Study on the tissue clearing process using different agents by Mueller matrix microscope

: In this paper, we monitor the in vitro tissue clearing process of mouse dorsal skin immersed into two types of agents using Mueller matrix microscope. By Mueller matrix polar decomposition, we can see that the major difference between polarization changes due to two kinds of agents is the opposite trend of phase retardance with clearing. For the insight of the connection between different agents with the microstructural and optical changes of cleared tissues, we establish various models to mimic the dynamic process of microphysical features of tissues with clearing time. The mechanisms considered include refractive index matching, collagen shrinkage, more orderly fibers and birefringence variation. We compare the experimental results with simulations based on a single mechanism model and a combined model, respectively, which confirms that an individual possible mechanism cannot explain the polarization phenomena due to clearing. Also by simulations of various clearing models involving two possible mechanisms, we can speculate that formamide and saturated sucrose as agents have respective impacts on tissue features and then cause different polarization changes with clearing. Specifically, collagen shrinkage plus birefringence reduction can better explain the tissue cleared by formamide, and refractive index match plus increased birefringence model is likely to be a proper description of tissue cleared by sucrose. Both simulations and experiments also validate the potential of Mueller matrix microscope as a good tool to understand the interaction between clearing agents and tissues.

(LSCI) [12], 3D-confocal microscopy [13], polarized microscopy [14], and multiphoton imaging [15] can be combined with TOC. These advanced optical methods promote studies on the molecular and microstructure and function of tissues in vivo or in vitro also can be used to evaluate the efficacy of TOC.
There are several possible mechanisms about the explanations of OC process [21]. The first and the most common explanation is refractive index matching. The second common explanation is collagen shrinkage. There are some corresponding experiments showing that sample's thickness changes significantly with clearing [27]. The third common explanation is more orderly collagen, which implies that scatterers' near-order spatial correlation is enhanced, and thus the lateral scattering is minimized or even eliminated [28].
Polarization imaging techniques are sensitive to microstructural changes in tissues, and can therefore be regarded as potential and label-free tools for physiological process monitoring and pathological diagnosis. Recently, polarization techniques are attracting more and more attention in biomedicine [29][30][31][32][33]. As a comprehensive description of polarization characteristics of scattering samples, Mueller matrix polarimetry has demonstrated promising potential in abnormal tissues detection for both backward scattering imaging of bulk tissue samples and transmission imaging of thin tissue slices [34][35][36]. In our previous research, we studied the influence of TOC on tissue polarization imaging [37,38], and give some preliminary explanations using our Monte Carlo simulations [39] combined with our anisotropic tissue model [40]. Also by Mueller matrix polar decomposition method, we show a semi-quantitative description on the polarization optical change due to tissue clearing.
In this paper, we focus on the tissue clearing process using two types of OCAS and try to explain the difference of dynamic microstructural change of cleared tissue and understand how these two OCAS make tissues clear respectively. Collagen fibers are typical anisotropic tissues and can generate obvious polarization optical phenomenon, and there are studies showing that hydroxyl molecules in OCAs will interact with collagen and affect the clearing results [10]. So we use skin rich in collagen fibers as the research object, and observe the polarization features by Mueller matrix imaging and MMPD method combined with Monte Carlo simulation of various tissue clearing mechanisms.

Sample and OCAs
Our experimental tissue samples are taken from seven-week-old nude mice from Guangdong Medical Lab Animal Center. They were fed under specific pathogen-free conditions. After mice being sacrificed, a 3x3 cm skin with a thickness of 1mm was cut from mice's back. Then, we divided the skin into 9 small pieces on average, with each piece 1x1 cm. Subsequently, these pieces of skin are immersed in two different types of OCAs and measured by polarized forward microscope to obtain their Mueller matrix images.
Polarization status is an effective way to study the microscopic changes in tissues and is especially sensitive to those optical anisotropic features, such as birefringence or fibrous microstructures. Collagen fibers are the main components of dermis, and show an apparent anisotropic scattering capability in our previous research work [38,41]. There are studies that the hydroxyl molecule in OCAs will interact with collagen and cause various possible microcosmic changes of fiber content and arrangement. We select formamide and saturated sucrose as our agents, because the molecular structure of the latter does contain hydroxyl groups and th becoming tran without hydro two agent: fo refractive inde

