2.1.

Fiber Optic Lightguides

An optical fiber for spectroscopy consists of a core, a doped cladding, and a protective jacket. Light is transmitted based on the principle of total internal reflection. The half-angle (α) of the light cone that a fiber can accept is characterized by the numerical aperture (NA), which is defined by the difference in the refractive indices (n) of the core and the cladding material (Figure 2):

Figure 2

Fiber optic cable for spectroscopy consists of a core and a cladding with a lower refractive index and a rugged supportive jacket. Light is transported by total internal reflection. The light acceptance angle of the fiber (α) is defined by the refractive indices of the media, core, and cladding (n _media , n _core , n _cladding ) .

For transmission in the visible wavelength range, the optical fiber core is made out of glass or plastic (e.g., acrylic or polystyrene). The doped cladding is usually made of a similar material but with a lower refractive index. Since light is not propagating in the cladding, losses due to absorption are less important. As soon as background signals introduced by the fiber itself become critical for the application, or the wavelength range must be extended to the ultraviolet (UV) and infrared (IR), high-grade fused silica are used for the core material. All-silica fibers have a doped silica cladding, while plastic clad fibers have a silicone cladding. Optimized fiber preform manufacturing^{7 8 9 10 11} enables transmittance from 200 nm (solarization-resistant-UV grade fiber, Polymicro Technologies Inc., Phoenix, Arizona, or Ceramoptec, East Longmeadow, Massachusetts) up to 2500 nm (low-hydroxyl fiber) and sapphire fibers (Saphikon, Milford, New Hampshire) extend the transmission in the IR above 3000 nm. This enables the application of fiber optic probes for UV resonance Raman (UVRR) and IR Raman spectroscopies.¹² Due to bending of the fibers and defects in the fibers causing scattering, light may exit the core and hit the jacket. Most plastic jackets, such as Nylon and Tefzel®, produce significant autofluorescence when irradiated with UV light. Polyimide- and metal-coated fibers, such as gold and aluminum, exhibit minimal fluorescence. During intense UV irradiation, defects may form in quartz fibers that are accompanied by autofluorescence in a broad band at¹³ both 450 and 650 nm. The formation of these defects is partly reversible through diffusion processes occurring in the quartz glass. It is further known that silica produces intrinsic Raman signals at near-IR (NIR) excitation which interfere with in vivo Raman spectroscopy. If background signals in spectroscopic measurements are present, the dynamic range of the detector can become critical.¹⁴ For spectroscopic applications, multimode fibers with core diameters of 50 to 600 μm are usually used. Short pieces (1 to 2 m) of fibers with larger diameters can be produced in custom runs. The bending radius of quartz fibers in which no long-term defects result is approximately 100 times the fiber diameter and the momentary bending radius is approximately 50 of the long-term bending radius.

2.2.

Fiber Packing

To manufacture flexible fiber optic cables with a large optically active area, fibers with diameters of 100 to 200 μm are packed into bundles. The amount of fibers (n _fiber ) that can be packed into a round cross section is illustrated in Figure 3. If fibers are packed hexagonally, according to ring a in Figure 3, the number of fibers is determined with

Figure 3

Hexagonal packing of fibers into the smallest possible cross section.

2.3.

Beveled versus Flat Exit Surface

To ensure optimal coupling, the end of the fiber optic cable is cleaved or polished. If the exit surface is polished with an oblique angle with respect to the fiber axis, the output will be deflected [Figures 4(a) to 4(c)] (beveled fibers). If the critical angle for total internal reflection is reached, the light will leave the fiber through the cylindrical side^{15 16 17} [Figures 4(d) and 4(e)]. The critical angle [Figure 4(d)] for the silica-air interface is 43.3 deg and for the silica-water interface is 66 deg. A fiber with a combination of a beveled and a flat polished tip¹⁸ (Gaser Technology, Visionex Inc., Georgia; product no longer available) can also be used for deflection [Figure 4(f)] as one part of the beam exits in the direction of the fiber axis while another part is guided sideways. To enable a larger range of steering angle, the beveled surface of the fibers can alternatively be coated (e.g., aluminum).

Figure 4

(a) Output of an optical fiber is described by the opening half angle α or the NA; (b) and (c) oblique polishing of the fiber tip deflects the output beam; (d) when total internal reflection is achieved for a part of the light, the output will be split into two parts, one exiting in the axial direction and the other in the radial direction; (e) when all light is totally internally reflected at the fiber end surface or the beveled end surface is reflectively coated, light leaves the fiber through the side wall; (f) a fiber based on a combination of a beveled and flat polished surface has a partially deflected output beam; (g) a fiber optic probe that illuminates and collects through the same fiber has the highest collection efficiency; (h) a dual-fiber bundle is used for many spectroscopy application, and this concept can be easily extended to hexagonal packed fiber bundles; (i) a dual-fiber probe based on beveled fibers achieves a better collection efficiency compared to the flat tip dual fiber configuration, while collection and illumination channels are still separated; (j) fiber probes based on a fiber with a combination of a flat tip and a beveled surface have excellent collection efficiencies; and two probe designs for submersion (k) and measurements on surfaces (l). Here a central fiber (labeled a) illuminates the sampling volume or the surface and six surrounding fibers (labeled b) collect the emitted light. Collection and illumination channels are interchangeable. Housing with a thin shield (quartz, sapphire) enables a constant sampling distance for surface measurements.⁶

Beveled fiber applications for probe designs have been theoretically and experimentally analyzed by Cooney et al.;^{19 20} The sensitivity and sampling volume of beveled fibers was compared to other designs, such as single fibers [Figure 4(g)] flat tipped probes [Figure 4(h)], and probes with lenses. The sensitivity and sampling volume were measured in a clear medium and results were reported for the collection of Raman-scattered light and are also applicable for fluorescence collection. A probe consisting of two beveled fibers [Figure 4(i)] is at least by a factor of 1.5 more efficient than a dual fiber with a flat tip probe (NA=0.22) and the sampling volume is smaller and located closer to the probe tip. A single fiber probe [Figure 4(g)] is 1.8 to 4 times more efficient, depending on fiber diameter, than a dual fiber with a flat tip probe [Figure 4(h)].

