We present a preliminary investigation of macroscopic polarimetric imaging of uterine cervix. Orthogonal state contrast (OSC) images of healthy and anomalous cervices have been taken in vivo at 550 nm. Four ex vivo cervix samples have been studied in full Muller polarimetry, at 550 nm and 700 nm, and characterized in detail by standard pathology. One sample was totally healthy, another one carried CIN lesions at very early stage (CIN1) in its visible exocervical region, while for the other two samples more advanced (CIN3) lesions were present, together with visible glandular epithelium (ectropion). Significant birefringence has been observed in the healthy regions of all six samples, both in vivo and ex vivo. Standard treatments of the Mueller images of the ex vivo samples allowed to quantify both retardation and depolarization. Retardation reached 60° in healthy regions, and disappeared in the anomalous regions of the other three ex vivo samples. The depolarization power was largest in healthy regions, and lower in CINs and ectropion. Possible origins of the observed effects are briefly discussed.
© 2013 OSA
Cervical cancer is the second most frequent cancer affecting women, and causes 275 000 deaths per year worldwide , mostly but not exclusively in developing countries. In the vast majority (95 to 98%) of cases the disease is due to infection by various types of Human Papillomavirus (HPV) . The dysplastic lesions, also called CIN (for Cervical Intraepithelial Neoplasia) start at the basal membrane at the bottom of the malpighian epithelium and in many cases they disappear spontaneously, as the immune system takes control on the infection. If not, the malignant cells progress towards the surface of the epithelium. The disease is staged CIN1 or CIN2 when one third or two thirds of the epithelium are respectively transformed and CIN3 when the transformation affects the whole epithelium. At this point, spontaneous regression is highly unlikely. The basal membrane eventually gets broken and the disease evolves into invasive cancer.
However, the overall evolution is very slow, with typically five to ten years between the infection and the onset of invasive disease. As a result, cervical cancer ideally meets the criteria defined by the World Health Organization for effective management by screening. The current recommendation in France is an organized screening of women aged 25 to 65 with Papanicolaou (Pap) smear every third year, after two normal Pap tests. Women with anomalous Pap smears are referred to colposcopy: the cervix is examined by means of a low magnification binocular microscope (colposcope), with successive stainings by diluted acetic acid and Lugol’s iodine. The dysplastic regions show some whitening with acetic acid and are iodonegative. These criteria are used to identify the most suspicious regions which are then biopsied to get a pathology diagnosis. Finally, when a high grade CIN is diagnosed, the patient is treated by a conization, which is the surgical removal of the diseased part of the cervix.
Colposcopy is notoriously difficult and operator-dependent. Taken alone, its sensitivity for the detection of high grade dysplasias (CIN 2 and CIN3) is about 60% - 70%, for a specificity of the order of 50% [3–5]. Moreover, due to the difficulty to correctly visualize the lesions, whenever a conization is necessary its margins are difficult to define accurately.
The recent introduction of vaccines is not likely to suppress the need to improve colposcopy, as these vaccines do not offer full protection against all types of HPV. Moreover, these vaccines are preventive but not curative, implying that the need to manage the disease for large populations will certainly remain for decades. Last but not least, if colposcopy performance and ease-of-use were dramatically improved, this technique might provide a simple, low-cost and efficient alternative to Pap test as a first screening step in low resource areas.
Several techniques have been investigated to improve the performances of colposcopy, such as spectrally resolved reflectance, fluorescence imaging, in vivo confocal microscopy, optical coherence tomography (OCT) [6–13] but unfortunately, so far none of them seems to have shown any significant advantage in everyday clinical practice, which is thus not really satisfactory.
Polarimetric imaging is a possible alternative in this respect. This technique can be implemented in various ways, among which the relatively simple Orthogonal State Contrast (OSC) and the most complete (and complex) Mueller polarimetry. OSC has proven useful to image skin pathologies , and define the surgical margins for melanomas . Mueller polarimetry has been used to characterize healthy and cancerous human colon  and distended rat bladders , among other examples.
