1.

Introduction

Ovarian cancer is the fifth leading cause of cancer deaths¹ in women and the most lethal gynecological malignancy, with a 5-year survival rate of only 40%.² Because the disease is primarily asymptomatic at early stages or associated with nonspecific symptoms, the majority of ovarian cancers is diagnosed at a late stage when the disease has already spread within the peritoneal cavity and curative strategies are largely ineffective.^3,4 Screening methods, such as transvaginal ultrasound (US) and blood levels of biomarker CA125, have been ineffective with little impact in reducing mortality and have, instead, generated greater distress and risk of unnecessary surgery.^3,5 Consequently, both the US Preventive Services Task Force and the American College of Obstetricians and Gynecologists do not recommend routine screening for ovarian cancer in the general population or among asymptomatic women, and there is insufficient evidence to implement routine screening for women who carry genetic mutations that place them at high risk.^5,6 Women in the general population have a 1% to 2% life risk of developing ovarian cancer, whereas women with familial history or with mutations in the homologous recombination DNA repair genes BRCA1 and BRCA2 have a significantly higher predisposition to ovarian cancer. Women carrying BRCA1 and BRCA2 mutations face a lifetime risk of 50% and 20%, respectively.⁵ However, when detected at an early stage, localized ovarian cancer has a much improved 5-year survival rate that is $>90\%$.^3,5 For high-risk women, it is especially crucial to develop effective screening strategies for early detection and prevention of advanced disease.

The standard of care for advanced ovarian cancer involves cytoreductive (debulking) surgery with the goal of removing or reducing tumor burden.⁷ Alternatively, women may receive neoadjuvant chemotherapy followed by interval debulking surgery, particularly if they have advanced disease or are not fit for primary surgery.^7,8 Prophylactic surgeries, which include the removal of fallopian tubes and/or ovaries, are practiced as a prevention strategy in women at high risk as studies have demonstrated a drastic reduction in the risk of ovarian, fallopian tube, and peritoneal cancers for women carrying BRCA mutations.^3,9,10 Following surgery, a pathologist performs a thorough macroscopic and microscopic examination of the excised tissues. Even the most stringent histopathologic protocol (sectioning and extensively examining the fimbriated end) only samples 0.5% of the fallopian tubes. In this context, submillimeter pathologies can remain undetected. Moreover, surgical procedures and management can vary across hospitals, and the careful dissection of the entire length of the fallopian tube is not universally conducted. Imaging tools to guide histopathologic examination could reduce the risk of undetected suspicious or cancerous lesions. Imaging procedures have also been used to examine disease burden and include US, computed tomography, positron emission tomography, and magnetic resonance imaging. Although these techniques allow for deep tissue visualization, the resolution of these techniques is not suited for the early detection of small tumors and lesions,^2,11 both at the screening and diagnostic levels.

At the molecular and histological level, ovarian cancer is a heterogeneous disease and is clinically diverse, with the majority of cases classified as epithelial ovarian cancer (EOC). There are five main subtypes of EOC that differ in cell origins, pathogenesis, and pathology: mucinous, endometrioid, clear cell, low-grade serous (LGS), and high-grade serous (HGS).³ HGS ovarian cancer is the most common and most lethal subtype of EOCs.³ Although the ovarian surface epithelium was historically believed to be the etiological site for HGS carcinomas, evidence now indicates that precursor lesions originate in the fallopian tubes as serous tubal intraepithelial carcinomas (STICs).^3,5 While a significant proportion of HGS ovarian cancers arise from STIC lesions that develop in fallopian tubes, these lesions measure only a few hundred micrometers, escaping detection even on rigorous pathological evaluations. HGS ovarian tumors arise from the proximal end of the fallopian tubes, spread to the ovaries, and, in more advanced disease, metastasize to the abdominal cavity.^12,13 A dedicated detection instrument must consider the specific characteristics of ovarian cancer, such as the large area covered by the organs of interest (fallopian tubes and ovaries), the heterogeneity of the tumors, and the high metastatic potential of epithelial lesions. There is a critical need for an efficient tool allowing for internal visualization of suspicious lesions or morphological changes that indicate early stage of disease in the fallopian tube.

