1. Introduction

Optical coherence tomography (OCT) [1–3] is a noninvasive method for 2D imaging of internal micro structure of transparent [1] and highly scattering objects [2]. This method provides in vivo information on the internal structure of biological objects with high resolution and in real time. Multiple experiments conducted by different research groups have shown that OCT is sensitive to structural alterations in biological objects that occur at the level of cell groups and tissue layers [4–7]. Our research team studied more than 1000 patients with different pathologies [8–11]. The clinical studies show that morphological structures and tissue layers have differing optical parameters due to their spatial organization. In the OCT images healthy tissues are visualized as several layers with distinct borders. Because of the difference in scattering properties of the epithelium and fiber connective-tissue structures, OCT can image layer by layer the architectonics of healthy lining tissues. As pathology develops, the tissue structure changes. OCT can detect the morphological changes in different pathologies. Specifically, neoplastic processes are characterized by most pronounced reorganization of tissue structure. OCT images of cancer are structureless and, as a rule, show fast signal decay with depth. OCT is informative with regard to structural alterations in tissue, although it does not specify the origin of these alterations. Using OCT imaging, it is very difficult to differentiate inflammatory processes, papillomatosis, cancer, and scar changes (Fig. 1,4a,5a).

In many pathologies structural violations are preceded by biochemical and initial morphological changes. It is known that some structural components of biotissue, e.g., stromal collagen fibers that constitute the basis of healthy mucosa, can strongly depolarize incident radiation [12]. Also fibrous tissues such as collagens are linear birefringent, i.e., they change the polarization state of light, depending on the value of birefringence and penetrated tissue depth [13]. Both these processes lead to the appearance of cross-polarized component in backscattered light. Pathological processes with different origin are characterized by the difference in both the amount of collagen fibers and their spatial organization. Therefore, a comparative analysis of cross-polarization backscattering properties of biological objects may be taken as an underlying point of the technique for early diagnosis of neoplastic processes.

 

Fig. 1. Standard OCT images printed in logarithmic scale of different diseases a) laryngitis (chronic inflammation of larynx) b) laryngeal cancer c) laryngeal papilloma. White bar corresponds to 1 mm.

Download Full Size | PDF

The specificity of standard OCT can be improved by studying polarization properties of probing radiation when it propagates through a biological object. This approach was implemented in the polarization-sensitive OCT technique (PS OCT) [4,14–19]. PS OCT provided important information on the internal structure of biological objects that cannot be obtained with standard OCT, for example, burn depth [14–16], tooth demineralization in dentistry [8,17], early diagnostics of osteoarthritis [18]. At present, in the majority of papers on PS OCT, the criterion of pathological changes in tissue is a sharp decrease in its macroscopic birefringence. For early diagnostics of neoplastic processes, a reliable signal reception at depths 400-700 μm is required. To correctly determine phase characteristics, such as birefringence, the signal-to-noise ratio should be not less than 10-15 dB, which is difficult to achieve when studying layers at depths more than 300-500 μm [14,17]. For deeper layers (up to 1.5 mm) a variant of PS OCT – the cross-polarization OCT (CP OCT) can be employed. The CP OCT technique is based on the detection of a backscattered component that is orthogonal to linearly polarized probing radiation [8,19]. In our previous papers we pointed out the possibility of early diagnostics of neoplastic processes using CP OCT [4,20].

The aim of this paper is to study the possibility of differentiating between healthy, neoplastic and scar tissues using complementarily CP OCT and standard OCT. The differential diagnosis relies on obtaining maps of two independent physical characteristics – backscattering coefficient and cross-polarized backscatter degree of healthy and pathological tissues. We associate the difference in cross-polarization backscattering properties of biotissues in this case with the amount, localization and spatial organization of stromal collagen fibers.