In this work we present a case study of polarimetric imaging of cervical samples, both in vivo and ex vivo. The paper is organized as follows. In the next section we describe typical images obtained during a clinical evaluation of OSC for the detection of cervical dysplasias. This study revealed an unexpected optical anisotropy of healthy cervical tissue, which was confirmed by complete Mueller imaging of ex-vivo conizations, as shown in section 3. These results are discussed in section 4. Section 5 concludes the paper.
2. In vivo polarimetric imaging by orthogonal state contrast (OSC)
This investigation, realized at Institut Mutualiste Montsouris, was actually the first clinical evaluation of polarimetric imaging for the improvement of the performance of colposcopy. Our hypothesis was that, due to the lack of clearly visible anisotropy such as that seen in tendons or striated muscles, the cervical tissue would behave as an isotropic depolarizer, pretty much like the colon samples we had studied previously. Moreover, the few preliminary observations that we could carry out on ex vivo conizations seemed to support this hypothesis.
OSC then appeared ideally suited for this study: on the one hand it provides quantitative measurements of the depolarization power of pure depolarizers; on the other hand, it is relatively simple to implement even for real time operation. Its principle consists in acquiring two images, III and I⊥, with linear polarizations in both the illumination and the detection arms, these polarizations being respectively parallel (for III) or perpendicular (for I⊥) to each other. The Orthogonal State Contrast image IOSC is then calculated as
From its very definition, Iosc is independent of the sample overall reflectivity, and depends only on its polarimetric properties. Moreover, as the normalized Mueller matrix of spatially isotropic pure depolarizers takes the form
The setup used for this investigation is described in ref . Briefly, a removable optical system is added in front of a colposcope, with a linear polarizer in the illumination beam, and a ferroelectric liquid crystal and a linear analyzer in the detection arm. This stack of two elements is equivalent to a linear analyzer which could be switched between two perpendicular orientations in less than 1 ms. This system allowed real time display of OSC images at 8 frames per second. Moreover, the azimuth of the (common) linear polarization of III could be changed at will by rotating a gear. This additional parameter was added to allow optimization of the contrast in presence of the (weak) diattenuation possibly introduced by the conical shape of the cervix.
Figure 1 summarizes typical in vivo results, with two examples, one of a healthy cervix, and another one with a pathological part identified by an experienced colposcopist. The whole healthy cervix, as well as the healthy parts of the other one, exhibited a strong variation of the IOSC contrasts when the azimuth was varied, with a 90° periodicity. On the other hand, in the anomalous region the IOSC signal did not exhibit any visible dependence on the azimuth. In other words, only the pathological region behaved as expected for an isotropic depolarizer, while in the healthy tissue unexpected strong optical anisotropy is dominant.
Two essential conclusions may be drawn from these results
- • Our basic assumption that cervical tissue behaves like a pure depolarizer is obviously wrong at least for healthy parts,
- • OSC imaging is not adequate for such samples, as it does not provide enough information to extract the relevant polarimetric parameters for healthy and anomalous tissues.
To properly characterize the polarimetric response of healthy and pathological cervical tissue, full Mueller polarimetry is clearly needed. As this technique is not yet available for use in vivo, and due to the need of accurate histological characterization of the imaged tissues to correctly interpret the polarimetric parameters, we started a systematic study on conizations, whose first significant results are summarized in the next section.
3. Mueller polarimetric imaging of ex-vivo conizations
3.1 Surgical and sample preparation procedures
The samples were obtained by standard and recommended surgical methods . We made conizations using the loop electrical excision procedure under colposcopy. The samples were imaged quickly after surgery, and prior to fixation in formalin.
The subsequent pathology procedure was quite standard: the samples were cut in pieces a few millimeter thick, dehydrated and included in wax. Then from each piece a few 5 µm thick slice was cut with a microtome, rehydrated, stained with hematoxylin-eosin to make standard histological plates which were eventually examined microscopically.