The efficiency of a “Gaser-type” collection fiber [Figure 4(f), 300-μm core] combined with a flat tip illumination fiber [Figure 4(j), 400-μm core] were measured by Shim et al.;¹⁸ with a sapphire Raman standard. Two configurations were considered. First, the light deflection angle of the “Gaser” fiber was between 13 and 32 deg (low deflection), and second, the deflection angle was between 35 and 55 deg (high deflection). Compared to a flat tip probe, Shim et al.; found that, in air, a low deflection probe measures a 4 times and the high deflection probe a 16 times increased signal. In water, the factors were 6.5 and 28, respectively. Similar values were found in 0.25 and 0.50 Intralipid™ solution. For a flat tip probe with a 400-μm illumination fiber and a 300-μm pickup fiber, maximal signal was picked up at a distance between 1.7 and 1.4 mm from the tip in water or in the Intralipid™ solution. The low deflection probe received the largest signal at a distance of approximately 0.5 mm, and the high deflection probe at 0.25 mm from the fiber tip in water or the Intralipid™ solution.

The concept of dual-fiber arrangement for separate illumination and collection can easily extend to a multifiber design, as shown in Figures 4(k) and 4(l). The sampling volume of a single excitation fiber overlaps with concentrically arranged collection fibers. Six collection fibers can be arranged around a single illumination fiber, which is equivalent to six dual-fiber arrangements. To construct the probe tip, the fiber bundle is glued together and polished to a cone shape by rotating the probe axially while polishing. This design can be optimized for immersed operation [Figure 4(k)] or for the interrogation of a surface [Figure 4(i)]. For surface measurements, the tip needs to be enclosed and a quartz or sapphire window placed at the distal end [Figure 4(i)]. In general, collection and illumination channels are exchangeable.

2.4.

Material Choice and Safety Considerations

All fiber optic probes in this paper are reviewed and presented with a biomedical application in mind. The probe tip is either at a close distance to or in contact with the tissue surface or body fluids and, therefore, should be analyzed for potential hazards. An analysis of potential risks and protection against those risks is a requirement of human subject studies funded by the National Institutes of Health (NIH). A guidance document, which is also helpful for the evaluation of hazards of fiber optic probes, was developed by the U.S. Food and Drug Administration (FDA) for electro-optical detection of cervical cancer.²¹ The list of potential adverse events that are applicable to the use of fiber optic probes are optical radiation hazards, thermal hazards, electrical shock hazards, clinical hazards (transmission of diseases), and material toxicity hazards. According to a general evaluation of medical devices,²² fiber optic probes are most likely categorized as transient surface or transient external communicating devices and their materials should be tested for cytotoxicity, sensitization, and irritation, and depending on the application, for acute systemic toxicity. If the fiber optic probe is made of materials that have been well characterized in published literature and have a history of safe use, there is adequate justification to not conduct some or all of the suggested tests.

Optical threshold limit values (TLV) or maximum permissible exposure (MPE) have been established to assist in the control of health hazards^{23 24} by the American Conference of Governmental Hygienists (ACGIH) and the American National Standard Institute (ANSI). The biologically effective radiation (device emission weighted by the action spectrum, which is normalized at 270 nm) should not exceed the TLV for skin and eye (which is 3 mJ/cm ² at 270 nm for non-laser-based devices). The TLV for skin and eye has also been recommended for the cervix. Potential temperature increases should be evaluated using endpoints or by potential peak tissue temperature. Since materials (fused silica, acrylic, polystyrene, or silicone) used for fiber manufacturing are good electric isolators, a fiber optic probe with minimal shock hazard can be manufactured.

To avoid clinical hazards, such as the transmission of diseases, the fiber optic probe needs to be disinfected or sterilized prior to its use with common clinical practice. If the fiber optic probe cannot be detached from the spectroscopic equipment, the parts which may be in contact with tissue need to be soaked; for example, in a solution based on 2.4 gluteraldehyde (CIDEX, Advanced Sterilization Products, Irvine, California) or 0.55 orthophthalaldehyde solution (CIDEX OPA). If a fiber optic probe is used during a surgical procedure, it can be covered with sterile drapes developed to cover ultrasound probes. Low-temperature sterilization with ethylene oxide or hydrogen peroxide gas can be applied to a detachable probe. Most critical is the compatibility of the probe materials with the disinfectant. CIDEX OPA is a disinfectant that is compatible with most materials used to manufacture fiber optic probes, including adhesives cyanoacrylate (Super Glue) or Epotek 353 and Epotek 301 (Epoxy Technology, Billerica, Massachusetts).

A variety of materials that can be used to enclose the fiber optic probe have been well characterized and their biocompatibility has been previously published in the literature. Safe choices are materials created for implants or materials that were tested for U.S. Pharmacopoeia (USP), class VI. USP, class VI, is a base requirement for medical device manufacturers. A summary of materials that could be used to create fiber optic probes are listed in Table 1. One of the first alloys created for human use was stainless steel type 302 and the corrosion resistance improved²⁵ stainless steel type 316 (hypodermic steel). A large variety of standard tubing diameters and wall thicknesses are available to enclose fiber optic cables and optical elements. Aluminum oxide (Al ₂ O ₃) is an inert bioceramic and, when grown to a crystal (sapphire), is chemically inert and almost insoluble. Sapphire has a high thermal conductivity and is optically transparent between 200 nm and 3 μm. Epotek produces a variety of glues with excellent transmission (301-2 and 307) and with low autofluorescence (301-2FL) that can be used to bond optical elements within the fiber optic probe. Several glues from Epotek were tested for USP, class VI, and are autoclavable (e.g., 353 and 375). Many thermoplastic polymers have been used in the body, for example, polyethylene (PE), polytetrafluoroethylene (PTFE, Teflon), polymethylmethacrylate (PMMA), and polyester (PET). These materials can be used to create flexible and heat shrink tubing or plastic enclosures for fiber optic probes. High-temperature-resistant polymers include polyimide (220 °C), PTFE (230 °C), and silicone rubber (200 °C).