2. Materials and methods

A sketch of the experimental setup used to map the backscattering coefficient and cross-polarized backscatter degree is shown in Fig. 2. Using a multiplexer, low-coherence near-IR radiation (λ=1.3μm) from a superluminescent diode (SLD) with a coherence length l_c=21μm is combined with radiation from a semiconductor red laser (RL) used for alignment purposes. Then one of polarization eigenmodes of a polarization-maintaining (PM) 3dB fiber coupler is selected by means of Lefevre polarization controller (CP). PM fiber is used to transport radiation with a certain polarization state both in the signal and reference arms. When there is no Faraday rotator (F) in the reference arm, a co-polarized component of backscattered radiation is recorded. The influence of the Faraday rotator consists in the rotation of an arbitrary polarization state by a specified angle and the direction of the rotation depends only in the direction of the magnetic field inside the rotator and does not depend on the propagation direction of the radiation [21]. Therefore, in case of the 45° Faraday rotator, the radiation passes through it, gets reflected by a mirror, goes back through the rotator and becomes orthogonally polarized in the reference arm. As a result, only the light component which is cross-polarized by the biological object interferes. In [19] a quarter wave plate oriented at 45⁰ to the incident polarization was used for this purpose. We use the Faraday rotator instead because it does not need the angle alignment, therefore minimizing the realignment time for the whole system. The readjustment of the system takes 30 s. The acquisition time for one OCT image is 1 s. For all OCT images the logarithmic intensity scale is used. The lateral resolution of the system determined by the waist diameter of the probing beam is chosen very close to the longitudinal (in depth) resolution which is determined by l_c=21μm. It should be noted that when the system was readjusted to obtain images in the orthogonal polarization the position of the probe was fixed. Therefore, both types of images from one and the same place were obtained. Since this design is based on PM fiber, a portable setup with flexible probe can be created, making it easy to use in clinical applications, e.g. endoscopically [4].

 

Fig.2 The experimental setup for the cross-polarization OCT. CS – crossectional scanner, O – investigated object, PS – longitudinal piezo-scanner, L – lenses, PD – photodiode, SA – selective amplifier, LA – logarithmic amplifier, AD – amplitude detector, ADC – analog to digital converter, PC – personal computer. Bold line corresponds to single-mode fiber, thin line - to polarization maintaining fiber.

Download Full Size | PDF

OCT images of healthy and cancerous esophagus were acquired ex vivo on resected esophagus (no more than 60 min after extirpation). OCT images of scar changes during esophagoscopy were obtained in vivo. All experiments were conducted according to IRB approved protocol; the patients’ consent was obtained.

To verify OCT and CP OCT images we made parallel analysis of biopsy samples taken from the same tissue regions where OCT imaging was performed. The biopsy samples were H&E and Van Gieson stained [22]. The latter staining is specific for collagen fibers of connective tissue.

3. Results

The results of OCT study of healthy esophagus are presented in Fig. 3. Tomograms of unaltered esophageal mucosa obtained in both polarizations have a layered horizontally organized pattern. In the direct polarization (Fig. 3a) the epithelium is seen as a moderately scattering zone with distinct boundary with the higher backscattering underlying stroma. In the orthogonal polarization (Fig. 3b) the epithelium appears as a very poorly scattering layer. Nuclei of the epithelium in Fig. 3d,e have brown staining. The main fibrous component of the stroma is collagen fibers (red staining in Fig.3d,e), which provide, according to paper [12], efficient depolarization and according to paper [13] birefringence of tissue. These explain the presence of an intense signal in the CP OCT images. The signal level in the CP OCT image is 18-20 dB lower compared to the standard OCT image. In the stroma, stripes that are oriented along the epithelium are seen in the CP OCT image (indicated by arrows). These structures correlate well with collagen fiber bundles oriented along the epithelium (Fig. 3e). The transverse size of these collagen bundles in Fig. 3e and of the stripe structures in Fig. 3b is 70-80 μm.

 

Fig. 3. a) Standard OCT image, b) CP OCT image c) H&E histology d), e) Van Gieson histology of healthy esophagus. White bar corresponds to 1 mm.

Download Full Size | PDF

Images of carcinoma and scar tissue of the esophagus are shown respectively in Fig.4,5. Standard OCT images (Fig. 4a and Fig. 5a) are scarcely distinguishable. The both images are structureless. Therefore, using standard OCT it is impossible to differentiate between neoplastic and scar changes. The diagnosis in this case is based on endoscopy and histology (Fig. 4c and Fig. 5c). However, CP OCT images of these pathologies (Fig. 4 b and 5 b) are considerably different. Cancer cells almost do not cross-polarize probing radiation and the signal level is 10 dB lower on the average than that in CP OCT images of healthy tissue. In the CP OCT images vertically-oriented regions of stronger signal are noted against this background of weak signal (Fig. 4b). These images correlate with single vertically-oriented collagen fibers in Fig. 4d, where they are visualized as red elongated individual structures.

 

Fig. 4. a) Standard OCT image, b) CP OCT image c) H&E histology d),e) Van Gieson histology of cancerous esophagus. White bar corresponds to 1 mm where not specially marked.

