2.1.

Pig Bladder Analysis—Inverse Adding–Doubling Method

Bladders from locally slaughtered male pigs ($n=3$) were provided refrigerated and vacuum packed by an established supplier (WETLAB) and immediately refrigerated. Superior and lateral bladder surfaces (apical surface) were separated from the anteroinferior surface (basal surface) of the bladder using medical grade scalpel. Subsequently, 2 cm² sections (of 6- to 10-mm thickness) from each half were excised for analysis. Excised tissue was introduced into a specialized tissue clamp and fixed inside the spectrometer beam path for analysis. A Lambda 1050 spectrophotometer with a specialized integrating sphere module was used to record the absorption, transmission, and reflectance (diffuse) measurements. Samples were analyzed in the spectrophotometer for all three parameters in 1.5-nm increments across a wavelength range of 350 to 1800 nm. All spectrophotometric procedures were performed at room temperatures (22°C). Spectrophotometry data were visualized graphically using Origin Pro 8 software (Fig. 1). Laboratory facilities housing spectrometric devices used in this procedure were kept at steady room temperatures. To prevent drying and deterioration of tissue, samples were transported and maintained in small, covered petri dishes on a bed of dry ice at all times. These covers were kept on to reduce air exposure. Samples kept in the open for long periods of time between experiments were also sealed in their petri dishes with parafilm to further reduce exposure.

Fig. 1

Total transmittance (Tt) and diffuse reflectance (Rd) spectra obtained from 2-mm-thick pig urinary bladder.

To process the experimental results and determine the optical parameters, a combination of the inverse adding–doubling (IAD) method with a corrective Monte-Carlo calculation was used.^37,38 These methods are widely employed in tissue optics for processing data obtained by integrating sphere enabled spectrophotometry.

The IAD method allows for the determination of tissue absorption and scattering coefficients using the values of the diffuse reflectance and total transmission coefficients. The anisotropy factor ($g$) was fixed during the calculations. For the purpose of this paper, $g$ was taken as 0.9 due to this value being most typical for the majority of biological tissues in the visible and near-infrared spectral wavelengths.³⁹

The solutions to the series of Eqs. (1) and (2) were used as the initial values of ${\mu }_{\mathrm{a}}$ and ${\mu }_{\mathrm{s}}^{\prime }$

The Monte-Carlo method was used to correct for the light loss at the sample edges, which is not collected by the integrating sphere. The calculated lost light for a given set of optical properties was taken into account within the subsequent iteration of the calculation using the adding–doubling method. This process was repeated as long as the optical properties were not stable, i.e., until the changes of calculated scattering and absorption values were less than $0.01{\mathrm{mm}}^{-1}$ between iterations.

2.2.

Construction of Three-Dimensional Model

TracePro (Lambda Software) was used for the construction of the model. The creation of two layers (mucosa and muscle) and simulation of NADH and FAD was carried out following an identical procedure to that described in Rafailov et al.³⁶

2.3.

Simulation of Fluorescence

TracePro software was further used to simulate fluorescence from the above-constructed model. It employs fluorescent properties along with the material properties of the modeled tissue as well as the laws described previously in Rafailov et al.³⁶ The relative absorption $ab(\lambda )$ and relative excitation $ex(\lambda )$ are two major parameters that can be set and normalized to the molar extinction coefficient $K_{\text{peak}}$ and the relative emission $em(\lambda )$. Input of the molar concentration $C_{\text{molar}}$ is used to set the fluorescent material concentration. The fluorophore absorption coefficient in media is determined by

With the path length before absorption

Ascribing quantum efficiency to the system determines the number of photons involved in the simulation process.

2.4.

Fluorescence Measurement of Urinary Bladder Tissue

Urinary bladder tissue was obtained as previously described and was split between dome and trigone sections. The bisected bladder was “virtually” divided into 16 sections. Virtual sections of the inner mucosal layer of the bladder were sequentially subjected to optical analysis by the fluorescence spectroscope of an “LAKK-M” multifunctional laser-based noninvasive diagnostic device. The fiber of the device was run along the length of its stand and the probe was placed into direct contact with the bladder tissue. No extra force was applied to the probe. Tissue was scanned sequentially using four different excitation sources (UV 365 nm, blue 450 nm, green 532 nm, and red 635 nm). Five seconds of contact was allocated per wavelength measurement. Resultant tissue fluorescence was recorded across a wavelength range of 300 to 800 nm using an inbuilt spectrometer and processed using custom-made software (LDF 3v3.1.1.403, SPE “LAZMA”). Data were exported and visualized graphically using Origin Pro 8 software. The entire procedure was carried out at room temperature, in a windowless dark room to limit noise.

