Correlation between forward and backward generated SHG images of biological tissue relating to collagen structure and biomechanics

Second Harmonic Generation (SHG) microscopy has emerged as a powerful label-free tool for studying collagen microstructure in biological tissues, with increasing interest in the backward-detected signal due to its potential for translation to in vivo imaging contexts. The origin of backward-SHG signal is not fully understood and, as such, interpretation of images remains challenging, as it represents complex optical interactions. This has led to misinterpretations of the collagen structure of biological tissues. Using a custom optical chamber allowing bidirectional SHG imaging of the cornea under controlled pressure, we assessed the spatial relationship between features in backward images and the underlying structural features of collagen in the paired forward images. Collagen waviness (crimp) was identified to be a key structural feature responsible for periodic bands of fluctuating signal intensity in the backward SHG images. Additionally, we demonstrate the potential of SHG imaging for investigating the biomechanical mechanisms relating to collagen and build-upon previous theoretical and experimental frameworks explaining forward and backward signal generation and current limitations in their interpretation.

Second Harmonic Generation (SHG) is a nonlinear optical imaging technique in which two photons at a fundamental frequency interact with a non-centrosymmetric medium and combine to generate a new photon at twice the frequency (Chen et al., 2012). SHG is a coherent, label-free process that is dependent upon the organisation and symmetry of the material’s structure. The second harmonic phenomenon was first introduced in 1961, when it was observed that a second harmonic signal, which is half the wavelength of the incident laser, can be generated as it passes through a crystal (Franken et al., 1961). Since, applications of SHG-imaging have spanned many research areas and industries including photonics, optical communications, plasmonics, and biology (Martorell et al., 1997; Chen et al., 2012; Wang et al., 2023).

SHG-imaging has been applied extensively in the field of structural biology specifically, as it was recognised that some structures in the body, such as collagen, microtubules, and myosin filaments, exhibit non-centrosymmetric properties similar to a crystal, and therefore demonstrate non-linear optical phenomena and can be imaged in detail via taking advantage of this generation of the second harmonic. As such, SHG microscopy has proved to be a useful tool to better understand the structural and diseased patterns of collagen-rich tissues such as the skin, cornea, and bone (James and Campagnola, 2021).

Most research to-date has used forward-SHG performed on ultra-thin sections. Due to the coherent nature of the forward signal, the images generated provide a clear and detailed representation of collagen in the sample under study; however, due to the requirement for the sensor to be behind the sample the application of forward-SHG is generally limited to ex-vivo experiments and thin sections. Conversely, backward-SHG holds potential for in vivo imaging, as it uses the portion of the signal scattered from the tissue. While several studies have collected the backward signal, interpretation of the generated images remains difficult and presents several challenges that must be addressed for it to be employed effectively (Han et al., 2005).

Here we examine the interpretation of forward and backwards SHG signal with relation to collagen features and orientation in transparent tissue, specifically in the cornea. The cornea was selected as the imaging target for several reasons. It is a transparent, collagenous tissue, within which the collagen has a specific depth-dependent highly ordered arrangement that has been characterised previously using alternative techniques including, histology, SEM (Scanning Electron Microscopy), TEM (Transmission Electron Microscopy) and X-Ray diffraction techniques (Meek and Fullwood, 2001). The transparent nature of the tissue enables full-depth imaging (both forward and backward) of intact tissue, ensuring the structure is closely representative of the in vivo state, while the complex collagen arrangement, and access to previously acquired structural information from alternate techniques facilitates convenient validation and a means to assess the origin of imaging artefacts.

In addition to the other techniques, SHG has been previously used to image sections of the cornea in an unstained manner and has provided valuable structural information (Han et al., 2005). These studies have helped to elucidate collagen branching in detail in the anterior cornea (Mercatelli et al., 2017), and the overall depth-dependent organisation of corneal collagen (Morishige et al., 2006). Backward-SHG, specifically, has been used for corneal imaging across several studies, but due to the complexities in interpreting the backward signal, structural features of the collagen were commonly misinterpreted (Yeh et al., 2002; Teng et al., 2006; Wu and Yeh, 2008; Bueno et al., 2011; McQuaid et al., 2014; Wang et al., 2017; 2018; Ávila et al., 2019; Anton et al., 2023).