Special care was taken to determine as accurately as possible the position of the examined tissues on the initial image of the conization specimen. To this end, the surgical sample was marked with various colors prior to cutting, based on polarimetric contrasts, to guide the cutting process. The final overall uncertainty in the positioning of the pathology data on the polarimetric images is estimated to be about 2 mm which is much less than the typical size of the analyzed zones.
3.2 Mueller imaging and data treatment
The Mueller imager used to study the surgical samples is outlined in ref . The illumination arm includes a Polarization State Generator (PSG) comprising a linear polarizer and two liquid crystal cells. The output arm consists of a Polarisation State Analyser (PSA), a mirror image of the PSG, a spectral filter (FWHM 20 nm) to select the wavelength and a system of lenses to properly image the sample on the sensor.
The PSG and PSA modulate and analyze the input and output polarizations according to the scheme defined in ref , which also outlines how the instrument is calibrated. To acquire a complete Mueller matrix, the PSG generates four different incident polarizations by changing the retardations of the NLCs, and for each of these, the PSA determines the emerging light polarization by measuring the transmitted intensity through four different “polarization filters” generated by changing the retardations of the NLCs set before the linear analyser. The instrument thus provides a set of 16 elementary images which are transformed into the full Mueller matrix by suitable linear operations which take into account the calibration of the instrument.
Once this matrix is obtained, the elementary polarimetric properties of the sample may be retrieved by a number of procedures. In this study we used the most common one, proposed by Lu and Chipman in 1996 . Basically, the sample Mueller image M is decomposed pixelwise into a product of three matrices
Cervical tissues may exhibit many histological variations, such as the presence of glandular epithelium in the exocervix (ectropion), malpighian metaplasias, where glandular epithelium is replaced by squamous epithelium, the different stages of CIN…This variety makes sensitive and specific diagnosis of CIN lesions difficult. For this first report on polarimetric imaging of uterine cervix, we chose to present a few samples with well characterized tissue alterations. In all cases, our gold standard is usual pathology, which provides diagnostics only along the cuts made after fixing the sample. Those cuts were a posteriori positioned on the birefringence and depolarization images extracted from the initial Mueller images.
As a first sample, we chose a specimen of healthy cervix (taken on a complete hysterectomy made for the treatment of menorrhagia). The results are summarized in Fig. 2. The images were taken at 550 nm, and exhibit a quite uniform and strong depolarization, as well as quite strong retardance, up to about 60°, excepted in a narrow region delimited by a thin solid line in the retardance image (lower left panel) which seems to correspond to a “depression” in the exocervix surface. This region was not characterized histologically. The orientation α of the slow axis (the direction of polarization with the larger optical index), displayed on the lower right panel, appears to vary widely from spot to spot within the sample, in agreement with the results of our previous in vivo study.
The second sample was a conization where only healthy and CIN1 tissues were found in the observable region. The data are shown in Fig. 3. The CIN1 lesions are mainly found below the entrance of the endocervix on the posterior lip of the cervix in its central region, and also above it, at the extreme left of the figure (thus on the right side for the patient). Polarimetric images, taken at 550 nm as in the previous case, clearly show a strong decrease of scalar retardation in the central part, where most CIN1 tissue was found. Another CIN1 region was reported at the extreme left corner, where again retardation was found to be much lower than in healthy regions. However, in this region depolarization exhibits a rapid spatial variation, making any accurate comparison with histology difficult, due to the positioning accuracy of pathology cuts, estimated to be about 2 mm.
Finally we point out that the orientation of the slow axis becomes somewhat “erratic” close to the endocervix, on the posterior lip and in the upper left corner of the figure, i.e; in the regions where the scalar retardation is much smaller than in healthy regions.
