The Scattering Correction Based On Coherent Optical Adaptive Technique And Deep Learning

At present,fluorescence microscopy has been widely adopted in biomedical researches.In many researches especially those concerned with brain,fluorescence microscope often has to perform imaging deep through living biological tissue.However,the highly scattering as well as complexity of the tissue severely limit the depth and the accuracy of deep tissue fluorescence imaging.Thus,based on relevant researches,this work uses coherent optical adaptive technique(COAT)to correct the aberrations induced by the tissue scattering in two-photon microscopy.A new method is proposed to optimize this technique by making up for the shortcoming of the conventional modulation frequency selection method.The correction ability of COAT for twophoton microscopy and the distinct superiority of the new method are demonstrated by numerical simulations.For living tissue imaging,it is not enough to just achieve the correction of the illumination wavefront distortion caused by the tissue scattering.The dynamic scattering response resulted from the movement of the living tissue also need the correction process done in fast speed.Therefore,researchers introduced deep learning to accelerate the correction.Previous deep learning guided correction techniques have accomplished partly corrections to the aberrations.Based on this,a wavefront zonal correction method is put forward in this work to take place the former modal approach in the deep learning guided correction.A novel deep learning algorithm is designed to match the newly proposed method.Datasets are built by numerical simulation to train and test the deep learning network.According to the numerical simulation results,the correction of dynamic complex aberrations can be finished with fast speed by the deep learning guided wavefront partition correction method.To sum up,firstly,this work enhances the imaging contrast of two-photon microscopy by optimizing the modulation frequency selection method of COAT;secondly,a new deep learning guided wavefront zonal correction method is proposed to deal with the aberrations from the living tissue.Dynamic complex aberrations can be corrected by the new method with fast speed,which lies the foundation for the deep living tissue fluorescence microscopy of better imaging performance.