Establishing a correlation between backward and forward SHG images in the cornea has been challenging (Han et al., 2005; Morishige et al., 2009), despite its importance for potential clinical translation. Several studies have attempted to address this, but with limitations. Strong correlations have been reported between forward and backward signals but only in the superficial layers of the anterior stroma (Lo et al., 2008; Mercatelli et al., 2017), with a lack of correlation observed in deeper layers, without identifying the determinants responsible for this (Lo et al., 2008). A key limitation of these previous studies was that they were performed on thin tissue sections and limited to static imaging conditions. In the present work, we address this limitation using SHG-imaging on intact tissue subjected to different pressures. In-doing this, we highlight an important structural feature that influences the directionality and correlation of the two SHG signals throughout the depth.

Considering the complexity of interpreting the backward signal, we provide a detailed discussion of the factors that influence backward SHG signal generation across different tissues, outlining the current theoretical frameworks proposed to explain forward and backward signals and their limitations.

Ten fresh porcine eyes were obtained from a local abattoir, from pigs approximately 6 months of age. All eyes included in the study had visibly transparent corneas before and after imaging. Corneas were imaged (<72 h postmortem) in an intact and non-processed state to prevent any potential structural modifications induced by sectioning or fixation agents.

For SHG-imaging, a Zeiss LSM980 on an upright Examiner. Z1 microscope was used. Laser excitation was provided by a Spectra-Physics Insight DS + laser tuned to 820 nm. The laser was linearly polarised in the x-axis of the image plane. Two water-immersion objectives were used: ×10 NA0.5 and 20x NA1.0 UV-VIS-IR. The light path used a dichroic BS-MP 760++, a SP485 filter for reflected light, and a SP485/IR + filter and a high numerical aperture condenser (WD = 8.6 mm, NA = 0.8) for the transmitted light. Each image was scanned at 1024 × 1024 pixels resolution. A schematic of the setup is depicted in Figure 1.

Initial imaging was performed with the cornea remaining within the eye globe, maintained at normal physiological intraocular pressure (IOP), using reflection-mode SHG at two magnifications: 10× (NA 0.5) and 20× (NA 1.0). Imaging was performed on the central cornea (606 × 606 µm) for all samples.

The cornea was subsequently dissected from the globe, maintaining a 3-mm rim of sclera, it was then re-imaged to acquire simultaneous forward- and backward-detected SHG signals for direct comparison. Following this, a custom-designed chamber was produced, shown in Figure 1, to enable forward and backward SHG-imaging of the cornea at different pressures. The base of the chamber was manufactured from an optically suitable material (Falcon™ Tissue Culture Dish, Corning Inc.) for forward detection of the SHG signal using the Zeiss LSM980. The chamber was also designed to allow the top to be filled with PBS solution as the immersion medium for the objectives, this also ensured the cornea remained hydrated throughout imaging. The cornea was mounted onto the chamber which clamped around the scleral rim. The chamber incorporated an inlet and an outlet with a syringe connected to one port to fill the chamber behind the cornea with Phosphate buffered saline solution (PBS, Sigma-Aldrich, UK, ρ = 0.995 g/mL at 25 °C) so the tissue could be pressurised while also adjusting the pressure by injecting or withdrawing fluid, while a manometer (DPGW-05, Dwyer Instruments, United States) on the opposing port provides real-time pressure monitoring (Figure 1). This allowed us to investigate the effects of pressure variation on the structure and features of the collagen bundles and the subsequent effect on the images generated from the backward and forward signals. No significant change in corneal thickness was observed over the imaging time; hence it was assumed that hydration did not vary significantly across the experiment.

To the best of our knowledge, this is the first experimental set-up described that enables forward imaging of the cornea in its unaltered, pressurised state providing layer-by-layer detail and an ability to capture structural changes under different loads.

Figure 2a shows a backward-detected image of the posterior corneal stroma, with a structure similar to previously published images (Yeh et al., 2002; Teng et al., 2006; Wu and Yeh, 2008; Bueno et al., 2011; McQuaid et al., 2014; Wang et al., 2017; 2018; Ávila et al., 2019; Anton et al., 2023). The image shows highly organised, vertically aligned bands of high-intensity signal, with periodic spacing. These high intensity bands were previously interpreted as ‘collagen bundles’ (Bueno et al., 2011; Anton et al., 2023), with algorithms used to compute their predominant orientation in some publications (Ávila et al., 2019; Bueno et al., 2022). Similarly, the areas of low signal were believed to be regions absent of collagen, corresponding to areas of ‘void’ or ‘interlamellar spacing’ (Yeh et al., 2002; Wu and Yeh, 2008). However, higher magnification of a region from Figure 2a, resolves speckled, elongated patterns, oriented perpendicularly - rather than aligned - to the structures previously identified as ‘collagen bundles’ (Figure 2b). It is also noticeable that the speckled features oriented parallel to the horizontally polarised laser excitation (white arrow in Figure 2b), displayed higher signal than those perpendicular to the polarisation direction (yellow asterisks in Figure 2b).

