Study On Signal Intensity Correction For Three-dimensional Microscopic Brain Image Based On Background Estimation

In neuroscience research,the study of neural circuit structure at a single neuron level helps to understand the operating mechanism and working principle of the brain.Images that contain brain fine structures like nerve fibers can be acquired by micro-optical sectioning tomography system.With these images,we can study the neural circuit at a single neuron level.However,due to the uneven illumination of the imaging system and the instability of sectioning,images collected by these systems often have the problem of uneven image signal intensity,like uneven background brightness and sectioning artifacts.These will bring difficulties for quantitative data analysis.To correct uneven image background caused by different factors mentioned above,different correction methods are proposed in this thesis.For the uneven background brightness,this thesis proposes an improved retrospective method.Firstly,the model is built for the uneven background.Then the optimization algorithm is used to estimate the optimal model parameters.Finally,correct the images with the optimal parameters.For the sectioning artifacts,consider the uneven background is not regular,we can’t build a model.This thesis proposes a processing method based on gray-scale morphology,which extracts the uneven background through the combination of morphological opening and closing operations,and finally the corrected results are obtained by subtracting the original image and the background.The whole brain-wide fluorescence images and Nissl-stained images are corrected by the correction methods proposed in this thesis,and multiple quantitative indexes are used to accurately and objectively evaluate the images quality before and after processing.The results show that the improved retrospective method can correct the uneven background brightness,and the neuron fiber orientation and cell are clear after correction;the gray-scale morphology method can effectively correct the sectioning artifacts in Nissl-stained images,and the background is uniform after correction,the cell boundary is very easy to distinguish.At the same time,the quantitative evaluation indexes also show that both images qualities have been significantly improved after correction.In summary,the proposed image intensity correction methods can both correct the uneven image background and improve the images quality on the whole brain-wide images.