Research On Multi-scale Multimodal Biomedical Imaging Techniques Of Chronic Wound Optical Phantom

The 3D printing and characterization system of biomimetic phantoms for biomedical optical imaging integrates techniques such as multimodal imaging, micro-fabrication and 3D phantom printing, to produce and characterize multimodal traceable phantoms for standardized biomedical optical imaging. Optical phantoms can be used to validate and calibrate clinical medical imaging devices for the improved performance. The characterization module of the system utilizes multimodal imaging techniques to acquire the structural and functional information of biological tissue in real time, making use of its advantages of low-cost, fast speed and noninvasive data acquisition.In this paper, we proposed two multimodal imaging systems in both macroscopic and microscopic levels, especially for application on quantitative wound tissue imaging. The 3D surface, oxygenation distribution and micro vessel reconstruction were obtained. We systematically studied the performance of the multimodal techniques we use in the systems and experimentally measured the key parameters, providing valuable new ideas for the setup of the optical characterization module in the phantom printing system. The major research content and contribution of this paper are listed below.(1) We built up a 3D reconstruction system based on multiview imaging. By introducing a mirror set with a rectangular cross-section, cohered with four inside-facing flat mirrors, we successfully extended the only one digital camera into five sub-cameras (one real sub-camera and four virtual sub-cameras) locating at different positions. These sub-cameras could capture images of the subject simultaneously from different view points. With the camera calibration method used in computer vision, we calibrated our multiview imaging system and calculated the intrinsic and extrinsic parameters.(2) We built up a macroscopic scale multimodal imaging system integrating multiview imaging and hyperspectral imaging. The system is able to obtain the structural 3D surface topography and functional oxygen saturation map of a biological sample. It consists of a high resolution CMOS DSLR camera, a mirror set, a high sensitivity CCD camera, a hyperspectral light source and a fiber based light ring. It can obtain the functional as well as the structural information within a very short time and generate a 3D functional map of the biological subject. The system performance was quantitatively validated by several experiments:a 3D reconstruction test set-up was built to measure the accuracy and axial resolution; a foot wound model was used to visualize the reconstruction ability of biological tissue; a blood phantom was reconstructed with both 3D surface and blood oxygen saturation information to show the possibility of in-vivo measurement; and an in-vivo healthy human figure blood flow occlusion experiment was conducted to test the ability of dynamic measurement. All the experimental results validated the great performance of our macroscopic multimodal imaging system.(3) For the microscopic imaging level, we built up a multimodal microscopic system that integrated three modalities of structured illumination microscopic imaging (SIM), fluorescent microscopic imaging and hyperspectral microscopic imaging. The combination of these modalities can image the structural and functional information of the sample at the microscopic scale. The system consists of a hyperspectral light source, a high sensitivity air-cooled camera, a DMD chip with its control module, multiple objectives and filter sets. A control software interface developed in LabVIEW environment was designed to control all the components systematically. The parameters that a user can modify with real time image display include the output light intensity and wavelength control of the hyperspectral light source, exposure time control of the camera, motion control of the sample stage and the pattern control on the DMD chip.(4) We proposed a method of acquiring the hyperspectral data cube using the SIM acquisition mode in our system. SIM imaging technique can generate a stack of sectioned image while the max projection image of the stack in Z direction can eliminate all blurred area in the sample image due to defocus. However, in the hyperspectral imaging, intensity of the sectioned image cannot retain all the reflectance spectrum information. So we developed a method that combines the sectioned image stack with the bright-field image generated by the three raw pattern images, gets the in-focus pixels out of the bright blurred images and assigns them to the max projection image. At last, we can get a hyperspectral data cube where each layer of image taken at a specific wavelength has all of its pixels in-focus and their intensities has the original hyperspectral reflectance value.(5) Several experiments were conducted to characterize our microscopic multimodal imaging system. The standard microsphere surface reconstruction experiment showed the SIM mode has an accurate surface reconstruction ability. An Ibuprofen bead and two letters on a dime coin were reconstructed with the SIM mode. The fluorescent SIM module was tested in the ICG crossed channel reconstruction experiment. Also we used the fluorescent SIM to reconstruct the auto-fluorescence of a stamen and several pollen grains. At last, to validate the hyperspectral microscopic imaging module, we reconstructed the oxygen saturation of a micro-channel filled with differently oxygenated blood sample. Two or more modalities of this system can be combined together to acquire much more useful characteristics of the sample, like true color surface reconstruction, optical biopsy and 3D oxygenation mapping, etc.