3.1.

Pig Bladder Analysis—Inverse Adding–Doubling Method

For the construction of the 3-D model, the required transport coefficients of tissue were calculated from the experimentally obtained transmission and diffuse reflectance values. Thus, using IAD method-based modeling, absorption and scattering coefficient spectra were reconstructed [Figs. 2(a) and 2(b), respectively].

Fig. 2

(a) The absorption coefficient of pig urinary bladder and (b) the reduced scattering coefficient of pig urinary bladder. Experimental data from Staveren⁴⁰ are portrayed on the spectrum.

Experimental data presented in Ref. 40 are provided for comparison in Fig. 2. The result of calculated tissue penetration depth is displayed on Fig. 3. The evaluation of this depth in biological tissue was achieved using

Fig. 3

The light penetration depth in urinary bladder tissue.

3.2.

Three-Dimensional Model

Due to the complex structural nature of organic tissue (in our case the urinary bladder), for the purposes of the theoretical modeling described in this paper, we constructed a simplified dual-layer model (mucosa and muscle).

Figure 4 displays the side and 3-D views of the constructed model simulating UV (a) and blue (b) light irradiation within the tissue. As it is evident from Fig. 3, the diagnostic depth into the tissue is around 1 to 1.2 mm.

Fig. 4

(a) Side view and full 3-D view of model depicting simulated UV light within tissue. (b) Side view and full 3-D view of model depicting simulated blue light within tissue. For both sets, the source, detector, and the tissue layers are labeled. The lines indicate the passage of individual photons through the tissue.

3.3.

Measured and Modeled Fluorescence

The results of experimental urinary bladder fluorescence at 365 and 450 nm excitation were compared to a simulated curve produced using the 3-D model (displayed in Fig. 5). This simulation incorporates NADH + collagen or FAD + collagen presence in the tissue.

Fig. 5

Comparison of (1) experimental and (2) model-based spectra, calculated using the Monte-Carlo method. Model simulates presence of (a) NADH + collagen and (b) FAD + collagen.

4.1.

Pig Bladder Analysis—Inverse Adding–Doubling Method

This study aimed to create a computer simulated optical 3-D bladder model for observing the propagation of light through bladder tissue. Looking at Fig. 2, the absorption bands of blood hemoglobin (420, 540, and 580 nm) and water (1450 nm) are clearly visible in the absorption spectra. The water peaks at 975 and 1195 nm, however, are less pronounced.

A drop in the scattering coefficient value is noticeable as wavelength increases, which is consistent with the general behavior of biological tissue scattering characteristics. The deviation from the monotony can be explained by the influence of strong absorption in the abovementioned areas.

As previously obtained data from the literature⁴⁰ are presented with our experimentally calculated transport coefficients, we are able to compare these outcomes. It is evident that they are in good agreement, with some of the observable differences attributed to the employment of different methods of experimental data processing, as well as differences in the properties of specific biological tissue samples.

It should be noted that the resulting dependence of the scattering coefficient, within the range of interest, could be approximated with high accuracy ($R^2=0.97$) using the Gunary power function⁴¹

Light penetration depth is one of the most critical characteristics for determining the capabilities of various methods of noninvasive optical diagnosis, including FS. Looking at Fig. 3, we can tell that this light penetration depth is around 0.5 to 2 mm in the spectral region of interest for fluorescent spectroscopy (350 to 600 nm). This allows for the stimulation of major biomarker (NADH, FAD, collagen, and so on) fluorescence.

Furthermore, use of experimentally obtained optical property data poses a distinct advantage over using literature-based data as displayed in our previous work.³⁶ Tissue samples spectroscopically measured for their reflectance and transmission properties were treated and kept at identical conditions to those undergoing fluorescence analyses, greatly reducing potential errors that arise from variation in laboratory conditions and different experimental practices. This also provides an avenue for further improvement of the calculated data by continuously expanding the sample size of experimentally measured tissue properties.