Table 1

3.1.

Diffuse Reflectance Probes

Several recent studies have suggested that differences in the optical properties assessed using diffuse reflectance spectroscopy can be used to discriminate normal and abnormal human tissues in vivo in the urinary bladder,^{33 34} pancreas,³⁵ esophagus,^{36 37} colon,^{38 39} ovaries,⁴⁰ breast,⁴¹ and the skin,⁴² using simple fiber optic probes.

To extract absorption and scattering properties from reflectance measurements, spatially resolved reflectance intensities must be obtained. These data are then fitted to an analytic expression, which is based on diffusion theory with the assumption that the sample is homogenous and semi-infinite.^{29 32 43 44} When this approximation is compared with Monte-Carlo-based photon propagation simulations,^{45 46 47} it can shown that at source detector separations smaller than 1/μ_s (Ref. 43) to 1.5/μ_s (Ref. 44), significant deviations occur.

Therefore, optical properties are normally derived from reflectance profiles measured at six to nine source detector separated locations. For tissue, the separation distances are in the range of 2 to 20 mm. The fiber optic probes used for these studies consist of a single excitation source and several spatially distributed collection fibers [Figures 5(a), 5(b), and 5(c)]. A probe with linear alignment was constructed by Wang and Jacques⁴⁸ and Lin et al.;⁴⁹ and places the detection fibers over a range of 1 to 10 MFP [Figure 5(a)]. Bays et al.;⁵⁰ used a scanning approach with a side-deflected illumination fiber and collection fiber that can be translated axially [Figure 5(b)]. A black mask was printed on the measurement window and defined measurement locations. The scanning range was 27 mm. Alternatively, a circular fiber arrangement [Figure 5(c)] by Nichols et al.;⁵¹ enables a simple calibration of the system by placing a source fiber in the center of all fibers. Measuring the spectrally resolved reflectance of all fibers simultaneously requires a dynamic range of four orders of magnitude. Neutral density filters in the fiber optic path reduce the required dynamic range of the detector. Our approach to measure reflectance with a source-detector separation from 200 μm to 3 mm requires three different integration time settings (150 ms and 2 and 9 s), sufficient spacing in between the fibers on the CCD detector to prevent blooming, and an increased number of detection fibers for larger source-detector separations. Fiber tracks are electronically binned on the CCD to detect lower light levels.⁵² Lin et al.;⁴⁹ and Marquez and Wang⁵³ modified their linear probe design with an oblique incident source fiber and measured the shift of the reflectance profile. This removes the necessity of measuring in absolute units. All fiber optic probes that measure reflectance profiles in the diffusion regime have a diameter of approximately 2 cm, and the assumption is made that the optical properties do not vary over this range. However, the optical properties of currently investigated target areas (e.g., cervix, ovaries, and oral cavity) vary over orders of millimeters (lesion size), and the organ size is similar to the diameter of the proposed probes. Therefore, an accurate measurement of optical properties from small, premalignant lesions requires other approaches.

Figure 5

(a) Fiber optic probe for reflectometry. The light scattered from a single excitation fiber is detected by a linearly arranged array of collection fibers. Tilting the excitation fiber shifts the profile along the surface by an amount that is determined by the sample optical properties. The distance between the source and detector fibers must be adjusted for the expected intensity profile sampling.⁶ (b) Scanning fiber optic probe for reflectance measurements: label a, illumination fiber; label b, SELFOC lens; label c, deflecting prism; and label d, photomask on exit surface (with permission from Ref. 49). (c) Fiber optic probe with a circular fiber arrangement for reflectometry. A 2-cm probe head consists of a central calibration fiber, an illumination fiber, and non–equally spaced fibers. A variable neutral density filter adapts the light intensity in a transfer array to a smaller dynamic range (with permission from Ref. 51 and suggestions from M. Patterson and coworkers). Parts are a, illumination fiber; b, circular probe head; c, calibration fiber; d, detection fibers; e and f, linear transfer array; g, neutral density filter; and h, a spectrograph input ferrule. (d) Fiber optic probe for single point reflectance measurements. Hexagonal packed fibers are randomly assigned for collection (label a) and illumination (label b). An outer tubing (label c) combines the fiber bundle (label d) with a spacing wedge (label e). An absorbing coating (label f) based on carbon powder or black paint absorbs specular reflected light from the tissue glass interface. (e) Polarized reflectance system. White light is transported from the light source with a fiberoptic cable, collimated (label a) and passes a polarizer. Reflectance is collected (label b) with an imaging system. In the collimated beam path, reflectance is filtered with rotatable polarizer (label c) for parallel or cross polarization. (f) Frontal view of a fiber optic polarized reflectance probe. A simple probe consisting of one illumination channel (label a), two detection channels for parallel (label b), and cross-polarized (label c) light detection. (g) Frontal view of a fiber optic polarized reflectance probe. Ideally all four possible polarization combinations are measured. This can be achieved when the illumination fiber (label a) and two detector fibers (labels b and c) covered with the parallel polarization filter and two fibers (labels d and e) with the cross-polarized filter. Note that the intensity of channel (label c) will be lower than channel (label b). (h) Schematic of a backscattering event on a small particle and the intensities associated with various pick up geometries. The scattering probability is indicated by the mesh surface and the distance from the center (particle) to the surface is equivalent with the scattering probability in that direction. Incident light is polarized in vertical direction. Detectors are positioned at the same location as in (f). (i) Simple polarized reflectance probe consists of optical fibers (label a), a mechanical placeholder for the fibers (label b), the polarization filter (label c), a cover glass (label d), and a housing (label e). Two fibers are covered with the same polarization filter while the other fiber is covered with a 90 degrees rotated filter. (j) An assembled version of the probe described in (i). The probe tip consists of three 200-μm-diam fibers. The gap formed by the two perpendicularly oriented polarization filters can be identified in both pictures. (k) To measure at the same spot on a surface, a polarized reflectance probe could be constructed with a conical reflector. There is a large difference in angle of incidence and angle of collection with this design and the polarization state changes at the reflecting surface.