Optical imaging techniques are good candidates to address these requirements. They can be implemented through endoscopic probes small enough to be inserted inside the uterus and fallopian tubes. They also allow for fast imaging, with the ability to perform optical biopsies without damaging the tissue. Among morphological imaging techniques, confocal imaging has been used to image ovaries in vivo^14–16 and in fallopian tubes¹⁷ using contrast agents such as acridine orange. Multiphoton imaging, which does not require a contrast agent, has been used on ovaries to find optical biomarkers.^18,19 On the other hand, molecular imaging techniques can provide more specific information, regardless of tissue structure. Fluorescence imaging of ovaries and fallopian tubes has been used intraoperatively using fluorescein-binded folate-alpha.²⁰ Ex vivo fluorescence imaging was also performed using the natural autofluorescence of ovaries^21–23 and fallopian tubes.^24,25 Raman spectroscopy is another molecular imaging technique that can distinguish normal from malignant ovaries.^26,27

Optical coherence tomography (OCT) is a laser-based interferometric technique that allows for transverse visualization of tissue without any fixation, slicing, or staining.²⁸ This imaging technique identifies morphological features of the sample and can yield qualitative and quantitative information on tissue architecture. State-of-the-art OCT implementation allows for fast imaging speed^29,30 and is robust enough for use in clinical settings.^30–33 OCT is now performed as a routine clinical evaluation of the retina in ophthalmology^33–35 and has been used for Barrett’s oesophagus detection in gastroenterology^31,36,37 and in cardiology applications for vulnerable plaque detection.^38,39

OCT has also been used in gynecology for specific applications, such as in vivo inspection of the cervix⁴⁰ and endometrial tissue in both ex vivo^40,41 and in vivo studies.⁴² Ovaries and fallopian tubes have also been investigated in vitro using bench-top systems.^43,44 A laparoscopic OCT probe was built to assess in vivo the condition of the isthmus, the smallest part of the fallopian tube,⁴⁵ and ovaries.^46,47 The design and the validation of an endoscopic OCT probe is primarily application driven. Recently, a promising flexible probe for fallopian tube imaging has been demonstrated, combining OCT with reflectance imaging.⁴⁸ Scanning mechanisms can also be included into the imaging probe to enhance the field of view including rotary junctions,^37,49 mounted mirrors,^30,46 and motorized pullback.^46,48,50 The design of very small probes (outside diameter of 1 mm or less) has been demonstrated in some of these implementations.^30,48,51

Recent progress in optical imaging, as previously highlighted, can be beneficial to ovarian cancer screening and diagnostics. Rigorous inspection of fallopian tubes with an instrument capable of identifying submillimeter suspicious lesions can effectively reduce the burden of extensive histopathologic examination protocols. In a long-term perspective, a similar tool could be envisioned for direct in vivo screening, adding flexibility to management of high-risk patients.

In this study, we investigate the potential of OCT as a morphological imaging technique for qualitative and exhaustive analysis of the mucosal layer of fallopian tubes for diagnostic purposes. Using a bench-top OCT system, we present a preliminary comparison of OCT images of human ovaries and fallopian tubes with corresponding histology. We analyze the five most frequent subtypes of ovarian cancer, which may involve the fallopian tube at some point in the cancer’s progression. We also investigate the potential of OCT for intraluminal imaging of human fallopian tubes. The OCT system was coupled to an all-fiber endoscopic probe, specifically designed for direct insertion into the fallopian tube and was capable of imaging epithelial and subepithelial structures over 10 cm in length. We present the design as well as the optical and mechanical characterization of this probe and demonstrate its application for OCT imaging of intact, freshly excised human fallopian tubes from five patients.

2.

Methodology

2.1.

