3.1.1.

IR-Microscopic Measurements on Air-Dried Tissue Samples (Absorption/Reflection, Transmission)

For transmission measurements using a distinct path length, soft tissue samples are either squeezed between transparent materials or homogenized at low temperature and filled into a precise measurement cuvette. In either case the microscopic structure of the tissue sample is disturbed. In order to avoid mechanical alteration of the native tissue structure of colorectal cancer, cryosectioning was performed. In previous investigations done by our group, 10 μm thin tissue sections were mounted onto calcium fluoride slides and air dried at room temperature. Neither chemical fixation (e.g., formaldehyde or a freezing medium) nor whole tissue embedding methods were used because this leads to alteration of the microscopic tissue specific structure. Each step of the preparation procedure was carefully checked in order to get reliable and serious wave number ranges for tissue analysis.

The absorption/reflection measurement is a standard procedure for the investigation of thin layers, which can be mounted onto a variety of substrates. Gold is widely used as a reflective coating for investigations in the infrared region.^{19 20} Using this method it has been possible to characterize the conformation of abnormal proteins found in pathological specimens of thin tissue sections of human origin.¹⁹ To enhance the absorbance and the detection of sensitivity investigations of absorption/reflection were carried out using an IR microscope (high spatial resolution) and IR fiber-optic spectroscopy whereby the preparation of the samples differs from the methods given in the literature.^{19 20}

The reliability of an infrared map is limited by several factors: by the system itself, by the measuring parameters and, more often than not, by the tissue quality. The sequence and results of a mapping experiment are shown in Figure 2. To avoid errors in data interpretation the infrared map and the IR map obtained were compared after analysis and mathematical conversion to the HE-stained parallel cryosection. For complete illustration, the native thin tissue section which was used for the IR microscopic mapping is shown in Figure 2(D).

Figure 2

(A) Visual microscopic image (IR microscope) of the HE-stained parallel 10 μm thin section (adenoma carcinoma of the colon). (B) IR map, intensity ratio of the second derivative at wave number ratio of 1223/1209 cm⁻¹, deduced from the original full spectra map. Due to the histological assessment this IR map is superimposed by the use of black and white squares to indicate tumor tissue (black) and connective tissue, (white), e.g., stroma. (C) Pseudocolored IR map, the result of image reconstruction using the correlation coefficient approach. For this the IR-mean spectra (second derivative, range of 900–1450 cm⁻¹) were built up for different types of tissue (see the legend for the color coded assignment). (D) Visual microscopic image (IR microscope) of the native 10 μm thin specimen (adenoma carcinoma of the colon) mounted onto a gold-coated slide.

Derivative techniques are of use for complex spectra and can enhance the sensitivity of the spectrum’s outcome. Figure 2(B) shows the infrared map based on the intensity ratio of the second derivative at a wave number ratio of 1223/1209 cm⁻¹, deduced from the full spectra map. Due to the tissue’s water content before drying and irregularities in tissue preparation normalization of the spectra is an essential step prior to quantitative analysis. Therefore the spectra were normalized in the fingerprint region (1590–900 cm⁻¹) for compensation of sample thickness. To simplify the illustration tumorous parts of the specimen are represented by closed squares, whereas nonaffected tissue appears colorless, which can be assigned mostly to connective tissue (e.g., stroma).

A second map was constructed from the initial one using the correlation coefficient analysis approach [Figure 2(C)]. For image construction of the tissue structure, discriminant analysis of more than two groups of spectra is necessary. In this case, all spectra (A) of a map are compared to a set of IR-mean spectra (second derivative, M_k ) which correspond to a definitive type of tissue. Mixed states were even used in the analysis. For the assignments, pixels (100×100 μm ² ) of the IR map were compared to the HE-stained tissue section. This allowed the corresponding IR spectrum to be correlated with the histological image (IR microscope). The unknown spectrum of the IR map is allocated to the mean spectrum, which results in the largest correlation coefficient. The correlation coefficient is used for definition of similarity:

Figure 3

FTIR-mean spectra of tumor and connective tissue (second derivative, range of 900–1450 cm⁻¹).

