This PDF file contains the front matter associated with SPIE Proceedings Volume 12820, including the Title Page, Copyright information, Table of Contents, and Conference Committee information.

We present an endoscopic system allowing to perform 3-photon excitation fluorescence and third harmonic generation imaging. The ultrashort pulses required for multiphoton excitation are delivered from an ultrafast laser system to the endoscopic probe using a connectorized hollow core delivery fiber.

We present here the concept of a MEMS-mirror based nonlinear endomicroscopic probe for coherent anti-Stokes Raman scattering (CARS), two-photon excitation fluorescence (TPEF), and second harmonic generation (SHG). The rigid probe head is 5 mm in diameter and 4 cm in length, offering a large field of view (FOV) with a high numerical aperture (NA). It incorporates a double-core fiber delivering two excitation wavelengths of CARS in isolated cores, with a large cladding area to increase the collection efficiency of the nonlinear signal from the tissue. A diffractive grating element is included in the probe to compensate for the spatial offset of two emission cores.

Early indicators of cancer manifest as cellular abnormalities that can be imaged via high resolution imaging modalities such as label-free Multiphoton Microscopy (MPM). Implementing MPM endoscopically can aid in the early diagnosis of cancer. We are developing a minimally invasive microendoscope system capable of simultaneous co-registered multiphoton imaging (two- and three-photon excited fluorescence, second and third harmonic generation) of the epithelial layer in small diameter ductal tissues via helical scanning of a 1.0mm diameter endoscope distal end with a fixed focus. The endoscope working length is comprised of a stationary outer sheath housing a proximally driven endoscope distal optical system. The lenses are 0.5mm in diameter and side viewing, requiring a novel optical design with power on the exit surface of the fold prism. Additive manufacturing (three-dimensional [3D] printing) opens significantly more possibilities for distal end microendoscope optical design. We present the design of the distal end outer sheath, housing, and optics, as well as an evaluation of the feasibility of 3D printed optics for a high numerical aperture (HNA) MPM microendoscope system. The selected distal end outer sheath meets flexibility, size, and optical requirements suitable for a first iteration lab prototype. The distal end housing was custom designed to be easily attached and removed from the proximal system and to ensure stable and consistent helical motion of the inner probe distal end when it is flexed and bent in manners needed for use in curved tissue lumens. We have also designed and manufactured (via two photon polymerization) multiple configurations of a 0.5mm monolithic multi-element lens system containing an aspheric surface on the exit face of the prism and evaluated its optical performance.

High performance multimode fiber (MMF) endoscopy is a possible thinnest endoscopy for widely applications, The main problem is how to real time overcome the mode coupling during the movement of the fiber and to get the high performance imaging. In this paper, we present our method to high speed compensating the mode coupling during the fiber movements, and present the possibility of the multimode fiber can not only be the finest endoscopy imaging, but also can be used to realize multi-modal imaging.

The magnetically controlled capsule endoscopy (MCCE) is an emerging modality for assessing gastrointestinal disorders due to its advantages. However, current assignments of MCCE rely on manual controlling and gastric landmarks, which are prone to omissions. We improve the scanning protocol of the MCCE in human gastric using both manual and automatic controlling methods. We design a quantitative scanning coverage ratio to measure the process of MCCE scanning within the human gastric. The proposed scanning coverage ratio is capable of guiding the manual and automatic scanning process of human gastric. Moreover, we design a deep reinforcement learning (DRL) controller for automatically navigating the capsule. Our DRL controller achieves a higher coverage ratio compared to previous research.

Endoscopic optical coherence tomography (OCT) offers near histologic quality visualization of tissue microanatomy in vivo, circumventing the challenges of traditional biopsy by enabling volumetric sampling without tissue removal. Miniaturized probes have been designed to overcome the limited imaging depth of conventional OCT, thereby facilitating minimally invasive imaging. Visible light OCT (vis-OCT) endoscopy has the potential to achieve ultrahigh resolution of less than 2μm with enhanced image contrast. However, current vis-OCT endoscopes, which rely on achromatic lenses and distal motors, are cumbersome and pose safety concerns in clinical environments, underscoring the need for ultracompact, current-free alternatives. Additionally, conventional fabrication methods for high-performance, ultrathin vis-OCT endoscopes have limitations. In this study, we introduce a submillimeter monolithic vis-OCT endoscope, created by directly coupling microlens to the fiber tip using a liquid shaping technique. Our ultrathin vis-OCT endoscope of 0.4mm in outer diameter enables ultrahigh-resolution (1.4μm × 4.5μm in axial and transversal directions) interstitial imaging in vivo.

Inline holographic microscopy has recently been demonstrated through fiber imaging bundles, opening up the possibility of ultra-miniaturized microscopy probes. In order to minimize artefacts arising due to the multimode behaviour of the fiber bundle cores, a partially coherent light source was used: an LED coupled into a multimode fiber. However, partial coherence limits the maximum working distance between the bundle and the sample before the resolution begins to degrade. The resolution is also limited by the finite core spacing in the fiber bundle, leading to under-sampling of the finer details of the hologram. Here, we demonstrate and evaluate several techniques for improving the resolution and working range, including tailoring the source coherence and using multiple sources, demonstrating that we can achieve at least a two-fold improvement in performance.

To detect disease in organs with small lumens like the pancreas or fallopian tubes, where lumens may be collapsed or filled with mucus, cilia, or plicae, minimally invasive submillimeter endoscopes must meet unique optical requirements. They must provide a moderate angular field of view (AFOV), a close working distance (WD), and provide adequate resolution for the application. Additionally, when using miniature fiber bundles for image relay, the effect of the honeycomb pattern and potential fiber crosstalk must be considered. We compare the optical performance of a gradient refractive index (GRIN) singlet, a 3D-printed doublet, and a 3D-printed triplet, as well as discuss the consequences of fiber bundle use. We show that for our designs, both GRIN lenses and 3D printed aspheric lens systems can meet our application requirements, with to the ability to resolve less than 10μm at 10% contrast. The effect of fiber bundle crosstalk is shown to be negligible because of limited mode coupling in realistic manufactured fiber bundles.

Endoscopic examination has the advantage of being minimally invasive, but the field of view is narrow and the shape of the gastrointestinal tract creates blind spots that may cause lesions to be overlooked. It is thought that these problems can be solved by performing 3D reconstruction from images using Visual SLAM (Simultaneous Localization and Mapping) and grasping the structure in three dimensions. However, when this method, which requires matching between images, is applied inside the gastrointestinal tract, the similar environment in the gastrointestinal tract continues repeatedly, which prevents appropriate matching and appears as noise during 3D reconstruction. Therefore, we propose a method that evaluates the reliability of feature points using epipolar constraint equations and optical flow for matching points of images obtained from a monocular camera, and classifies whether they are correct matching points by machine learning. The specific methods are 1) image matching, 2) calculating epipolar constraint formulas and distances between matching points, and 3) classifying whether matching points are correct by machine learning. 4) Perform 3D reconstruction using only correct matching points. In order to demonstrate the effectiveness of this method, we conducted experiments using simulation images with known three-dimensional structures. For machine classification, K-means clustering method and nonlinear SVM (Support Vector Machine) were used for comparison. We also conducted a similar experiment with a real object. These results suggest that the method can perform correct 3D reconstruction even in the gastrointestinal tract and contribute to the identification of lesions such as early esophageal cancer.