Optical Coherence Tomography System

The OCT system (Fig. 1) is a wavelength-swept micro-electro-mechanical system vertical-cavity surface-emitting laser (MEMS VCSEL; Thorlabs, New Jersey) centered at 1300 nm and coupled to an interferometer and a data acquisition module ($500\mathrm{MS}/\mathrm{s}$; ATS9350, Alazar Technologies, Canada). The swept-source has an average output power of 35 mW with an A-line scan rate at 100 kHz. The first of two configurations [Fig. 1(a)] involved a conventional bench-top commercial version, using a Michelson interferometer and a bulk imaging head. The scan lens is an LSM03 $5\times$ objective (Thorlabs, New Jersey), with an effective focal length of 36 mm. The imaging head of the bench-top configuration yields a lateral resolution of $25\mu \mathrm{m}$ and depth resolution of $15\mu \mathrm{m}$ in air. The field of view covers a volume of $10\mathrm{mm}\times 10\mathrm{mm}\times 5\mathrm{mm}$. Data acquisition of a volume takes 17 s ($10\mathrm{mm}\times 10\mathrm{mm}\times 5\mathrm{mm}$, $1024\times 1024\times 512\text{voxels}$). The second configuration [Fig. 1(b)] uses the same laser and data acquisition module but uses an endoscopic probe in a common-path interferometer with an AC-coupled dual-balanced detector (PDB410C-AC, Thorlabs, New Jersey). The probe yields resolutions of $23\mu \mathrm{m}\times 25\mu \mathrm{m}\times 15\mu \mathrm{m}$ ($x\times y\times z$).

Fig. 1

System configurations for the acquisition of fallopian and ovarian tissue images. Schematic diagram of the commercial bench-top version is presented in (a), with the addition of a custom external reference arm. The probe configuration is presented in (b). The probe images using common-path interferometry, using the fiber tip as the reference surface. Additional details on the construction of the probe are presented in Fig. 4. C, circulator; G, galvanometer-mounted mirror; DB, dual-balanced detector; Det, AC-coupled detector; SMF, single-mode fiber; NCF, no core fiber; and GRIN, graded-index fiber.

3.

Results

3.1.

Paraffin-Embedded Specimen Imaging

In the bench-top OCT configuration, six types of specimens representing EOC pathologies and normal tissue were selected and imaged. A total of 52 samples were scanned resulting in 180 OCT volumes. Representative en face images with corresponding histology are shown in Fig. 1 with a single image presented per type of pathology. Figures 2(a) and 2(b) show transverse images of a healthy fallopian tube by OCT and histology (H&E staining), which revealed multiple structures such as the tubal lumen (Lu) and the surrounding muscle wall (M), as well as a cluster of blood vessels (V) outside the fallopian tube. Figure 2(c) shows an OCT cross-sectional image of an LGS tumor with corresponding histology [Fig. 2(d)], and demonstrated stroma (S) in distinct contrast to regions of papillary architecture (PA). Pathology of an endometrioid tumor was confirmed by OCT and histology [Figs. 2(e) and 2(f)] showing that the sample was mainly constituted of glandular growth (G), very similar to glandular endometrial tumors. For an HGS tumor, both OCT and histology [Figs. 2(g) and 2(h)] showed a carcinoma with solid architecture, which appeared dense in comparison to the stroma due to the higher number of nuclei (purple regions in the histology translated into brighter regions on OCT). In opposition, the histological section of a clear cell tumor shown in Fig. 2(j) had unstained cytoplasm (CC) and presented a tubulo-cystic architecture, whereas the corresponding OCT section [Fig. 2(i)] appeared more heterogeneous. Finally, Figs. 2(k) and 2(l) present the OCT image and histology, respectively, of a part of a mucinous tumor and show that the stratified columnar epithelium (Ep) lining was distinct from the ovarian stroma (S).

Fig. 2

En face OCT images of paraffin-embedded human EOCs or normal specimens paired with the corresponding histology (H&E) section. All OCT images are represented by (a, c, e, g, i, and k). All H&E sections are represented by (b, d, f, h, j, and l). (a) and (b) show healthy fallopian tube in transverse sections, (c) and (d) show LGS tumor, (e) and (f) show endometrioid tumor, (g) and (h) show HGS tumor, (i) and (j) show clear cell tumor, and (k) and (l) show mucinous tumor. Punch holes in the tissue are marked with a circle (o). Lu, lumen; m, muscle layer; V, blood vessels; pa, papillary architecture; S, stroma; G, glandular structure; In, stromal invasion; cc, clear cells; and Ep, epithelium. All OCT images are $10\mathrm{mm}\times 10\mathrm{mm}$ and scale bars are $500\mu \mathrm{m}$.