3.1.2.

IR Microscopic Measurements on Moist Tissue Samples

Before starting IR-fiber-optic measurements, measurements were conducted with an IR microscope, keeping the samples moist, using remission and ATR. The main purpose of these experiments was to see if there was a significant wave number shift compared to the measurements on dry specimens. A measurement chamber was specially designed and tested for this purpose. Initially, it was not clear whether the specimens were subject to irreversible changes during IR measurement, e.g., cell lyses, a state that could subsequently influence the histological information obtained. However, subsequent histological staining and examination with the microscope showed little or no deterioration as a result of repeatedly conducting the standard procedure.

ATR spectroscopy. ATR measurements were made on thick, dry samples of colorectal tissue. In doing this it was possible to compare the results to equivalent measurements made in transmission mode. The penetration depth of the IR radiation was a limiting factor in this measurement procedure which was of the order of the IR radiation wavelength. The penetration depth can be increased by technical means such as the use of different optic materials or variation of the angle of incidence. The spectral information obtained is the result of a range of tissue components near the surface (mucus, epithelium). A contact method such as the ATR crystal has the additional disadvantage of contaminating the crystal, which affects its chemical resistance and requires cleaning after use, as well as a possible loss of cell material in the process. These problems can be partly solved by using inert materials (e.g., fibers with a diamond tip or a polymer coating in the contact area). Figure 4 shows the results of ATR measurements conducted with the IR microscope and IR fiber sensor on the same specimen.

Figure 4

Left: FTIR spectra obtained using an IR microscope: ATR mode, aperture 100×100 μm ² , 300 μm thick muscle tissue (bovine). Right: Fiber-optic IR spectra (ATR mode, diameter of 700 μm), the sensor element was pressed onto the surface of fresh tissue (muscle, bovine).

The IR spectra of moist muscle tissue (bovine) recorded with the IR microscope and silver halide IR wave guides, respectively, show the typical intense absorption bands for proteins (amide-I and amide-II). The contact area of the tissue/ATR crystal for the IR microscope was 100×100 μm ² . However there is no way of controlling the exact size of the contact area or the pressure of the microcrystal on the soft tissue surface, which was mechanically stressed. The IR-sensor head, equipped with a disposable silver halide fiber (11 cm long, 700 μm in diameter) was connected to an excitation and emission waveguide. The sensing area (tissue/fiber) was approximately 7 mm². As a result minor shifts occur in the absorption maximum. After multiple measurements at the same position (time interval of approximately 2 min) additional absorption bands can be seen, for example, at 1743 and at 1160 cm⁻¹. This is partly due to the intensive contact which leads to a change in the contact area size (the roughness of the tissue surface due to wrinkles, microscopic gaps, etc.), and to a change in water content throughout the course of the measurement (displacement, drying). It results in more intense contact between the medium and the measurement device. The different absorption behavior in the contact area mainly leads to a change in the penetration depth of the IR radiation. As a consequence, the IR spectrum at any one measuring point can change over time. Micro-ATR measurements were carried out on the same native targets. Even the ratio of the intensity of the amide and the lipid absorption bands alters. This process is much more intense for micro-ATR measurements and occurs less frequently in measurements with IR fibers as a result of the large contact area.

However good resolution can be achieved with both ATR techniques using the IR interferometer in the range of 1400–1000 cm⁻¹, therefore making tissue differentiation possible.

Reflection spectroscopy. Systematic experiments using the IR microscope in remission to establish the penetration depth of IR radiation by the use of gelatin sections as a tissue phantom resulted in values of 250–300 μm. As a result, the investigations were conducted on moist tissue samples 500 μm thick to avoid any unwanted interference effects from the slide.

Representative results of one study are shown in Figure 5. Reflection measurements occur in a special measurement cell. The fresh, native, malignant tissue material was measured IR spectroscopically for 3, 24, and 72 h after sample preparation. The first measurement was carried out under moist conditions, whereby the sample was dry by the time the last measurement was made (not shown). To avoid the latent danger of biochemical alteration, the measurement time was reduced during measurements 1 and 2 by using a larger aperture (400×400 μm ² ) with the result that spectroscopic information was averaged over a larger area of tissue.