To further investigate these speckled features and characterise the regions of low or absent signal, the cornea was dissected from the globe to enable simultaneous acquisition of backward and forward images. Figures 3a,b,d,e show backward and forward images of the posterior corneal stroma at 457 μm and 524 µm depth from the anterior surface, at ×10 and ×20 magnifications, respectively. Depth was quantified based on the first detectable SHG signal which was designated as the reference surface, corresponding to the surface of Bowman’s layer. Because the cornea was removed from the globe and was no longer in its native curved shaped, there is a relative disorganisation of the patterns in the backward images compared to Figure 2a. However, the aligned, high signal, periodic structures are still evident and mirror closely the previously reported backward-SHG images of the cornea. In the corresponding forward images, the collagen bundles are clearly visible, running in an organised, undisrupted, and wavy pattern (Figures 3b,e). This waviness is commonly referred to as ‘crimp’ in previous structural studies of collagen (Franchi et al., 2007; Jan et al., 2018) and it is postulated to be an important biomechanical mechanism in the cornea and in other collagenous tissues in the body (Franchi et al., 2007).

Two regions where the aligned structures are clearly visible and periodic (red and yellow boxes from Figures 3a,d), were overlayed with the corresponding regions from the forward images (red and yellow boxes from Figures 3b,e). The merged images reveal that the aligned bands in the backward image correspond to regions of the wavy bundles observed in the forward images (Figures 3c,f). This indicates that the structures previously identified as ‘collagen bundles in numerous studies’- and used to compute predominant orientations in an image plane - are, in fact, oriented perpendicularly to the actual lamellar direction and originate from signal contributions from different regions of the wavy lamellae. Similarly, the dark regions do not correspond to void, but to an absent or weak backward signal due to the specific collagen position along the wavy lamellae. While a precise correlation between the two signals is difficult to establish across the entire image plane for reasons detailed in the discussion section, it is clear from Figures 3a–f that the forward-detected bundles consistently run perpendicular to the backward-detected high-signal bands throughout the entire imaging plane.

Due to the relative absence of collagen waviness - the anterior cornea - which has been shown previously to be made up of interwoven, mesh-like collagen bundles (Meek and Knupp, 2015) - shows a consistently strong correlation between the forward and backward signal (Figure 4).

Based on these findings, it is evident that the wavy features of collagen bundles in the mid and posterior stroma, referred to as crimp, give rise to the periodic features of high and low signal in the backward SHG-images. To further demonstrate this, additional imaging of the cornea was conducted after placing the cornea in the chamber described in the methods section and shown schematically in Figure 1. In this instance, images of the cornea were acquired throughout the depth firstly at low (∼5 mmHg) and then higher (∼60 mmHg) pressure. Supraphysiological pressures at both ends of the spectrum were deliberately used for the purpose of analysing the origin of features in the forward images and their relationship to the backward images. Despite being outside the normal physiological range, these pressures remain relevant, as they can occur in certain disease states such as glaucoma and hypotony (He et al., 2006; Costa et al., 2007).

It has been theorised, but not yet proven in the literature, that the waviness (crimp) of the collagen is a structural mechanism governing the corneal response to increasing intraocular pressure (Jan et al., 2018; Wilson et al., 2020). Hence, it is expected that at increasing pressures the collagen bundles would straighten which will be evident on the forward images, while the periodic features in the backward images will disappear at higher pressures, as the collagen fibres become straighter.

Figures 5a,b shows the backward and forward SHG images of the cornea (560 µm depth) mounted in the transparent chamber at a low pressure (∼5 mmHg). The collagen bundles appear wavy and crimped, with periodic features clearly evident in the backward images. The overlayed image of the two channels show that the backward signal matches the forward signal in some regions, periodically (Figure 5c). The intra-chamber pressure was then raised to supra-physiological levels (∼60 mmHg) to observe how both signals would be affected. Figures 5d–f shows the corresponding images after increasing the pressure. The forward image (Figure 5e) reveals straightened bundles, while the backward image (Figure 5d) shows a more uniform forward-to-backward signal and loss of the periodic bands previously observed. This indicates that collagen waviness is the main factor driving the difference between backward and forward SHG signals and the periodic dark bands in the backward image. When pressure increased and the bundles uncrimped, the two signal contributions became consistent across the entire image plane (Figure 5F).