To overcome limitations of the diffusion theory, heuristic approaches, such as Monte-Carlo-trained neural networks^{32 54 55} were investigated to obtain optical properties. Accurate numerical light propagation can be modeled outside the diffusion regime with various forms of multivariate analyses.⁵⁶ Also hybrid approaches^{57 58} were proposed that combine diffusion theory and Monte Carlo calculations to predict reflectance profiles over a wide range of source-detector separations. Recently, Dam et al.;⁵⁹ presented a probe design with six short source-detector separations between 0.6 and 7.8 mm. The probe measures at four different wavelengths. Using a multiple polynomial regression model, absorption and scattering could be predicted with an error of roughly 3 for absorption and 1.5 for scattering. Bevilacqua et al.;⁶⁰ and Bevilacqua and Depeursinge⁶¹ investigated short source-detector separated reflectance measurements and explained their sensitivity to the scattering phase function. Their probe diameter is 2.5 mm and incorporates eight linearly aligned 300-μm-diam fibers. They found that for source detector separations between 0.5/μ_s ^′ and 10/μ_s ^′, the reflectance profile not only depends on the average cosine of the scattering phase function, but also on the second momentum of the phase function. It is still difficult to determine absorption accurately if only reflectance profiles are available,⁶⁰ however, if the ratio of second and first momentam is known, absorption and scattering can be derived accurately from the reflectance intensity and the spatial slope of the reflectance intensity profile.

Mourant et al.;^{62 63} showed that for certain source-detector separations, the path length of the collected photons does not depend on scattering parameters. For experimentally verified μ_a of 0 to 0.86 cm ⁻¹, a source-detector separation of approximately 1.75 mm is mainly sensitive to changes in absorption. Scattering parameters for this investigation were in the tissue-relevant range for visible light (7.5<μ_a ^′<15 cm ⁻¹ and 0.8<g<0.95) . The NA (0.2 or 0.34) of the fibers did not affect these findings. Drezek et al.;⁶⁴ found that changes in neoplastic cells mainly affect high scattering angles, which can best be detected with a source-detector separation smaller than 500 μm and an NA of the fibers smaller than 0.4.

Both scattering and absorption properties of tissue can contain diagnostic information relevant to tissue pathology. Variations in scattering are due to inhomogeneities in the refractive index; recent results^{64 65} have shown that tissue backscattering is altered as the size of the nucleus increases and the nuclear texture becomes coarser. In case of spherical scatterers, a periodic structure in the reflectance spectrum is characteristic and depends on relative scattering size and collection and illumination geometry.^{37 66} This periodic structure can be theoretically derived by evaluating the scattering phase function at the angle corresponding with the angle between illumination and collection direction for every wavelength.

A probe to collect data which measures reflectance at a single point is shown in Figure 5(d). A wedge is put in front of a fiber bundle to reduce specular reflection in the detection path.⁶⁷ Reflected light from the glass-tissue interface is absorbed on the sides of the wedge (black paint or glue mixed with carbon powder). If the angle of the wedge is larger then the acceptance angle of the fiber (e.g., >9 deg for an NA of 0.22), light that is specularly reflected back onto the fiber bundle will reach it with an angle that will not be transported to the detector.

3.2.

Polarization Reflectometry Probes

Polarization techniques have been successfully used to gate detection depth in reflectometry. This is particularly interesting because the origin of epithelial neoplastic changes is within the first 100 to 300 μm of tissue and neovascularization originates in deeper tissue layers. Reflectance measured in a parallel-polarized fashion contains mainly light scattered from upper tissue layers, since light from the deeper tissue layers is multiply scattered, which randomizes the polarization status. Light loses its original polarization status after approximately 20 scattering MFP lengths (1/μ_s) and becomes unpolarized in the diffusing regime.^{68 69} Cross-polarized reflectance originates from deeper tissue layers since the polarized illumination light must undergo several scattering events until significant components are created in perpendicular polarization direction. Subtracting the perpendicular polarized reflectance from the parallel polarized, removes 90 of light originating from deeper tissue layers.⁷⁰ Dividing this subtracted reflectance by the sum of the parallel and perpendicular polarized reflectance cancels common attenuation and the spectral characteristics of the light source and detector. Jacques et al.;⁷⁰ used this technique for polarization imaging to detect basal cell carcinoma in highly pigmented tissues. Syris Scientific, LLC (Gray, Maine) and Cytometrics, Inc. (Philadelphia, Pennsylvania) use an orthogonal polarized imaging technique to enhance signal from deeper tissue layers and to reduce surface specular reflections.⁷¹