Experime
The  [31], as e incident 1 with a luminates ed by the nally, the pixels). e-cylinder gnize and possible ng effects ed on our polarization scattering calculation program and SCBM tissue model and compare them with experimental results in the following section.
In sphere-cylinder birefringence model (SCBM), as shown in Fig. 3, sphere scatterers represent cells and other isotropic microstructures, such as nuclei and other organelles. Cylindrical scatterers represent fibrous microstructures like collagen fibers in skin tissue. Birefringence is introduced in our tissue model considering the optical anisotropy at molecular level. By combination of Monte Carlo simulation and SCBM, we can mimic various tissue types and simulate the transmission and scattering of polarized light in biological tissues.
In our simulation program, variable parameters for scatterers include scattering coefficient, diameter of the spheres and cylinders, the mean value and standard deviation of the orientation distribution function for the cylinders. For the ambient medium, variable parameters include the refractive index, the absorption coefficient, the optical activity coefficient, and the value and orientation of birefringence. The initial state of skin tissue before clearing can be set as follows [42,43]: the tissue thickness is 1mm; the diameter of spheres and cylinders are 0.2μm and 1.5μm, respectively; the scattering coefficient of spheres and cylinders are 20 cm −1 and 180cm −1 ; the refractive indices of the interstitial medium and the scatterers (including spheres and cylinders) are 1.35 and 1.43, respectively; the cylinders are along the x-axis direction with a FWHM of 18 degree in the orientation distribution considering collagen fibers packed in bundles and arranged in a lamellae structure; the birefringence value is 3e-5 and its optical axis is along the x-axis; the wavelength of the incident light is 633nm and the simulated photon number is 10e7.
To mimic tissue optical clearing process by different agents, we consider various dynamic models corresponding to several possible clearing mechanisms including refractive index match, tissue shrinkage by dehydration, fluctuation of birefringence effect in intercellular substance and ordering of fiber arrangement. In the following studies, we will start with the simulations based on one single mechanism and compare them respectively with experiments, and then we will observe whether the parallel simulations based on two major mechanisms can explain the experimental phenomena better.

Experimental results of tissue clearing with different agents
Simple immersion is a very convenient and useful way to make tissue transparent. In Fig. 4 the skin samples are put on a 1951 United States Air Force (USAF) resolution test target before and after treatment with two kinds of agents: formamide and saturated sucrose. Whitelight images make clear that these two OCAs we choose have a rapid and apparent effect to make tissue tr cover the pat treatments, m the pattern of Fig Fig. 8(a) s dels, where m tively. By con obvious fluctu nt, Fig. 8 According to the above simulations, we can see that a combined model involving multiple mechanisms can better explain the trend of phase retardance with clearing than any single mechanism model. The differences between simulation results of various combined models are the change of depolarization parameter. Then in next simulations we focus on three diagonal elements of Mueller matrix: m22, m33, m44, which are closely related with the depolarization phenomena of measured tissues. Figure 10 shows the simulated diagonal Mueller elements based on different clearing models. By comparing experimental data with simulation results, we can deduce the similar possible clearing models. Specifically, the experimental depolarization parameter,Δ, cannot be close to 1 after formamide clearing (shown in Fig. 10(a)), which supports the model involving collagen shrinkage plus birefringence reduction again. From Fig. 10(b), only the combined model involving refractive index matching and increased birefringence can better mimic the increase of Δ using saturated sucrose as agent than the other two models. Figures 7, 8, and 9 demonstrate rather regular oscillations of depolarization Δ and retardance δ parameters of the skin during optical clearing at sucrose application. These oscillations for studied nude mouse skin occurred with time-period of approx. 2 min, which is well fit to temporal characteristics of oscillations of optical properties found as 2.5 and 3.5 min for hamster and rat skin at clearing by anhydrous glycerol, respectively [23]. Such quasiperiodic oscillations were described for the first time in Ref [2]. for collagenous tissues like human sclera and then experimentally proved for coherent and polarization (linear parallel-and crossed-polarization) properties of human sclera for which the period of 1.5-2 min was found at application of x-ray contrast trazograph solution [44]. In this paper, it was also hypothesized that the oscillations are the result of temporal-spatially irregular optical clearing agent diffusion driven by a local multi-step dehydration of collagen and dilution of the interstitial fluid. There is a high degree of evidence that a similar mechanism may underlie the temporal oscillatory behavior of the depolarization and retardance parameters.

Conclusion
This paper focus on the influence of different clearing agents on polarization features of tissues. By Mueller matrix microscope, we select formamide and saturated sucrose as agents and observe the Mueller matrix images and MMPD change of mouse skin with the clearing time. The differences in experimental phenomena may originate from their respective clearing mechanisms of two kinds of agents. To find out the key factors of making the measured tissue transparent, we consider several possible clearing mechanisms and then establish corresponding single factor models and combined models. We can mimic the dynamic tissue clearing process by simulating the gradual changes of microphysical attributes using our Monte Carlo simulations, and then compare our experimental results with simulations results based on different clearing models.
Firstly, the simulations involving only one mechanism cannot approximate the experimental MMPD, which confirms that tissue clearing is very likely due to multiple mechanisms working together. Further investigation and comparison indicate that different agents have respective influence on the cleared tissue, such as shrinkage due to dehydration, changes in Fiber Orientation and birefringence variation in intercellular substance. Based on the dynamic variations of diagonal Mueller elements, depolarization and phase retardance parameters, this paper try to explain the clearing process by formamide and sucrose. Specifically, tissue clearing by formamide is mainly due to collagen shrinkage plus decreased birefringence, and clearing by saturated sucrose can be mainly due to refractive index matching plus increased birefringence. The investigation including experiments and simulations provides a way to understand the clearing process according to the connection between polarization changes and microphysical features of tissues. In addition, this paper also verify that Mueller matrix imaging is potentially a powerful method applied in tissue clearing.