In some imaging planes, there were some areas, where the collagen did not uncrimp completely, and in these instances periodic patterns are evident in the backward images in specific regions directly corresponding to the crimped areas in the forward images (Figure 6).

A complete layer-by-layer recording of the cornea across a thickness of 985 μm, starting from the posterior surface and progressing towards the surface, before (Supplementary Video S1) and after (Supplementary Video S2) increasing the pressure is supplied in the Supplementary Appendix. In these recordings, the collagen becomes noticeably uncrimped after the increase in pressure and the differences in the backward signal are clearly visible between the crimped and uncrimped states. It should be noted that images of the anterior cornea showed no noticeable change between the pressurised and unpressurised states, as shown in the Supplementary Material.

In this paper, we present a novel method to image intact corneas ex vivo, through the entire tissue depth, under different pressures using SHG-imaging, while detecting both the forward and backwards signals simultaneously. In doing so, we address highlight an important feature that appears to govern differences in forward and backward SHG signal generation recurring misinterpretation of the backward images of biological tissue generated from SHG microscopy and relate it to a specific structural characteristic–collagen waviness - that appears to govern differences in forward and backward SHG signal generation. We also highlight the importance of understanding this mechanism, as it is a key contributor to corneal biomechanics, and one which is expected to extend to other biological tissues.

Some previous SHG studies, where only the backward signal was obtained, led to several incorrect conclusions about the structure of the cornea (Yeh et al., 2002; Teng et al., 2006; Wu and Yeh, 2008; Bueno et al., 2011; McQuaid et al., 2014; Wang et al., 2017; 2018; Ávila et al., 2019; Anton et al., 2023). Features seen in backward images were sometimes mistakenly identified as collagen fibres. Their apparent ‘diameter’ was measured across different depths to infer fibre diameter variation, and although it was correctly noted that these ‘fibres’ appeared larger than reported in other collagen studies (Bueno et al., 2011), this was because the measurements reflected an optical effect of SHG arising from collagen waviness rather than the actual fibre dimensions.

Importantly, a key biomechanical mechanism governing the corneal response to increased pressure was overlooked: collagen crimp. A previous study using backward SHG-imaging to evaluate corneal response to IOP changes (Wu and Yeh, 2008), observed a narrowing of the dark ‘low signal’ bands - which was attributed to changes to ‘interlamellar spacing’. Our study clarifies that changes to these periodic intensity levels observable in the backward images, occur due to straightening of the crimped lamellae under increased pressures. The decrease in the size of low signal regions is due to extension of the fibres, leading to a reduction in the angle between the imaging plane and the out-of-plane orientation of the fibre, leading to an eventual loss of these periodic patterns in the backward direction as the fibres become fully extended (Figure 5).

Though a correlation is clear between forward and backward images accounting for specific collagen features, a direct and consistent correlation between the forward and backward images across the entire image plane remains hard to establish due to the complexity of signal generation, which is influenced by several factors and remains a highly debated topic across SHG research (Chen et al., 2012). The complexity arises from the nonlinear behaviour of SHG in non-centrosymmetric structures, where signal strength and directionality depend on multiple inter-related factors, including the mixed origin of the backward SHG images from both the backscattered forward SHG signal and the directly generated backward SHG signal (Shen, 2013), fibril length and diameter (Williams et al., 2005), axial arrangement (LaComb et al., 2008), polarisation state of the excitation light (Mazumder et al., 2017), numerical aperture of the system (del Barco and Bueno, 2012), and refractive index mismatches that influence scattering and phase coherence (Mazumder et al., 2017), the complexity of which is even more exacerbated in thick tissues, such as the cornea.

Backward and forward SHG signals have been analysed across different tissues, including skin, tendon and corneas (Sivaguru et al., 2010; Pendleton et al., 2020). Theoretical frameworks have been proposed to explain the directionality of SHG and therefore lay the groundwork for interpretating both signals in normal and disease states (Williams et al., 2005; LaComb et al., 2008). However, no single unified theory has yet been established that can fully account for differences in SHG signal generation directionality across different tissue types.