Fiber optic probes that measure polarized reflectance consist of a polarization-filtered white light source and a polarization-filtered detection system with an optical multichannel analyzer [Figure 5(e)]. Since multimode fibers do not conserve the polarization status of light, the polarization filters must be placed at the fiber optic probe tip. Johnson and Mourant⁷² presented a system that measures reflected light at four positions through a linear polarizer. Linear polarizing laminated film (Edmund Scientific, Barrington, New Jersey) was placed in front of the fibers. Suppression of different polarization modes was in the order of 98 to 99. Ideally, in a polarization filtered probe, two or more pieces of polarization film are placed on the probe tip [as shown in Figures 5(f) and 5(g)]. It is expected that light from fibers b and c do not return the same intensities. Also fiber positions d and e will collect less light than locations b and c. This is illustrated in Figure 5(h) for the case of one small scattering particle. It can be deducted that parallel polarized location (label b) will exhibit the highest intensity. In this simplified model, location c should receive sin ² (θ) less light and locations d and e would require several other scattering events until the scattering plane is aligned with the polarizing filter. The light intensities detected in these fibers can be calculated with Monte Carlo simulations that consider the propagation of polarized light.^{73 74} The probe design from Johnson and Mourant⁷² incorporates a single polarization film covering all five fibers. For diagnostic purposes, the polarization ratio of light collected from fiber b and c was formed. This design is similar to that in Figure 5(g), with fibers d and e covered with the same polarization film as fibers b and c. The distance between the fibers is critical for such measurements, as described by Gurjar et al.;⁷⁵ and Backman et al.;⁷⁶ It is desirable to keep the angle between the axis of the illumination and collection path small. This prevents photons from entering the detection channel after one scattering event in a cross-polarized setup. If the angle is larger, after one scattering event, the projection of the field vector can result in a component that passes the polarization filter. Small source-detector fiber separations are most sensitive to changes in the scattering phase function at high angles.⁶⁴ Sokolov et al.;⁷⁷ presents in vitro results with a system similar to that in Figure 5(e). Data from phantoms and cell suspensions was represented in a linear combination of forward and backward scattering components which were determined by the Mie theory. A fiber optic probe was designed based on this study [Figures 5(i) and 5(j)] and characterized in vivo.⁷⁸ A laminated polarization film was mounted with microscope guidance onto a fixture holding three 200-μm-diam fibers. Clear tape with adhesive on both sides was very useful in mounting the filters, because the tape is transparent and the filters could be visualized with a light microscope. A glass cover and rigid tubing encapsulate the construction. If the measurement and illumination spot must overlap on the sample surface, the shield thickness could be increased. An increased shield thickness results in specular reflections from the shields exit surface. Tilting the reflective surface, with respect to the optical axis, prevents reflected light from coupling back into the fibers. Alternatively, an optical lens can be placed in front of the probe similar to Figure 14(a) in Sec. 8.1, or a conical reflector [Figure 5(k)] could align the beam path. If reflectors are used to align the illumination and measurement path, linear polarized light will be converted into elliptical polarized light if the light is not purely linear polarized. This will make construction with mirrors challenging.

4.1.

Single-Pixel Measurements

The classic fiber optic probe to measure fluorescence consists of at least one excitation and one collection fiber [Figures 4(h) or 6(a)]. A quartz shield placed at the distal end of the fibers [Figure 6(a), item f] enables the illuminated and probed area to overlap. The fraction of overlapping increases with an enlargement of the numerical aperture of the fibers and the thickness of the shield. A larger shield thickness requires a larger diameter shield. A typical shield thickness is 1 to 7 mm. If a shield is omitted, fluorescence can still be detected but will originate from deeper layers because the average photon path length is increased. Such a design has been used to measure skin fluorescence.¹⁰⁶

Figure 6

(a) Classic fiber optic probe for fluorescence spectroscopy is based on two or more fibers. Hexagonal packing is a dense arrangement and enables the selection of multiple fibers for different excitation sources and collection channels. A quartz shield permits the overlap of excitation and collection areas. Parts include a, excitation fibers; b, collection fibers; c, tubing; d, carbon-filled glue; e, sleeve; and f, shield.⁶ (b) Single-fiber fluorescence probe with a spacer (optical shield). (c) typical fluorescence probe for single point sampling [legend according to (a) (Ref. 6)].

The collection efficiency β_t for the fiber optic probe described in Figure 6(a) can be described as

Pfefer et al.;¹⁷⁵ showed the influence of the fiber diameter and shield thickness to probing depth and remittance at 337-nm excitation and 450-nm emission and at 400-nm excitation and 630-nm emission. The simulation consisted of a single fiber delivering excitation and collecting fluorescence [Figures 4(g) and 6(b)]. The Monte Carlo simulations assumed typical optical properties for esophageal tissue. The remittance (337-nm excitation, 450-nm emission) increased with increasing fiber diameter and was 0.11 for 0.1-mm fiber diameter and was 0.4 for 1-mm fiber diameter when the fibers were placed in direct contact with the tissue. Eighty percent of the fluorescence signal originated between the tissue surface and a depth of 0.175 mm when a 0.1-mm-diam fiber was used. A 1-mm-diam fiber collected from deeper tissue layers and 80 of the fluorescence signal originated between the tissue surface and a depth of 0.375 mm. Placing a shield in front of the fibers increased the probing depth further. With a 5-mm-thick shield, 80 of the fluorescence was collected between the surface and 0.425 mm for fibers with 0.2 to 1-mm diameters, while the collected fluorescence dropped by a factor of 20 for the 0.2-mm fiber and by a factor of 5 for the 0.6-mm-diam fiber. Similar results were found at 400-nm excitation wavelength with the probing depth increased by approximately 20. This indicates that by varying the excitation and emission aperture one may be able to identify depth variations in fluorophore concentrations.

A typical probe with the design presented in Figure 6(a) consists of excitation fibers, collection fibers, carbon-filled or low-fluorescent epoxy, and tubing. A rigid type uses metallic tubing and a flexible type uses shrink tubing (Zeus Industrial Products, Inc). A probe with an outer diameter of 4 mm, as presented in Figure 6(c), was successfully used by Ramanujam et al.;⁹⁵ in a study with more than 100 patients. The shield and a sleeve are detachable to disinfect the probe. With the classical design, a probe based on seven 200-μm fibers has a diameter of at least 1.5 mm.

4.1.1.

Minimizing the outer diameter

For an improved detection and illumination spot overlap, the shield in front of the probe is replaced with a coated glass rod or a thick piece of optical fiber [Figure 7(a)], which acts as a mixing element homogenizing the output of the probe with two major benefits. During manufacturing, illumination and collection fibers can be randomly mixed and the outer diameter of a probe is reduced and limited by the number of excitation and emission fibers only. Simple probes with minimal diameter can be assembled by stripping the jacket from the excitation and emission fibers, bundling the fibers with glue, and attaching a glass rod with shrink tubing [Figure 7(b)]. The length of this rod should exceed 2R _fiber /NA to enable a uniform illumination.