Figure 5

Microphotograph (IR microscope) of (A) the HE-stained parallel 10 μm thin tissue section and (B) the IR reflection map according to KK conversion for the ratio of the wave number ranges of 1280–1185/1140–995 cm⁻¹ from a 500 μm thick tissue sample (adenocarcinoma of the colon). For orientation the visual picture and the map was marked.

For comparison, Figure 5(A) shows the counterpart to the thick tissue section which was taken before the mapping experiment—the microscope picture of the HE-stained 10 μm thin tissue section. The result of the mapping measurement is shown in Figure 5(B). The visual microscopic image (IR microscope), i.e., the histopathological information, was compared with the resulting full spectra map. The IR spectra were converted according to the Kramers–Kronig (KK) method and then combined onto an IR reflection map. The ratio of the wave number regions (1280–1185/1140–995 cm⁻¹) was evaluated. Only minor changes to the IR maps could be observed between the first and second measurements (3 and 24 h after sample preparation, respectively). The IR map of the same specimen made after 72 h shows changes to the tissue surface which are a result of effects caused by drying (not shown).

Visual inspection of the HE-stained specimen allows the tumorous parts to be differentiated. Even the result of the reconstructed IR map from the experimental data showed that differentiation between malignant tissue and other parts of the specimen is possible. In conclusion, the wavelength range chosen for the tissue differentiation in transmission can be transferred for measurements in remission mode.

Fiber-optic based laboratory setup. Before the ultimate aim of differentiating tissue using an endoscopic device could be achieved, it was essential to complete a fiber-optic based laboratory setup. The IR fiber was connected to the FTIR interferometer. Commercial silver halide fibers were used as the transmission material of choice because of their excellent physical properties with regard to transmission, numerical aperture, damping, flexibility and aging.

Description of laboratory setup. Coupling of the IR radiation was carried out by a closed unit connected to the interferometer with either a preadjusted KRS 5 lens system or a ZnSe lens. On the excitation side, an IR waveguide ( L=1.5 m, d=700 μm ) was prepared on both ends with a Ti-SMA switch. An external MCT pigtail detector was used (active detector area of 500×500 μm ²; Infrared Fiber Sensors, Germany) on the detection side. A permanent fiber/detector coupling ( L=1.8 m, d=700 μm, Ti-SMA switch) was fitted on the detector side by a silver halide microlens. The sample compartment is shown Figure 6.

Figure 6

View of the sample compartment and the fiber-optic based laboratory setup.

Results: Preliminary experiments were carried out. An example will be explained in detail. The moist tissue material was examined with both the fiber-optic based laboratory setup and the IR microscope. The following results were obtained. Just for documentation purposes, the visual microscopic image of the native, moist 300 μm thick tissue section was prepared by the IR microscope [Figure 7(A)]. Then an IR map (remission mode) of a restricted area of the tissue section was recorded. On the basis of these data, and in accordance with the previously successful KK conversion, a pseudocolor map was made for the area ratios of the wave number region (1280–1185/1140–995 cm⁻¹) using the converted IR-spectral data. The recorded reflection spectra of the “L” marked regions after KK conversion are shown in Figure 7(B). These spectra can be compared with the fiber based measurements that were carried out on the same tissue sample [Figure 7(C)]. IR fibers with a diameter of 700 μm were used. Optimal results were obtained using an experimental setup in which both the excitation and detection fibers had an angle of incidence of 60° to each other. The fiber to tissue distance was approximately 0.5 mm. A gold standard background was recorded every 10 measuring points, analogous to the IR microscopic measurements. Taking the spectral characteristics of the IR fibers into account, the fiber-optic based IR map that resulted from the previously described data processing [the ratio of the wave number regions (1262–1130/1103–972 cm⁻¹)] corresponds well with the IR microscopic result. Areas without color in the IR map are assigned to the slide material.