Our macroscopic comparisons showed good structural agreement between the histology and OCT images. We then examined samples on a microscopic level to determine if relevant diagnostic information could be identified. En face OCT images of HGS and LGS cancer samples at $5\times$ magnification are shown in Figs. 3(a) and 3(e), respectively, whereas corresponding histological sections are shown in Figs. 3(d) and 3(h). Insets represent $10\times$ magnifications of each tumor. Since the lateral resolution of the bench-top system was $25\mu \mathrm{m}$, details smaller than this limit were not visible. Carcinoma with solid architecture was revealed in the HGS tumor and marked with an asterisk (*), whereas PA was evident in the LGS tumor, marked with a cross (†). Healthy stroma, as indicated by arrows, appeared slightly darker than cancerous tissue.

Fig. 3

En face OCT images of HGS and LGS ovarian tumors at $5\times$ magnification (a, e) and $10\times$ magnification (b, f) with the corresponding histology (H&E). (a) OCT segment of HGS ovarian tumor with magnification of a cancerous region (b). The histology slice is shown in (d) with magnification (c), corresponding to (a) and (b), respectively. This type of tumor has carcinomas with solid architecture (*) around the stroma (arrow). (e) OCT segment of an LGS ovarian tumor with an enlargement of the cancerous zone shown in (f). The corresponding histology and magnification are shown, respectively, in (h) and (g). LGS tumors present a PA (crosses) around the stroma (arrow). Holes (circles o) represent sections punctured from the sample for external analyses. Scale bar: $500\mu \mathrm{m}$.

3.2.

Custom-Built Optical Coherence Tomography Probe for Intraluminal Imaging

To examine the potential of OCT for intraluminal imaging of freshly excised human fallopian tubes, we designed a custom-built imaging probe to use as a replacement of the bulk scanning head of the OCT system previously described. To withstand insertion into the fallopian tube without assistance of an outer canule, the probe was designed as a rigid cylinder of 10 cm in length. The probe wall consisted of hard glass tubing and the tip was capped with optical glue to prevent perforation and minimize tissue trauma during insertion. Within the glass tubing, the optical fiber was covered with a protective hytrel jacket (extending to the fiber connection) to prevent mechanical damage during manipulation. Figure 4 presents the main features and characteristics of this endoscopic probe. The final outer diameter was 1.2 mm and allowed for both contact and noncontact imaging.

Fig. 4

Descriptive diagrams and photographs of the probe. (a) Schematic diagram of the assembly in the tip. (b) Photographs of the complete probe. (c) Dimensions of the three fibers used as focusing optics. (d) Photograph and a magnified view of the gold-coated mirror with a Canadian dime for size comparison. (e) Full length of the rigid part of the probe with an inset showing a closer view of the tip.

All the focusing optics were directly implemented into the optical fiber. The probe consisted of 2 m of SMF-28 fiber (Corning, New York), a spacer of no core fiber (NCF, custom-drawn pure silica fiber, COPL, Laval University, Canada), and a graded-index fiber (GRIN—GIF625, Thorlabs, New Jersey). The SMF-28 has a core of $8.2\mu \mathrm{m}$, an outer diameter of $125\mu \mathrm{m}$ and a numerical aperture (NA) of 0.12. The NCF is a highly multimode fiber with a guiding coating jacket with a diameter of $125\mu \mathrm{m}$ and a 0.5 NA. The GRIN fiber is a graded-index and bend-tolerant multimode fiber with an inner cladding of $62.5\mu \mathrm{m}$, an outer cladding of $125\mu \mathrm{m}$, and a 0.275 NA. The length of the spacer ($180\mu \mathrm{m}$) and the GRIN ($300\mu \mathrm{m}$) were selected through simulations to provide optimal imaging at a focal distance of 2 mm away from the outside wall of the probe with a simulated diffraction limit of $<60\mu \mathrm{m}$ over the imaging range of 2 mm. To provide a side view, an angled-polished rod mirror (#47-628, Edmund Optics, New Jersey) was mounted at the tip of the probe within the glass tubing. A gold coating was applied to the mirrors to increase the reflectivity. The coating was a three-layer structure of high reflectivity and of high surface quality that is composed of 30 nm of chromium, 200 nm of gold, and a protective layer of 20 nm of pure silica. With this modification, the probe’s transmission of near infrared light (1250 to 1350 nm) was improved from 65% to 95%.

