Abstract:
Early therapy of malignant neoplasia (cancer) leads to clearly better survival rates, however, early stage neoplasia are generally harder to detect and diagnose correctly. Diagnosis is based on cellular differentiation (grading), which up to now can only be determined in vitro by microscopy of thin sections of a biopsy sample after histologic processing. Epithelial neoplasia (carcinoma) would be accessible to histology in vivo, but conventional microscopes cannot image single cellular layers because of the strong scattering within tissue volumes. Confocal microscopy and optical coherence tomography are tomographic methods, which effectively filter scattered light that otherwise would contribute to image intensity. They can be miniaturized using optical fibers. Optical coherence microscopy combines the advantages of both methods, i.e. high lateral resolution provided by confocality and long information depths in biological tissue provided by the interferometric coherence gate. Images are acquired in lateral direction at a fixed axial position of the coherence gate. Therefore, a phase modulation has to be employed in the interferometer's reference arm to measure the degree of coherence and thus the image intensity. This three-part work develops optical coherence microscopy towards a system for endoscopic in vivo-histology.
In the first part an optical coherence microscope is set up on a laboratory scale to investigate to which degree confocal low-coherence reflectometry can be applied to image epithelial layers of the gastrointestinal tract. With a center wavelength of 1310 nm and a coherence length of 14.4 µm, lateral and axial resolutions are 1.45 µm and 3.0 µm, respectively. Scanning of the sample as well as the phase modulation in the reference arm are performed piezoelectrically. The optical coherence microscope provides a 28 dB improvement in dynamics compared to conventional confocal reflection microscopy and is less sensitive to aberrations. Robust and clinically relevant algorithms for a mostly automated operation of an optical coherence microscope are developed. The point spread function of the optical coherence microscope is measured as a function of the reference arm length and interpreted regarding the autocorrelation function of the light source spectrum. It is, among other techniques, used to propose ways to reduce speckles. The fringe-like artifact related to the object phase is described for the first time. A very simple method to determine surface topologies is introduced, with which highly resolved isophase lines separated by an eighth of the center wavelength can be derived. Cellular membranes can be identified for all epithelial layers of the gastrointestinal tract. Columnar and squamous epithelium can be automatically distinguished based on azimutally averaged power spectra of images recorded close to the surface.
In the second part mathematical foundations for the computer-based simulation of the imaging of an optical coherence microscope are worked out. For each line of an image the signal of the time-dependent detected intensity is simulated as well as the demodulation of this signal. The tissue model comprises an arbitrary number of reflecting boundary faces. The spectral power density of the light source can be modelled as a sum of an arbitrary number of gaussian distributions. The simulation is applied successfully to provide evidence for the cause of the fringe-like, object phase-related artifact. Besides being compared to an experiment, the result is substantiated theoretically, which leads to a refined discrimination between two causes of the artifact, one of which is inherent to the imaging method itself while the other is due to alignment issues. A strategy to minimize the critical fringe-like, object phase-related artifact caused by misalignment is shown.
In the third part a realizable concept for an endoscopic optical coherence microscope is developed based on the results yielded before. A suitable confocal probe is presented. Using a gimbal-mounted miniaturized mirror for scanning, it achieves an image rate of 0.5 Hz. For an electrooptic phase modulation with a lithium niobate crystal at 10 MHz frequency an impedance-converting circuit is desiged to enable the use of a low-power oscillator to drive the modulation. For the high-frequency demodulation of the interference signal an economical analog circuit is set up. The concept includes additional adaptions to the increased bandwidth of image acquisition, which is significantly higher compared to the laboratory setup. In addition, boundary conditions for a clinical application are taken into account.