Download Full Size | PDF

CP OCT images of scar tissue of the esophagus show considerable signal level, comparable to that in healthy tissue (Fig. 5b). At the same time, in the CP OCT image one can note quite a large number of chaotically oriented regions of both intense and weak signal. This is because the nature of scar tissue organization is different than in cancer. It is seen from Fig. 5d that collagen fibers are one of the main components of an immature scar (pink regions correspond to maturing collagen). In Fig. 5d the collagen regions are alternating with regions of cell accumulation of granular tissue, which correlates well with the signal behavior in the CP OCT image of scar. The difference in structural features of collagen fibers in cancer and scar tissue forms the basis of CP OCT differentiation of these pathologies since their cross-polarization backscattering properties are determined, in a considerable degree, by anisotropic structures, i.e., by collagens.

 

Fig. 5. a) Standard OCT image, b) CP OCT image c) H&E histology d),e) Van Gieson histology of scar esophagus. White bar corresponds to 1 mm where not specially marked.

Download Full Size | PDF

4. Discussion

The value of macroscopic birefringence of native collagen as measured by Maitlaind and Walsh in rat tendon is Δn = (3 ± 0.6) ∙ 10^-3 [13]. The period of intensity oscillations in CP OCT image due to this should be equal to the quarter of the beat length which covers the range from 360 μm to 540 μm for a wavelength of 1.3 μm. Thus macroscopic birefringence could be the reason of the stripes in Fig.3,b because the distance between them is about 100-140μm. Since in other CP OCT images such periodic structures are not seen, we attribute the signal on them to depolarization properties of biotissues [12,19]. The non-periodical signal in Fig.3,b we also attribute to depolarization properties of healthy esophagus.

In the technique presented herein the sample illuminates by plane polarized light. Therefore it is sensitive to orientation of fiber-like structures, such as collagen, with respect to the polarization plane. Indeed we observed the difference of signal intensity of up to 15 dB in CP OCT images of kevlar fibers when we rotated the probe in the plane parallel to the orientation of fibers. Thus it could have been a disadvantage as compared to other PS OCT techniques where the sample is either illuminated with circularly polarized light or subsequently with different polarization states. But regardless of the fact that stromal collagen fibers are oriented along the epithelium, they are chaotically oriented in the plane parallel to the tissue surface [12]. It means that CP OCT images qualitatively should not depend on the direction of probing beam polarization. This statement is confirmed by our studies of all three types of tissues considered above: healthy, cancerous and scar esophagus.

Compared to other PS OCT systems that record two polarization states simultaneously and therefore get the total information on intensity and polarization simultaneously [14–16], the method presented in the paper needs two subsequent measurements with a realignment time of 30 seconds. For in vivo measurements the influence of reflex movement of the object should be avoided. The usual way here is to reduce the acquisition time up to 1 second and less. In our technique we avoided movement artifacts by gently touching investigated tissue by probe, i.e. during the process of image acquisition a probe is moving together with the patient’s movement. During the realignment of the reference arm to process the image in the orthogonal polarization the probe is held on tissue, therefore no relocation of the same area is needed. Thus the acquisition time of 30 seconds for two polarizations measurements for in vivo studies is not convenient but accessible. At the moment we are working on creation of a new setup for simultaneous recording of standard and cross-polarization OCT images. For this purpose instead of the 45⁰ Faraday rotator (see Fig.2) a 22.5⁰ Faraday rotator in the reference arm will be employed. In this case light reflected by mirror MR and double passed through the 22.5⁰ Faraday rotator will be plane polarized at an angle of 45⁰ to the eigenmodes of the PM fiber in the reference arm. Thus the eigenmodes will be excited with equal weight. The amplitudes of the interferometric signal in the cross-polarization and co-polarization axes will be determined by tissue backscatter components in corresponding polarization. To simultaneously record the interferometric signal in both polarizations spatially resolved by Wollaston, Glan-Laser polarizers or by a polarizing beam-splitter, two photodetectors will be used, like in other PS OCT techniques [14–16].

5. Conclusions

The experimental setup provides successive maps of backscattering coefficient (standard OCT images) and cross-polarized backscatter degree (CP OCT images) from one and the same site on a sample at readjustment time of 30 s. It is shown that standard OCT and CP OCT, when used complementarily, can differentiate between healthy mucosa, neoplastic and scar tissues of the esophagus. The cross-polarized backscatter degree of probing radiation in neoplastic tissues is on the average 10 dB lower than in healthy and scar tissues. In the CP OCT images, scar tissue does not have the layered structure that is evident in CP OCT images of healthy tissue. Van Gieson histology shows that the changes in cross-polarization backscattering properties of biotissues correlate well with the quantity, localization and spatial organization of collagen fibers in stroma. CP OCT can provide additional information on cross-polarization backscattering properties of biotissues, thereby improving the diagnostic value and informativity of standard OCT imaging.