Figure 5 presents simulations of the main optical characteristics of the probe using optics software for layout and optimization (OSLO, Lambda Research, Massachusetts). The spot diagram [Fig. 5(a)] of the output beam was computed for five distances away from the output window, ranging from 0 to 2 mm away from the probe’s external wall, at 0.5 mm increments. An Airy disk of $43\mu \mathrm{m}$, drawn in black on each sketch, shows that all the paraxial rays are included within the diffraction limit. Figure 5(b) represents the simulated point spread function for each axis and shows a relatively small presence of astigmatism. Coddington sags factors obtained from the simulations indicated a nearly collimated beam in both axes $(x,y)$, with a residual focal length of $-37\mathrm{mm}$ in the $x$-plane and $-2\mathrm{mm}$ in the $y$-plane. The schematic diagram of the simulation is shown in Fig. 5(c) and represents the beam propagation ($z$-axis) corresponding to the imaging direction (A-line). The probe has an effective focal length of $256\mu \mathrm{m}$ and an NA of 0.024. The chromatic focal shift over the spectrum of the OCT source is $10\mu \mathrm{m}$ between 1250 and 1350 nm. The inset shows an image of a human finger acquired with the probe.

Fig. 5

Optical specifications of the probe. (a) Spot diagram computed with OSLO for five distances from the probe’s outer surface to 2 mm away from the probe. (b) Point spread function of the beam in the $x$ and $y$ planes at the plane of best focus. (c) 2-D layout of the simulated probe, as well as the coordinate system. A-lines are acquired in the $z$ direction and the inset outlined in the blue box presents an image of a human finger acquired with the probe. For (b) and (c), units are in mm.

The three-fibered parts of the probe were sequentially spliced together using a filament fusion splicing system (FFS-2000, Vytran Corporation, New Jersey). Each piece of NCF and GRIN was carefully spliced and cleaved to the right length, then measured and inspected with the splicer’s visualization camera. This part of the probe was then inserted into three clean, clear fused quartz capillaries (#CV2033, CV4055, and CV6084, VitroCom, New Jersey) that were held together with optical glue. This encapsulated fiber probe was inserted into the clean, clear fused quartz capillary (#CV1012, VitroCom, New Jersey) and placed close to the gold-coated mirror at the tip of the final assembly. When an optimal placement of the fiber tip and the mirror was reached, components were glued together with optical glue. The tip of the probe was rounded off and polished to minimize tissue damage during insertion. A protective $900\mu \mathrm{m}$ hytrel jacket was added on the fiber outside the tubing for additional mechanical protection.

After assembly, the probe was optically characterized to ensure its performance met the design specifications. Measured lateral resolutions were 23 and $25\mu \mathrm{m}$ in $x$ and $y$ planes, respectively. These values corresponded to edge responses acquired using a USAF1951 resolution target and compared well with the simulated resolutions in air of 22 and $26\mu \mathrm{m}$, respectively. Axial resolution ($z$) was $15\mu \mathrm{m}$ in air. Imaging depth in diffusive soft tissue samples, such as human finger and uterus, was $\sim 2\mathrm{mm}$.

3.3.

Imaging of Freshly Excised Fallopian Tubes

Reproductive tissues and specimens were obtained from five patients undergoing prophylactic surgery and were imaged within 30 min before pathological examination. Human fallopian tubes were imaged using the custom endoscopic probe. Images of uteri, ovaries, and fallopian tubes were also acquired using the bench-top configuration. All imaging sessions included a macroscopic examination of the sample. The bench top was then used to image the exterior part of the fimbriated ends of the tube. The probe was inserted into one fallopian tube from the proximal end (close to the ovary) toward the uterus. Complete insertion time ranged from 20 s to 1 min, depending on the tube’s condition and length. Probe insertion up to the isthmic part in both fallopian tubes was successful in all five patients’ samples. After insertion, a pullback was performed to acquire a 2-D image over the entire length of the tube. The process was then repeated at other angles to collect four scans per fallopian tube. The last centimeter of the isthmus and uterotubal junction was not acquired due to size constraints. Regions of interest were marked with small surgical tacks in the peritoneum surrounding the tubes for pathological examination.