Figure 7

(a) Reduced outer diameter is achieved with a mixing element instead of a quartz shield. The mixing element consists of a coated quartz rod or a thick piece of optical fiber. The outer diameter does not exceed the diameter of the hexagonal packing. The parts are a, excitation fibers; b, collection fibers; c, tubing; d, epoxy; and e, coated glass rod (mixing element). (b) Assembling of a reduced diameter fiber probe: Illumination and collection fibers (label a), are packed hexagonally (label b) and glued together (label c). The glass rod (label d) is placed in front of the polished fiber bundle and held in place with shrink tubing (label e).

Innova Quart (Phoenix, Arizona) successfully manufactured fiber optic tips using CO ₂ laser micromachining. Melting and compressing fibers in a fiber bundle eliminates the dead space in between the fibers. An increase of power density and an almost complete overlap in between illumination and collection areas was reported. However, this technique is currently limited to fiber bundles with only a few fibers.

4.2.

Multipixel Measurements

With a single-pixel measurement one can obtain spectroscopic information of a small tissue volume, and, based on the performed analysis, a classification of the measurement location is possible. However, for many biomedical applications, a tissue surface area must be investigated, either to determine the extent of a lesion or to identify the location of potential lesions. Pitris¹⁷⁶ and Agrawal¹⁷⁷ successfully adopted the principle of single-pixel fluorescence spectroscopy measurement to a measurement that interrogates 31 locations simultaneously [Figures 8(a) and 8(b)]. For that purpose, the traditional single-pixel measurement is conducted in parallel and data are obtained from a static pattern of pixel locations. Several spectroscopic measurements can be conducted simultaneously with modern aberration-corrected imaging spectrographs. Combined with a large focal plane CCD detector (e.g., 30×12 mm) , these spectrographs can disperse up to 30 and more input channels with minimal crosstalk. The average precancerous lesion size, for example, of the cervical epithelium is between 1 and 2 mm. This means that for the screening of the cervix one does not require a high spatial resolution and can use fiber optic multipixel probes. The outer diameter of the probe in Figure 8(b) is 1 in. and the shield thickness is $\frac{1}{4}$ in. To cover the whole cervix (which is approximately 3 cm in diameter), this probe was placed onto four quadrants separately.¹⁷⁸ Figure 8(a) shows a schematic arrangement of a multipixel probe consisting of several 6 around 1 single-pixel probes, while Figure 8(b) shows the actual implementation with a single excitation and collection fiber. As proposed in Figure 8(a), using several excitation fibers for each collection fiber increases the illumination intensity. An estimation based on commonly used equipment and the previously introduced formula for hexagonal packing shows that more than 500 fibers with a core diameter of 200 μm can be successfully located in a spot created by a standard 300-W xenon lamp. The focal spot size of the excitation source at the coupling site is determined by the arc size, which is typically of the order of 0.5 to 3 mm. A 300-W xenon lamp with a parabolic reflector (Compact Illuminator 6000CI, ORC) produces a spot diameter of approximately 7 mm when a focusing lens with a focal length of 50 mm is used.

Figure 8

(a) Multipixel fiber optic design based on the design of Figure 6(a) (frontal view). At each sample spot, one collection fiber is surrounded by one to six illumination fibers. Crosstalk in between sample spots can be influenced by the separation distance of the collection and emission fibers.⁶ (b) Typical fluorescence probe for simultaneous multipixel measurements with 32 sampling locations. Black shrink tubing holds the quartz shield. Emission and collection fiber pairs are arranged hexagonally.⁶ (c) Separated illumination and collection spot. Fluorescence generated in a 1-mm spot area spreads by scattering inside the tissue and is also detectable outside the illumination area. Since the average path length of the light detected outside the illumination spot is larger, absorption processes are increased and can be observed at increased oxyhemoglobin absorption valleys at 540 and 580 nm (adapted from Ref. 158).

Fiber optic multipixel systems were recently presented by the private sector.^{179 180} An advanced system that can measure fluorescence and reflectance spectra at 120 locations was introduced by Nordstrom et al.;¹⁷⁹ This system incorporates imaging optics and a laser scanning device that can simultaneously illuminate six excitation channels out of 120 rectilinear aligned fibers. To investigate even larger surface areas, researchers have proposed systems that do not rely on fiber optic probes, such as a flying spot scanner,¹⁸¹ or the use of imaging techniques.¹⁸²

4.3.

Influence of Illumination and Detection Spot Separation

Keijzer et al.;¹⁵⁸ described scattering and detection of fluorescence from arterial tissue outside the illumination spot experimentally and theoretically [Figure 8(c)]. The β and α absorption bands of oxygenated hemoglobin alter the fluorescence spectra. Avrillier et al.;¹⁸³ described the same effect and called the fluorescence distortions a “media function.” In experiments with a separate excitation and fluorescence collection fiber (200 μm in diameter), they showed that when brain is excited at 308 nm, the fluorescence ratio at 360- and 440-nm emission decreases with increased fiber separation (2.8 for 250 μm, 2.5 for 500 μm, and 2.4 for 750 μm). Pfefer et al.;¹⁷⁵ investigated the average probing depth in esophageal tissue when fluorescence is collected outside the excitation spot. They simulated an excitation fiber with a diameter of 100 μm and found that inside the excitation fiber area, the collected fluorescence was very sensitive to superficial layers, however, the first 50 μm outside the excitation area, the detected fluorescence, is mainly originates from a depth of 125 μm, while the intensity dropped by approximately a factor of 2.5 compared to the intensity within the excitation area. Absorption in the fluorescence distortion function and, hence, the spectral shape variations are enhanced when the excitation and collection fibers are separated further.¹⁸³ This may be useful to detect neovascularization with fluorescence spectroscopy. In contrast, when multipixel fluorescence measurements are conducted in parallel, the spacing of the collection spots and the shield thickness must be chosen carefully to minimize crosstalk between the measurement channels.