Theoretical frameworks based on fibril size suggest that fibrils with diameters on the order of the SHG wavelength generate predominantly forward-directed signals, whereas smaller fibrils (less than λ/10) produce nearly equal forward and backward signal (Williams et al., 2005). However, this theory cannot account for the difficulty in correlating the backward and forward signals in the cornea across different layers, which is known to have a consistent diameter of collagen fibrils across the depth, on the order of 31 nm in pigs (Hayes et al., 2017); nor the intensity differences observed with single lamella, since fibril size would not be expected to vary periodically within the same lamellar bundle.

A more comprehensive theory has been developed in which the fibril is no longer treated as an isolated SHG generator. Instead, the model considers the collective packing of fibrils within a domain, which influences the phase-matching conditions and, consequently, the continuity of the coherent second harmonic generation (LaComb et al., 2008). This theoretical framework is based on a model of phase matching and relaxed phase matching conditions in collagenous tissues with respect to SHG-directionality, which improved upon the models based on fibril size. This has been used to differentiate between normal and diseased states, since diseased tissue is known exhibit disrupted collagen organisation. A detailed explanation of the model can be found elsewhere (LaComb et al., 2008). Briefly, in biological conditions, a perfect phase matching where the wavevector mismatch Δ k between the fundamental and second harmonic fields fulfils the following criteria: Δ k = k 2 w − 2 k w = 0. is difficult to achieve. Therefore, a quasi-phase matching is usually achieved where Δ k is small but not zero.

The generation of the backward signal is possible only when the axial momentum K is sufficiently high to induce a change in the direction of the photons (meaning K b > K f). Consequently, the phase mismatch in the backward and forward directions conforms to Δ k b > Δ k f and therefore the coherence length in the backward direction is smaller than in the forward direction L c b < L c f (Equations a, b). This sets the minimum phase mismatch at Δ k f, governed by dispersion Δ n (Δ n = n 2 ω − n ω) (Equation c) and the maximum coherence length at L c f.

For a domain size L, defined by a stack of fibrils packed up homogenously in the axial direction in which SHG signal accumulates (Figure 6a), the SHG conversion efficiency of the fundamental beam in each direction can be determined from Equation d:

Assigning backward mismatches Δ k b to integer multiples of Δ k f, i.e., m Δ k f, and normalising the intensity I, and the domain size L with respect to L c f, we can calculate the percentage of the forward signal for different m. This gives that the forward signal approaches 100% as domain size approaches the coherence length (L c f), while the backward signal is generated mostly when the domain size L is smaller than the coherence length in the forward direction L c f. This suggests that the directionality of SHG is directly influenced by the domain size relative to the coherence length, therefore a change in the domain size, will eventually affect the ratio of the forward and backward. It should be noted that this model takes into consideration the bulk properties of the tissue, including the tissue scattering properties and a calibration of the forward and backward signal, therefore extracting the local SHG ratio of forward to backward. However, in our case the recorded forward and backward signals are not calibrated, and scattering effects cannot be fully accounted for. What we measure experimentally are the ‘collected’ forward and backward signals, which differ from the intrinsic SHG forward and backward emissions. However, the cornea is a weakly scattering tissue (Sardar et al., 2009),which underlies its excellent optical performance, and therefore the scattering effects are minimal. This has been further demonstrated by flipping the cornea upside down, for which the same periodic patterns seen in the previous backward images can be observed (Supplementary Figure S3).

This quasi-phase matching model can explain the strong backward signal observed in the anterior cornea, where collagen fibres are interwoven and anchored to Bowman’s layer. Their disordered arrangement might introduce phase mismatch, which accounts for the increased backward signal in this region. Moving deeper into the corneal stroma, the periodic pattern observed in the backward signal can be analysed using this model, by extending the current proposed framework to introduce an additional parameter - tilt, defined as the out-of-plane angle between the collagen bundles and the axial propagation direction. Tilt is inherent to the crimp pattern, as the sides of the crimped collagen waves are inclined relative to the imaging plane (Figure 7b). This inclination induces a change in the ‘effective’ domain size, affecting phase-matching conditions and thus the forward-to-backward signal distribution. This inclination also alters the orientation of the individual collagen molecules with respect to the imaging axis (Figure 7b), potentially influencing the directionality of the generated SHG signal.