Photographs and images of a freshly excised specimen with fallopian tubes and probe insertion are presented in Fig. 6. A transverse histology of the fallopian tube after four insertions is presented in Fig. 6(d) with the insertion site noted by the (*) as in Figs. 6(a) and 6(b). An OCT image of a pullback covering 10 cm in length is presented in Fig. 6(e) with magnifications of the ampulla and fimbriated end of the tube [Figs. 6(f) and 6(g)]. The mucosal layer of the fallopian tubes (L), in contact with the probe, appears in the center region of Fig. 6(e). The peritoneum (P) and muscle (M) form the external layers of the tube and are located in the exterior of the image. The pullback image was acquired from left to right on the image, beginning from the isthmus to the fimbriated end of the tube. The OCT image revealed the isthmic part as very thin with a small muscle layer, and the outer layer of peritoneum was visible as a brighter lining on the bottom. Microvilli were compacted around the probe, creating lines that were brighter than the surrounding muscle (black arrowheads). From the isthmus, the tube transitions to the ampulla [Fig. 6(f)], represented as a conic structure that expanded the internal diameter from 2 to 4 mm. This section of the ampulla had more microvilli that were less packed in the presence of the probe, and a slight increase in muscle thickness was observed. The infundibulum shown in Fig. 6(g) had a lumen that widened into a diameter of 8 mm toward the end. In this case, the fallopian tube was manually held by the end of the infundibulum during the insertion and pullback (${\mathrm{D}}_1$ and ${\mathrm{D}}_2$ represent fingers holding the tube), to preserve the fragile fimbriated end. The probe was closer to the muscle wall in this section of the pullback, explaining the apparent restriction in tube diameter. The right end of the image in Fig. 6(e) represents the fimbriated end, imaged from the internal lumen, which correlated well with the histology and external imaging results using the bench top (results not shown). Small dark inclusions (white arrows) were visible in the fimbriated end and arose from edema that leads to lymph accumulations. These structures differ from blood vessels (white star), since blood has a greater effect at attenuating the OCT signal.

Fig. 6

OCT pullback of a freshly excised human fallopian tube. (a) Photograph of the patient’s specimen and (b) photograph taken during insertion of the probe in the fallopian tube. (c) Drawing of corresponding fallopian tube illustrating the isthmus (Is), ampulla (Amp), infundibulum (In), and fimbriated end (Fim). (d) H&E slide of the tube in the region of the infundibulum. Repeated probe insertion at the site identified by (*) on images (a), (b), and (d) did not create damage. (e) Pullback on a length of 10 cm. Magnifications of OCT pullback show the ampulla (f) as well as infundibulum and fimbriated end (g), both of which are highlighted in green and red boxes, respectively, in (e). Layers of microvilli are pointed by the black arrowheads. White arrows point to edema regions, which differ from blood vessels (white star). FT, Fallopian tube; Ov, ovary; Ut, uterus; Cv, cervix; Lu, lumen; M, muscle; P, peritoneum; and D, finger. Scale bars: $500\mu \mathrm{m}$ (vertical) and 5 mm (horizontal).

4.

Discussion

The imaging capabilities of OCT offer an approach for the diagnosis and detection of ovarian cancer. This paper presents a preliminary comparison of OCT images of ovarian cancer specimens with histology, and validates the potential of OCT imaging to distinguish different ovarian pathologies. This study demonstrated a good correlation between OCT images at $5\times$ and $10\times$ magnifications and histology sections for six types of specimens representing healthy fallopian tubes and the most frequent ovarian tumors. To capture the heterogeneity of each subclass would have required a large cohort of hundreds of patients with ex vivo tissue imaging to obtain the same database size as with the tissue bank. The main limitation of these comparisons was the lateral resolution ($25\mu \mathrm{m}$ for OCT compared to 500 nm for histology), which did not allow for the assessment of fine cellular structures. Although OCT at 1300 nm cannot replace conventional pathological evaluation, it is an attractive modality to rapidly locate small benign or malignant conditions within fallopian tubes, and bridge the gap between macroscopic evaluation of the specimen as a whole and microscopic evaluation of histology slides. The various pathologies can be distinguished from one another with qualitative inspection of the OCT images throughout the acquired volumes, providing an initial assessment of the specimen before further microscopic evaluation. In addition, quantitative analysis of OCT data could potentially lead to discriminative metrics between healthy and pathological tissue requiring medical attention. Such classification can speed up the analysis of large areas (or large surface scans) of fallopian tubes, ovaries, or uterus.