4.4.

Combined Probes for Fluorescence and Reflectance Spectroscopy

It is well known that the absorption and scattering properties of tissues in vivo affect both the intensity and line shape of measured fluorescence spectra.¹⁶⁵ Furthermore, measuring both fluorescence and diffuse reflectance spectra may provide additional information of diagnostic value.¹⁸⁴ Durkin¹⁸⁴ described a probe that combines the measurement of fluorescence and optical properties. Scattered light is collected with a white light “transmission” measurement. This light travels through the same tissue volume, which is excited for fluorescence measurements. The probe, as illustrated in Figures 9(a) and 9(b), consists of a total of 21 optical fibers (200-μm diameter, NA=0.2) arranged in concentric bundles. The center bundle contains 7 fluorescence excitation fibers and 12 fluorescence collection fibers. At the distal end of the probe, the fibers that excite and collect fluorescence are sealed with a quartz shield. This shield is placed in contact with the sample surface and ensures that the area from which fluorescence is collected is the same as that illuminated for reflectance measurements. The reflectance fibers are flush with the tip of the central shield. Similar attempts are described in the literature.^{169 185}

Figure 9

(a) Fiber optic probe that combines fluorescence and reflectance measurements: White light scatters through the volume interrogated during fluorescence measurements. Parts include a, excitation fibers; b, fluorescence collection fibers; c, carbon-filled glue; d, outer tubing; e, quartz shield; f, reflectance collection; and g, white light illumination.⁶ (b) Manufactured probe according to the design of (a). Parts included are the same as in Figure 8(a). (c) Combined fluorescence and reflectance probe. Illumination of the area around the measured spot enables visual feedback during measurement: (label a) Fluorescence fibers (label b) and reflectance fibers (label c) are arranged in the center of the probe. An integrated glass rod minimizes the outer diameter (label d) of the fluorescence channel. The reflectance fibers (label c) are flush with the tissue. (d) Frontal view of the probe and its connectors from (c). Fluorescence excitation and collection fibers (label b) are randomly arranged behind the quartz rod. Reflectance illumination (label E) is facilitated with one fiber, and reflectance is collected at four different source-detector separations (0, 1, 2, and 3). (e) Fluorescence excitation fibers from the design represented in (c) and (d) are arranged in two lines and coupled to the output of a monochromator filtered xenon light source. Fluorescence and reflectance collection fibers are linearly arranged for the imaging spectrograph ferrule. Sufficient spacing between the fibers eliminates blooming on the CCD detector, while measuring low light levels with fiber positions 2 and 3. (f) Compatibility of probe with laparoscopic equipment. A trocar shaft (label a) enables the passage of probe with a diameter smaller than 5 mm for measurements during abdominal surgery. The probe (label b) and shaft are sealed with a flexible lip (label c). Illumination fibers (label d) enables a remote-controlled illumination of the measured spot while endoscopic illumination is dimmed. A second trocar must be placed for endoscopic supervision.

The author’s attempt^{51 186} to measure combined fluorescence and reflectance resulted in a probe with a mixing element for the fluorescence channel, four different source-detector separations and visual illumination for probe placement [Figure 9(c)]. The current probe consists of 46 fibers: 25 for fluorescence excitation, 12 for fluorescence, and 9 for reflectance collection. Reflectance is measured in a circular fashion with fibers placed at 200 μm and 1.1, 2.1, and 3 mm distances to the source fiber [Figure 9(d)]. The detection fibers are linearly aligned in the spectrograph ferrule [Figure 9(e)]. For fluorescence measurements, all emission collection fibers are binned on the CCD chip, while for reflectance measurements all fibers are imaged with appropriate resolution. The outer diameter is smaller than 5 mm, which enables us to measure through a trocar shaft in the peritoneal cavity [Figure 9(f)].

6.1.

A Sapphire-Tip-Based Ring Probe

Sampling along the circumference of the probe is useful to identify the angular position of obstructions in tubes (e.g., arteriosclerosis) or for identification of lesions in tubular structures (e.g., dysplasia in the endocervical canal). A probe initially designed for radial ablation and simultaneous spectroscopic analysis²¹⁵ is illustrated in Figures 12(a) and 12(b). The fibers are glued onto a metallic ring (OD 2.4 mm, ID 2.05 mm). A hollow plug attaches internal flexible tubing for flushing and a guide wire. A custom-made sapphire prism (Swiss Jewel, Tenero, Switzerland) is attached to the plug. The output of the fibers with a NA of 0.22 is totally deflected at the back surface of the prism. An outer flexible silicone tube seals the tip from the liquid environment. Two different outer diameters were realized: 1.4 and 2.8 mm. Figure 12(b) shows the individual parts partially assembled. In the upper picture, the outer tubing is not completely mounted. For spectroscopic measurements, smaller 100-μm fibers were integrated in between the ablation fibers (lower picture). The sealing of such a tip is a tedious task since only small amounts of glue can be used. Similar tips can be produced with vee-, ring-, orifice-, cup-, and chisel-jewels (Swiss Jewel, Tenero, Switzerland; Sapphire Engineering Inc., Pocasset, Massachusetts; SWIP, Biel, Switzerland).

Figure 12

(a) Sapphire prism deflects the output of a fiber ring into a radial direction. The probe consists of a plug (label a) holding the prism (label b) and inner tubing. Fibers (label c) are glued into a metallic ring (label d). Outer tubing (label e) and glue seals the fiber tip.⁶ (b) Upper picture, the sapphire prism and the fibers are partially assembled; lower picture, a cross-sectional view showing the arrangement of illumination and smaller collection fibers in a metallic ring (label d).⁶

8.1.

