Endoscopic Frequency-domain Diffuse Optical Tomography Technology For The Diagnosis Of Cervical Cancer

As an extension of diffuse optical tomography (DOT), endoscopic diffuse opticaltomography (EDOT) is a new functional imaging technology which is noninvasiveand can detect internal organs correctly. The technology includes imaging system,endoscopic probe and image reconstruction algorithm, the last two of which arestudied in this paper for the diagnosis of cervical cancer.Firstly, an endoscopic reconstruction algorithm based on total detection points isdeveloped using both the amplitude attenuation and phase shift on frequency-domain,which is suitable for the diagnosis of cervical canal. The finite element method isemployed for numerically computing the forward problem, and the Gauss Newtonmethod is applied to solve the inverse problem. To cut down the computing time andsimplify the structure, the Jacobian matrix is built by combining the adjoint methodwith the modified Generalized Pulse Spectrum Technique. The Generalized MinimalResidual (GMRES) Krylov method is used for obtaining the iterative update factor tocut down the computer memory resources and computing time. The simulation resultsof the reconstruction algorithm show that the quantitative ratio can reach up to morethan80%, and the effect of the target depth could be eliminated when locating thetarget’s position.Secondly, for the propagation characteristics of light in the tubular model, aconcept of effective detection range (EDR) is proposed, and a corresponding imagereconstruction algorithm is developed to cut down the computer memory resourcesand computing time. The source and detection sites can be reasonablely arrangedaccording to the size of the EDR which is derived from the equation of the EDR. Thesimulation results based on the reconstruction algorithm show that the reconstructedimages with the EDR could achieve the similar accuracy which is obtained by usingtotal detection points, confirming the validity and general applicability of thealgorithm.Thirdly, a frequency-domain EDOT system based on EDR is developed. Anendoscopic rotary probe with two optical fibers whose heads are cut to45oisdesigned for delivering and collecting light perpendicular to the optical fiber atmulti-source and detection sites. The measurement system and the image reconstructed algorithm are evaluated on solid phantoms and tubular biological tissues.The experiment results demonstrate that high reconstructed image quality could beobtained by the measurement system, confirming the validity of the EDOT systemand the reconstruction algorithm.Finally, multi-frequency modulation, the diffusion equation for dual-pointssource approximation and the region of interest (ROI) are used to improve thereconstruction algorithm based on EDR. The simulation results show that thediffusion equation for dual-points source approximation is more accurate than thediffusion equation for single-point source approximation to simulate photon transportin biological tissue. The use of dual-frequency modulation is sufficient to improve thequantitative accuracy. The algorithm based on ROI can effectively enhance the qualityof the reconstructed images.