The third sample, shown in Fig. 4, has been characterized by pathologists in five planes. It exhibits a large part of healthy tissue (in the lower left part of the images), a CIN3 lesion close to and above the endocervix, and bare glandular epithelium in the lower right part of the images. We observe again strong scalar retardation and depolarization in the healthy part. In contrast, retardation decreases to very small values in both CIN3 and glandular parts. However, these two types of tissues are clearly distinguished in the depolarization image: depolarization indeed increases from glandular tissue (the lowest value) to CIN3 and then to healthy. Finally, we see again a strong variation of the retardation axes orientation around the endocervix.
We now describe our results on the fourth and last sample. Figure 5 shows the classical colposcopic images (left panel) taken in vivo before realizing the conization and the histological characterization of the extracted sample (right panel). On the colposcopic image taken after application of diluted acetic acid (Fig. 5(a)), the observed weak whitening is difficult to interpret. With iodine, a wide iodonegative zone is observed on the anterior lip (Fig. 5(b)), with a brownish yellow shade characteristic of CIN3, except of the very bottom of the lip, where a pink stripe is seen and interpreted as the bare glandular tissue found inside the junction zone. During surgery, only the anterior lip has been excised so that the endocervix is at the very bottom of the extracted sample. This sample has been histologically characterized in 8 planes, which are shown by colored lines superimposed on an intensity image taken with the polarimeter (right panel of Fig. 5). This analysis confirms the presence of CIN3 in the middle of the sample, and bare glandular tissue at the bottom, close to the endocervix. However, the top and the left parts of the sample feature healthy tissue, in zones which appeared iodonegative during colposcopy. A tiny spot of CIN1 has been found close to the healthy tissue in the middle of the sample. Finally, at the extreme right thick glandular tissue is found beneath the superficial CIN3. Two histological cuts are shown at the bottom right of Fig. 5: in the AB plane, the parts diagnosed as healthy, CIN3 and bare glandular, while in CD plane, the “buried” glandular tissue is visible beneath the CIN3.
Figure 6 shows the polarimetric images of the fourth sample, taken at 550 nm and 700 nm. At 550 nm we observe again strong retardation and depolarization in healthy regions. Retardation vanishes already at CIN1 stage, and everywhere else (with CIN3 or glandular tissue in surface). Depolarization seems to decrease from healthy to CIN1 to CIN3 to glandular tissue. When the wavelength is increased to 700 nm, we observe that:
- • The retardation image is practically unchanged, with respect to the shape of the observed patterns as well as the absolute values of R.
- • The depolarization increases everywhere, but with a significant change in its shape: the measured Δ seems almost the same for healthy and CIN1 on the left and CIN3 at the center of the figure, while it is lower at the bottom center and the bottom right, where glandular tissue is respectively present at the surface (cut AB) of the sample or a few mm deep (cut CD).
Let us consider birefringence first. This unexpected effect has been seen in the healthy regions of all six samples, and vanished in all anomalous spots (CINs, but also ectropion). Though much more work is needed to fully assess the conditions of appearance of birefringence and its real value for the diagnosis of CINs, the data shown here already deserve some tentative interpretation.
As no collagen, nor any other fibrillar and ordered protein is present in the malpighian epithelium, we propose that the observed birefringence comes from the connective tissue beneath the epithelium itself. The fibrillar proteins in this tissue would then exhibit a spatial organization which has been overlooked so far, and which would get disrupted if the nearby epithelium develops a cervical intraepithelial neoplasia. This hypothesis seems quite plausible, as there is growing evidence that a precancerous evolution strongly affects the neighboring stroma. Light scattering in stroma has been observed to decrease close to dysplastic epithelium, an effect attributed to a degradation of collagen fibers . Modifications of the stromal structure close to high grade CIN has also been inferred from multifractal analysis of stromal refractive index , or by transmission Mueller imaging of histological plates of cervical tissue with the epithelium and underlying stroma . This latter study showed a significant decrease in the retardation in the stromal regions close to dysplastic epithelium with respect to normal ones. In the same study, depolarization in the epithelium was found to increase in dysplastic regions, in apparent contradiction with our data. However, depolarization is very sensitive to the observation conditions, and cannot be compared in forward imaging of thin samples and backward scattering on thick samples. In comparison, the evolution of stromal retardance is more easily comparable.