Spherical Fiber Tips

Melting the end of the fiber shapes the exit surface of optical fibers. An almost spherical surface will be created by the surface tension of liquid quartz.²³¹ Its form is determined by the volume of melted quartz, which is related to the amount of absorbed thermal energy. A number of techniques are used to achieve this deformation: a microfurnace, a Bunsen microburner, an electrical arc, and a CO ₂ laser beam.²³²

The smallest possible surface curvature is a hemisphere. Theoretical results of a beam exiting a hemispherical fiber tip in air and water are shown in Figures 14(a) and 14(b) (modified from Ref. 15). Because the refractive index of silica is 1.46, the focusing power is limited in aqueous media. Figure 14(c) illustrates the focusing of a fiber with a 200-μm core diameter and a NA of 0.22 in water. The beam in water is reduced to a quasi-parallel beam. A light intensity concentration was found at a distance 3 times the distal radius of curvature from the vertex. In simulations²¹⁵ the maximal irradiance was found to be 3 times the irradiance within the fiber. A fiber optic probe for spectroscopic measurements could be manufactured with a single fiber [Figure 1(b)]. The sampling area would be smaller then the fiber diameter and the solid angle of collected light larger than the NA of the fiber.

Figure 14

(a) and (b) Hemispherically shaped ends of fibers focus the output beam in air and collimate the output beam in water (Ref. 6). (c) Maximal irradiance is found at a distance of 3 times the fiber radius in water for quartz fibers with a NA of 0.22. The irradiance increases⁶ by a factor of 3.

8.2.

Ball Lenses

Spherical ball lenses have been used as collimators in fiber optic connectors.²³³ Sapphire spheres with a high refractive power are industrially produced in a wide range of diameters (Sandoz SA, Cugy, Switzerland; Rubis Precis, Charquemont, France). Light emitted from a fiber can be focused with this lens type onto the sample site [Figures 15(a) and 15(b)]. Rol and Niederer et al.;¹⁵ found a maximal illumination intensity when the ratio of the back surface distance to the fiber and the radius of the sphere is in between 1.8 and 2.8 in water and air. Illumination analysis shows a maximal increase of the irradiance by a factor of 4.8 in water [Figure 15(c)] and 5.7 in air [Figure 15(d)]. With increased fiber diameter, spherical aberrations will degrade the focusing of the beam because light will propagate outside the paraxial regime.

Figure 15

(a) Ball lenses image the output of a fiber and focus it. The ball lens is centered and placed at a fixed distance to the fiber with metallic holder. If this probe is used in a fluid, focusing is limited. (b) Side-deflecting probe can be manufactured with a half sphere. The flat surface is inclined to the probe axis. If this surface is a sapphire-air interface, total internal reflection will occur. The flat surface could also be coated with aluminum to achieve a similar effect. The deflecting optic is housed in quartz tube. Focusing in water is not affected, since the beam path is determined inside the probe.⁶ (c) and (d) Illumination analysis demonstrates an up to fivefold increase of irradiance in the focal spot in (c) water and (d) air.⁶

An example fiber optic probe [Figure 15(a)] with a 200-μm core diameter fiber and a sphere with a radius of 0.4 mm will not exceed an outer diameter of 0.9-mm (modified from Ref. 16). The sphere and fiber can be held in a cylinder with a conic inner shape. For side-looking applications, the same principle is used with a half sphere. If the plane is inclined by 45 deg and the step in the refractive index is equivalent to sapphire-air, the beam will be deflected with total internal reflection at the interface and leave the probe at 90 deg with respect to the probe axis. A microquartz tube encloses the half sphere and provides air as refractive media. For spectroscopic measurements a setup as described in Figure 15(b) would be required.

8.3.

Spherical and Parabolic Reflectors

Mirrors can concentrate light.²³⁴ The complex refractive index of a dielectric material defines the absorption and the reflectance of a mirror. An example of a nonimaging parabolic light concentrator is presented in Figure 9(c). For side-looking applications, parabolic or spherically polished mirrors focus the output of several fibers onto the same area [Figures 16(a) and 16(b)]. Best performance is achieved with a parabolic surface [Figure 16(b)]. The manufacturing of such mirrors requires a high-precision, numerically controlled lathe. Surfaces with optical quality can be manufactured with single-point, diamond-cutting machines.

Figure 16

(a) Spherical or (b) parabolic mirror focuses and deflects the fiber output. Rotationally symmetric mirrors with optical surface quality can be manufactured with numerically controlled diamond-turning machines.⁶ (c) and (d) Fiber optic ring probe with a parabolic reflector according to (b). A plug with the reflective surface (d) is mounted into a ring of fibers and flexible tubing. A short piece of a quartz tube isolates the mirror from the probe environment.⁶ (e) Illumination analysis of the output from five fibers was performed with a reflector according the design of (c). The analysis shows a good overlap of the fibers on the exit surface (B) of the probe. The spot diagram on the exit surface (B) shows that the output of the optical fibers is spread in a circumferential direction.⁶ (f) Flux density profile was measured at the exist surface (B) and the five fibers produced an increase of a factor of 2 compared to the intensity inside a single fiber. A ray trace projection (A) shows that the output is focused to a small spot in radial direction.⁶

An example of a probe illuminating a ring is shown²¹⁵ in Figures 16(c) and 16(d). Multiple fibers are glued into a metallic ring. A plug with the optimized polished surface is centered into the ring and deflects the axial fiber output in a radial direction. A quartz tube shields the reflective surface. Illumination analysis shows that the output of five fibers (two fibers were arranged on an inner concentric ring and three on an outer location) overlaps on the circumference of the probe [Figures 16(e) and 16(f)]. The rotational symmetric design spreads the light from the fibers in an elliptical spot around the circumference. The inner ring of fibers could be used as collection fibers and the outer ring as illumination fibers. This design confines the sampling volume close to the probe surface since the light diverges rapidly outside the focal spot [Figure 16(f)].