail in Sec. 4, shorter lifetime of bubbles can likely be attributed to higher compression imposed by the dense and stiffened mechanical structure of scarred tissue.

Overall bubble lifetime did not change within a sample significantly with the incident pulse energy but showed sample-to-sample variability. While the bubbles had the shortest lifetime (10 to 20 min) for scarred tissue, they might still provide enough time for verifying the creation of voids and guiding the injection of biomaterials.