Our data also suggest a variation of depolarization power with the nature of the observed tissues. Again, this trend may a priori be due to changes in tissue properties both in epithelium and stroma, and may involve collagen fibers as well as other (presumably small) scatterers. However, in contrast with retardance, which does not change when the wavelength is increased to 700 nm, at this wavelength depolarization seems better correlated with deep structures rather than superficial ones. Therefore, the data taken at 550 nm certainly contain information about the epithelium histology, and this information should be retrieved more accurately by suitably combining the data at both 550 nm and 700 nm than from the images at 550 nm alone.
We investigated the polarimetric response of six uterine cervix sampeles, two of which in vivo, with Orthogonal State Contrast, and another four ex-vivo, with complete Mueller polarimetry. In all samples healthy malpighian epithelium with underlying stroma appears birefringent and strongly depolarizing. This birefringence, which is attributed to the organization of stromal collagen, disappears in the anomalous zones (CINs or apparen glandular epithelium). In non-birefringent zones, the depolarization varies differently in the green and red parts of the spectrum, with a stronger correlation with the epithelium histology in the green and with deeper tissues in the red. Of course, much more work is needed on many different samples to refine the observations reported in this work and to evaluate the possibilities of the technique for CIN diagnosis purposes.
This research was funded by the Institut National du Cancer (INCa) and the Cancéropôle, under contract 2012-GYN-01-EP-1.
References and links
1. J. Ferlay, H. R. Shin, F. Bray, D. Forman, C. Mathers, and D. M. Parkin, “Estimates of worldwide burden of cancer in 2008: GLOBOCAN 2008,” Int. J. Cancer 127(12), 2893–2917 (2010), Globocan.iarc.fr/. [CrossRef] [PubMed]
2. D. R. Lowy, D. Solomon, A. Hildesheim, J. T. Schiller, and M. Schiffman, “Human papillomavirus infection and the primary and secondary prevention of cervical cancer,” Cancer 113(7Suppl), 1980–1993 (2008). [CrossRef] [PubMed]
3. M. Underwood, M. Arbyn, W. Parry-Smith, S. De Bellis-Ayres, R. Todd, C. W. Redman, and E. L. Moss, “Accuracy of colposcopy-directed punch biopsies: a systematic review and meta-analysis,” BJOG 119(11), 1293–1301 (2012). [CrossRef] [PubMed]
5. S. B. Cantor, M. Cárdenas-Turanzas, D. D. Cox, E. N. Atkinson, G. M. Nogueras-Gonzalez, J. R. Beck, M. Follen, and J. L. Benedet, “Accuracy of colposcopy in the diagnostic setting compared with the screening setting,” Obstet. Gynecol. 111(1), 7–14 (2008). [CrossRef] [PubMed]
8. R. D. AlvarezT. C. Wright JrR. D. AlvarezT. C. Wright JrOptical Detection Group, “Increased detection of high-grade cervical intraepithelial neoplasia utilizing an optical detection system as an adjunct to colposcopy,” Gynecol. Oncol. 106(1), 23–28 (2007). [CrossRef] [PubMed]
9. R. D. AlvarezT. C. WrightR. D. AlvarezT. C. WrightOptical Detection Group, “Effective cervical neoplasia detection with a novel optical detection system: a randomized trial,” Gynecol. Oncol. 104(2), 281–289 (2007). [CrossRef] [PubMed]