This model can explain the appearance of the backward-SHG images in Figure 3. The backward signal alternates periodically between bands of high signal and low signal, and these periodic patterns match with sides of the crimped patterns from the corresponding forward images. This can be due to the specific angle of each of the sides of the crimp, which creates a periodic alternation between different Forward/Backward ratios. In the pressurised cornea, extension of the collagen bundles (reducing crimp) leads to close correlation between the forward and backward signals (Figure 5), indicating that out-of-plane waviness is a key parameter influencing SHG signal directionality, affecting the forward and backward components differently.

Polarisation is another factor that must be considered when evaluating the intensity of SHG-signal. The polarisation dependence of the SHG signal for in-plane fibres has been demonstrated both theoretically and experimentally (Mazumder et al., 2017), consistent with our observations (Figure 2b), where fibres aligned parallel to the incident light polarisation exhibit higher intensity than those oriented perpendicularly. Circular polarisation excitations have been shown to allow SHG signal detection of all in-plane orientations (Alizadeh et al., 2019). However, the polarisation dependence of out-of-plane collagen has not yet been thoroughly investigated. A recent study showed that polarisation-in, polarisation-out SHG can be used to determine the polar angle of collagen fibres, but only the forward signal was detected in this study, leaving the effect of out-of-plane orientation on forward and backward SHG signals undetermined (Harvey et al., 2023).

While valuable, previous studies comparing forward and backward SHG signals have generally focused on static samples and primarily on SHG generation at the fibrillar level, analysing how individual fibril properties affect directionality (Han et al., 2005).

Broader-scale structural changes, such as those occurring at the bundle level, including out-of-plane uncrimping give an additional insight to the previous research work, demonstrating that when the same tissue is imaged at the same depth under identical conditions, a change in pressure - and thus in bundle organisation - significantly alters the generated SHG signals contributions. Most importantly, the intrinsic structure of how the collagen is organised in a specific area remained unchanged; what varied was how this structure responded to the increase in IOP, influencing SHG generation in both directions. In practice, these structural changes can occur within very short timescales - during each heartbeat, when the cornea experiences pulsatile IOP variations. This dynamic behaviour must be considered when analysing backward signals, particularly if an in vivo diagnostic device or AI-based model is to be developed to distinguish between normal and diseased states (Anton et al., 2023).

It is worth noting however, that a key limitation of this work is that it does not include a quantitative analysis of the correlation between the two signals. Due to the current lack of an established understanding of all the mechanisms contributing to SHG signal generation and directionality in biological tissues, many confounding variables - both at the fibrillar and bundle level-are expected to affect both signals. Further to this, imaging thick, non-sectioned tissues subjected to pressure changes adds another layer of complexity, which prevents meaningful quantitative analysis at this time, but is a future aim of this continued work.

Due to its suitability for collagen imaging SHG-microscopy holds vast potential for understanding the structure and biomechanics of biological tissue. It has many advantages over alternate microscopy techniques, as it requires no staining or fixation; provides depth-resolved high-resolution images; and the wavelengths employed, and powers required to generate images, make it suitable for potential translation to an in vivo imaging setting. However, to achieve the latter requires the backward signal to be interpretable. Here, this work introduced a structural feature that governs the correlation between the images generated from forward and backward SHG-images of the cornea. Although further work remains in understanding this correlation in full, this work represents an important step forward with regards to the interpretation of backward SHG and the association between specific structural features and their representation in the image.

Further to this, SHG-imaging was highlighted as a promising technique that can be employed to understand biomechanical mechanisms and their association with structure and structural changes, overcoming the limitations of the static imaging techniques. Although many previous studies, using a range of techniques including SEM, TEM and Wide-angle X-ray scattering techniques, have provided us with a wealth of information on corneal structure (Meek and Fullwood, 2001), due to their limitations, evidence of how these structural mechanisms relate to biomechanics remains challenging.

With respect to the cornea, and in the context of in vivo translation and clinical impact - SHG imaging, and understanding the origin and interpretation of the backward signal holds vast potential. Diseases such as keratoconus and ectasia are associated with structural and biomechanical changes but are currently not fully understood and diagnosed at a late stage due to the lack of diagnostic tools capable of clinical evaluation of structure and biomechanics. In vivo SHG imaging of the cornea has already been demonstrated to be achievable (Ávila et al., 2019), and with continued advancements, such as those presented in this study, widespread clinical translation may be possible. In the shorter term, SHG can be employed exvivo to understand the structural and biomechanical changes that occur during common ocular surgeries, including refractive procedures, to optimise their application and safety profile.