Following the calculation of transport coefficients, they were used in TracePro software (Lambda Research Corporation) to build the 3-D model.

4.2.

Three-Dimensional Model

The construction of the 3-D model was based on the Monte Carlo methodology, which remains one of the most effective simulation tools when dealing with biological tissues. This is especially true for multilayered tissues. Thus, a number of other studies have applied this methodology to create models based on tissue optical properties.^42–45 These incorporate multiple layers ranging from a single⁴⁴ to five⁴⁶ different ones. However, to date, the models have concentrated mainly on breast and cervix tissues, with none being produced or optimized for endoscopically reached hollow organs such as the bladder.

For the construction of our 3-D model, a two-layer setup was chosen. The physiological structure of the urinary bladder is easily simplified into two major layers, the muscle and the mucosa layers, without suffering in accuracy. This may be explained by the relatively large size of the modeled layers in comparison to omitted ones such as connective tissue. As the predominant constituent tissues of the bladder, they will be host to most of the tissue–light interactions. Other layers (e.g., connective tissue) would not drastically alter the model properties and would not be efficient to add to the model at this stage. Thus, these layers were not incorporated into the model. As the model described in this study was ultimately directed at aiding the detection of cancer, the contributing fluorophores selected were NADH, FAD, and collagen. These play a crucial role in observing tumor metabolism (through the redox ratio) and the structural extent of the tumor cells. Most of the other similar models, with the exception of Pavlova et al.⁴⁶ do not concentrate on FAD. The authors have found no similar models that look at the urinary bladder and incorporating the same fluorophores.

4.3.

Measured and Modeled Fluorescence

Using the model, simulated fluorescence spectra can be produced. Upon comparison with experimentally measured tissue spectra, we can see a good consistency presented in Fig. 5. Some of the observable differences may be characterized by the presence of unaccounted for fluorophores within the real tissue. Additionally, other absorbing chromophores may contribute to the modification of the experimentally obtained spectrum. While these differences were also observable in the previously presented model,³⁶ this newer model displays marked advances. Figure 5 displays a higher level of accuracy to past work, where even the effects of blood (hemoglobin peaks at 540 and 580 nm) are clearly visible both on the experimental and modeled spectra. Furthermore, it is possible to conclude that the shape of the experimentally obtained bladder tissue fluorescence spectrum is heavily influenced by the fluorescence of present NADH and collagen (for 365 nm excitation) or FAD and collagen (for 450 nm excitation). A similar conclusion was reached by Drezek et al.⁴⁷ and Pavlova et al.⁴⁸ when studying cervical tissue using fluorescence microscopy. The studies reported that mitochondrial NADH and FAD dominate epithelial cell fluorescence.

4.4.

Future Prospects

These very promising and substantial results demonstrate the ability of spectroscopic techniques to provide useful information for disease classification in a noninvasive manner. Although each of the techniques discussed in this article shows great potential as a means of detecting dysplasia in bladder, their combination should allow us to create a comprehensive picture of the biochemical and morphological stages of tissue. Specifically, fluorescence changes of biochemicals such as NADH, FAD and collagen will provide details about urinary bladder tissue biochemistry. This paper details marked improvements to our initial 3-D model by incorporating measured tissue data in favor of literature-based values. With further expansion of data sets and tissue types, this model has the potential to greatly aid in optical diagnostics of urinary bladder diseases such as cancer, when used in conjunction with other optical diagnostics tools. Combined with established methods of detection, use of such a model with non- or minimally invasive fluorescence measurements can aid in increasing the detection rate of BC, while helping reduce the number of false-positive diagnoses. Such improvements can lead to not only an increase in the quality of life of patients but also a great cost reduction and thus a reduction in the economical strain on health services. Use of human tissue can further improve the functionality of the model, advancing it in terms of accuracy and relevance. The use of both healthy and cancerous tissues can drastically improve the viability of this model in potential medical applications.

Additionally, based on the developed optical model and further computer simulations, the general distribution of light within tissue can be predicted. This can be invaluable in a variety of noninvasive optical methods of investigation.