10. M. Cardenas-Turanzas, J. A. Freeberg, J. L. Benedet, E. N. Atkinson, D. D. Cox, R. Richards-Kortum, C. MacAulay, M. Follen, and S. B. Cantor, “The clinical effectiveness of optical spectroscopy for the in vivo diagnosis of cervical intraepithelial neoplasia: where are we?” Gynecol. Oncol. 107(1Suppl 1), S138–S146 (2007). [CrossRef] [PubMed]
11. T. DeSantis, N. Chakhtoura, L. Twiggs, D. Ferris, M. Lashgari, L. Flowers, M. Faupel, S. Bambot, S. Raab, and E. Wilkinson, “Spectroscopic imaging as a triage test for cervical disease: a prospective multicenter clinical trial,” J. Low. Genit. Tract Dis. 11(1), 18–24 (2007). [CrossRef] [PubMed]
12. R. A. Drezek, T. Collier, C. K. Brookner, A. Malpica, R. Lotan, R. R. Richards-Kortum, and M. Follen, “Laser scanning confocal microscopy of cervical tissue before and after application of acetic acid,” Am. J. Obstet. Gynecol. 182(5), 1135–1139 (2000). [CrossRef] [PubMed]
13. S. W. Lee, J. Y. Yoo, J. H. Kang, M. S. Kang, S. H. Jung, Y. Chong, D. S. Cha, K. H. Han, and B. M. Kim, “Optical diagnosis of cervical intraepithelial neoplasm (CIN) using polarization-sensitive optical coherence tomography,” Opt. Express 16(4), 2709–2719 (2008). [CrossRef] [PubMed]
15. S. L. Jacques, R. Samatham, S. Isenath, and K. Lee, “Polarized light camera to guide surgical excision of skin cancers,” Proc. SPIE 6842, 68420I (2008). [CrossRef]
16. A. Pierangelo, A. Benali, M. R. Antonelli, T. Novikova, P. Validire, B. Gayet, and A. De Martino, “Ex-vivo characterization of human colon cancer by Mueller polarimetric imaging,” Opt. Express 19(2), 1582–1593 (2011). [CrossRef] [PubMed]
17. S. Alali, K. J. Aitken, A. Schröder, D. J. Bagli, and I. Vitkin, “Optical assessment of tissue anisotropy in ex vivo distended rat bladders,” J. Biomed. Opt. 17(8), 086010 (2012). [CrossRef] [PubMed]
18. M. Anastasiadou, A. De Martino, D. Clément, F. Liège, B. Laude-Boulesteix, N. Quang, J. Dreyfuss, B. Huynh, A. Nazac, L. Schwartz, and H. Cohen, “Polarimetric imaging for the diagnosis of cervical cancer,” Phys. Status Solidi 5(5c), 1423–1426 (2008). [CrossRef]
19. P. P. Martin-Hirsch, E. Paraskevaidis, A. Bryant, H. O. Dickinson, and S. L. Keep, “Surgery for cervical intraepithelial neoplasia,” Cochrane Database Syst. Rev. 16 (2010).
20. A. Pierangelo, S. Manhas, A. Benali, C. Fallet, J. L. Totobenazara, M. R. Antonelli, T. Novikova, B. Gayet, A. De Martino, and P. Validire, “Multi-spectral Mueller polarimetric imaging detecting residual cancer and cancer regression after neoadjuvant treatment for colorectal carcinomas,” J. Biomed. Opt. (to be published).
22. S.-Y. Lu and R. A. Chipman, “Interpretation of Mueller matrices based on polar decomposition,” J. Opt. Soc. Am. A 13(5), 1106–1113 (1996). [CrossRef]
23. D. Arifler, I. Pavlova, A. Gillenwater, and R. Richards-Kortum, “Light Scattering from Collagen Fiber Networks:Micro-Optical Properties of Normal and Neoplastic Stroma,” Biophys. J. 92(9), 3260–3274 (2007). [CrossRef] [PubMed]
24. N. Das, S. Chatterjee, J. Soni, J. Jagtap, A. Pradhan, T. K. Sengupta, P. K. Panigrahi, I. A. Vitkin, and N. Ghosh, “Probing multifractality in tissue refractive index: prospects for precancer detection,” Opt. Lett. 38(2), 211–213 (2013). [CrossRef] [PubMed]