Figure 7

(A) Visual microscopic image of a 300 μm thick, native, moist specimen (adenocarcinoma of the colon). (B) IR microscope: Pseudocolored IR map (aperture 400×400 μm ² ) after KK conversion and IR reflection spectra, which correspond to the positions marked on the map. The wavelength region used for analysis is marked on the spectra. (C) IR fiber optic: Pseudocolored IR map and a number of corresponding IR reflection spectra after KK conversion that were used for construction of the reflection map. The wavelength range used for analysis is marked on the spectra.

Representative reflection spectra are shown in Figures 7(B) and 7(C). The wave number regions are marked separately for spectroscopic differentiation in the IR map.

It should be emphasized that the results (reflection mode tissue spectra) of a fiber based laboratory setup using an IR interferometer as the excitation source can be compared to the IR microscope results using moist tissue samples. The spectral information allows differentiation between healthy/malignant tissues. Moreover, from the results, the wave numbers for colorectal tissue differentiation could be verified compared to the IR measurements on dry tissue specimens. This aided in the choice of suitable IR diode laser for the experimental laboratory setup.

3.2.1.

Experimental Setup with IR Diode Lasers as the Excitation Source

Experiments using a lead salt diode laser were realized according to the schematic diagram shown in Figure 8.

Figure 8

Schematic diagram of the fiber-optic based laboratory setup using lead salt diode lasers.

For the laboratory setup two high power multimode lead salt diode lasers (Mu¨tek Infrared Systems, Germany) were used as the IR-excitation source [diode 1: 995–1045 cm⁻¹ (9569–10 050 nm)], [diode 2: 1145–1245 cm⁻¹ (8032–8733 nm)]. The bandwidth is between 20 and 30 cm⁻¹. The usable wavelength range is approximately 50 (diode 1) and 100 cm⁻¹ (diode 2). The actual wavelength depends on the diode current. The current is modulated at the laser driver in order to scan the wavelength range. After being collimated with spherical concave mirrors, both IR beams are overlapped and are then coupled into a fiber using another spherical concave mirror. The IR fiber and the MCT detector are described above. The signal at the detector is evaluated using a lock-in amplifier (model SR830, Stanford Research Systems, CA). To position the sample and to record a map, a translation stage with three axes is set up which moves the sample underneath the fixed IR fibers and is controlled by a PC. The resolution in either direction is 1 μm.

To evaluate the setup of the IR diode lasers, the recorded spectra were compared with the spectra of the FTIR interferometer. The results cannot be compared directly because the excitation sources differ. The interferometer may use a Nernst glowing bar which can be regarded as blackbody radiation.

The diode lasers emit at different modes while the diode current is scanned. This means that the spectral emission has to be taken into account at every wavelength in order to obtain a spectrum that is comparable with the spectrum from the FTIR interferometer. Deconvolution of the IR spectra is not necessary because the differentiation of tissue requires only the detection of relative differences.

3.2.2.

Results

Figure 9 shows preliminary results from thick, moist normal and malignant colorectal tissue specimens. The tissue area was scanned point to point in rows. Each point represents the raw data and the data normalized to the gold standard. All spectra of the raw data show almost the same course, the main differences being in intensity. From the index point 140 differences occur. Differences can be clearly seen between the spectra of normal and malignant tissue where diode laser 2 is the excitation source. Not only are there differences in intensity but also additional signals. The high level of noise in the normalized spectra is due to the low intensity of the excitation sources.

Figure 9

Preliminary results from measurements using IR diode lasers on moist, normal and malignant tissue samples. Each spectrum equals one measuring point on the tissue. (A) Raw data of normal (left) and malignant (right) tissue. (B) Data normalized to the gold standard of normal (left) and malignant (right) tissue.

The initial results indicate that measurements using IR diode lasers and moist tissue samples are possible and that typical tissue differences can also be detected. This proves, in principle, the feasibility of this measurement method. The measurement time could be further reduced by optimizing the conditions.