The potential of OCT was extended to intraluminal imaging of freshly excised human fallopian tubes. For this purpose, we designed, assembled, and characterized a dedicated all-fiber endoscopic OCT probe and demonstrated that the measured performances of the probe agreed with the simulations and desired specifications. A minor astigmatism was measured in the air and was reduced when used in biological medium (due to the smaller refractive index mismatch between the glass capillary and the surrounding medium). When imaging surgically removed human fallopian tubes, the probe showed adequate imaging depth and multiple anatomical structures, such as the muscle layer, the mucosal layer, and the peritoneum. To our knowledge, this study yields the first OCT images of intact and freshly excised human fallopian tubes with minimal to no damage to the epithelial or subepithelial structures. It also provides the first images of the full tube, as the probe can be inserted over 10 cm in length in close contact with the mucosal layer. Imaging can be performed with minimal to no damage to the epithelial or subepithelial structures.

Notably, this probe could be adapted to in vivo use and screening purposes. This would require reducing the probe’s external diameter and rigid length. For in vivo screening, the probe would have to be inserted through the uterotubal junction instead of the fimbriated end. This junction is the smallest opening of the tube with $<1\mathrm{mm}$ in diameter. Flexibility is required as the tube is curved and at an angle with respect to the uterine cavity. During this study, we have noticed a variability in probe insertion easiness. Specific tubal curvature and lumen microvilli density vary for each patient. Older patients tend to have empty tubes, whereas younger patients have higher microvilli density. Navigating this convoluted structure can be difficult; a steering mechanism at the tip of the probe would help with managing the natural tubal tortuosity. In addition, the tip of the probe plays a pivotal role in the insertion process, since a poorly designed tip may be too rough on the microvilli. We found that an elliptical structure ensures a smoother insertion than a spherical shape, since it seems to displace the microvilli more gently before the passage of the probe’s main body.

The design of our all-fiber endoscopic probe demonstrates promising potential as an imaging tool as it is compatible with a comprehensive imaging technology coupled with the addition of a radial scanning mechanism. This can yield a 3-D virtual fallopian tube that can be cut in any direction, under any desired/selected projection without altering the specimen. With adequate sampling, the data set of a whole fallopian tube can be represented by 20G voxels ($1024\times 2000\times \mathrm{10,000}\text{voxels}$). Once acquired, the data could be displayed according to any transverse axis for the pathologist to review and analyze. In addition, quantitative analysis of OCT data could potentially lead to discriminative metrics between healthy and pathological tissue requiring medical attention. Such classification can speed up the analysis of large areas (or large surface scans) of fallopian tubes, ovaries, or uterus. Finally, the probe can be easily adapted to include fluorescence²⁴ and/or spectroscopy measurements simultaneous to the OCT imaging^32,49,51 by adding double-clad fiber to the probe and a double-clad fiber coupler to efficiently separate the signals.^51,53,54

5.

Conclusion

This paper presents, to the best of our knowledge, the first OCT images of intact human fallopian tubes. The fast acquisition rate, the in-depth imaging without altering tissue, and the volumetric image acquisition showcase OCT imaging as an efficient tool for ex vivo visualization of fallopian tubes and for diagnostic purposes. Notably, this imaging technique allows the scanning of a large area of the human oviducts without damaging the specimen and can be performed both on a paraffin-embedded specimen and on fresh samples. Future work includes imaging of the inner fallopian tubes to detect benign and malignant tubal conditions in addition to comparative imaging with healthy tubes. This type of probe can also be used to image the inner cavity of the uterus through the cervix.