Research On PCLE Imaging Artifact Correction Method

Probe-based confocal laser endomicroscopy(pCLE)is an effective method for in vivo high-resolution imaging of biological tissue samples,however,artifacts introduced during the imaging process can significantly reduce the effects of pCLE.The resolution at which the original image was captured.The goal of this paper is to obtain higher quality pCLE images by developing specific innovative artifact correction schemes.This paper first analyzes the working mode of the pCLE device used and the causes of artifacts,and then studies the classical image registration schemes used in biomedical image analysis,especially a rigid registration algorithm based on the sum of conditional variances.This algorithm can achieve accurate correction only with the gray value information of the image,and has a large convergence radius.On this basis,combined with the L-BFGS algorithm,a rigid artifact correction method based on the minimum conditional variance sum is proposed,which can correct the rigid artifacts of fiber microscopic images at sub-pixel level.To address non-rigid artifacts introduced during imaging,we investigate a B-spline-based free-transform model.B-spline has good local characteristics,and the surface with nonlinear deformation can be adjusted precisely by using B-spline mesh.This paper combines it with the conditional variance sum algorithm,and proposes a non-rigid artifact correction algorithm using conditional variance and constrained B-spline free deformation.First,global coarse registration is performed on the adjacent frames of the pCLE image sequence to obtain tracking and rigid correction parameters that match the target sample,and then the local control points are adjusted by the L-BFGS optimization method to solve the optimal non-rigid artifact correction parameters to achieve higher calibration accuracy.Finally,we propose a motion compensation-based algorithm utilizing these correction building blocks.We study and calculate a physical model that works in front of a non-rigid artifact correction algorithm to compensate for motion distortions introduced by pCLE’s special imaging modality and mitigate non-rigid deformations.To improve efficiency,we design a parallel acceleration mechanism to execute the proposed algorithm.In the end,our algorithm can restore the globally consistent alignment of the input image sequence,and effectively implement artifact correction,and has a correction efficiency improvement of about 10 times.In this paper,the subjective and objective evaluation method is adopted,and the quality and efficiency of artifact correction of the proposed method are analyzed by using the difference map and RMSE,PSNR and other measures.Experiments show that the artifact correction scheme proposed in this paper successfully improves the image quality of pCLE.