Research On Swept Source OCT And Human Eye OCT Image Based Axial Length Measurement Method And Calibrator

Axial length is one of the key biological parameters of human eyes.The change of axial length is closely related to the diagnosis and treatment of glaucoma,cataract,and macular edema.Therefore,the axial length with accurate measurement is the data basis for the diagnosis and treatment of ophthalmic diseases.The axial length measurement methods based on ultrasonic and optical interference use ultrasonic reflection ranging and distance between two interfering optical signals to measure the axial length respectively.The measurement processes of these two kinds of methods are easily affected by the complex structure of human eyes,the different refractive index of different structures,and the transmittance of the diseased lens.This condition resulted in low measurement accuracy,poor numerical repeatability,and high measurement error rate,which make it difficult to obtain accurate and effective axial length data.Swept source optical coherence tomography(SS-OCT)is a kind of imaging technique based on the Michelson interferometer.Due to the advantages of noninvasive,high resolution and the measuring length by imaging,the SS-OCT technology was applied to the human eye structure imaging and axial length measurement.The axial length measurement method based on SS-OCT measures the axial length by segmenting the cornea and retina in the human eye OCT image and obtain the distance between anterior corneal surface and retina.This method can reduce the influence of interference factors such as complex structure and lens transmittance,to improve the measurement accuracy,improve the numerical repeatability and reduce the measurement error rate.However,the large amount of speckle noise in the human eye OCT image and the integration effect of the image signal acquisition equipment affect the human eye OCT image quality,the human eye OCT image segmentation accuracy is low and there is no calibration method for accuracy of axial length measurement.The above issues restrict the improvement of accuracy of the axial length measurement method based on SS-OCT.Therefore,research on SSOCT and human eye OCT image based axial length measurement method and calibrator to improve the accuracy of axial length measurement method based on SS-OCT has important theoretical significance and application value.Based on the analysis of the existing axial length measurement method,OCT image processing key technology and axial length measurement calibrator,this thesis researches SS-OCT and human eye OCT image based axial length measurement method and calibrator.The researches are mainly completed as follows:1.A reference variance and adjustment factor based wave atoms image speckle reduction method for human eye OCT image was proposed.First,a reference image and an adjustment factor were introduced,and the speckle noise reference variance of reference image was adjusted using the adjustment factor to obtain the accurate speckle noise variance of target image.Then,a wave atoms threshold equation including the reference variance and the adjustment factor was constructed to obtain an accurate wave atoms noise reduction threshold.On this basis,speckle noise was shrunk using the wave atoms noise reduction threshold to achieve speckle noise reduction in human eye OCT image,and human eye OCT image with clear small objects and sharp structure edges was obtained.2.A anisotropic total variation regularization(ATVR)based nonnegativity and support constraints recursive inverse filtering(NAS-RIF)blind human eye OCT image restoration method was proposed.First,ATVR was introduced into the cost function of NAS-RIF to improve the noise robustness and structural edge definition of the NAS-RIF blind restoration.The split Bregman iteration was used to optimize the ATVR-based cost function.On this basis,ATVR based NAS-RIF blind image restoration method was constructed.Then,the ATVR based NAS-RIF blind image restoration method was combined with the Poisson-distribution-based minimum error thresholding method for blind human eye OCT image restoration to improve human eye OCT image quality.3.An RM Multi-stage Mask R-CNN human eye OCT image segmentation method for axial length measurement was constructed.First,the method used the ResNet50 networks to extract image features.And the feature pyramid networks were used to fuse high-dimensional and low-dimensional object features.Modulated deformable convolution was used to generate irregular receptive field,which improves the fitting degree of the networks to the complex deformation of human eye OCT image objects.In the feature map extraction,the feature maps of the position relation between the human eye OCT image structures were learned by the relational model,and the feature maps were enhanced by the feature maps of the position relation.The multi-stage mask was constructed in the segmentation branch,and the segmentation error was corrected through the multi-step iteration of mask,to improve the segmentation accuracy of the networks to the human eye structure.Combined with the above structures,an RM Multi-stage Mask R-CNN human eye OCT image segmentation method was constructed for human eye OCT image.Then,the RM Multi-stage Mask R-CNN model was trained using the human eye OCT images and the trained model was used to achieve high accuracy segmentation of the human eye OCT image.Finally,axial length measurement achieved by the measure distance of segmented anterior corneal surface and retina.4.A set of axial length measurement calibrators were developed.The axial length measurement calibrator was designed as a cylindrical structure.The reflection and scattered light on the front and rear surfaces of the cylinder generated interference to realize the length measurement of the axial length measurement calibrator,and then the measuring accuracy of the axial length measuring equipment was obtained.According to the range of axial length,three kinds of calibrators length of 15 mm,20 mm and 30 mm are designed.polymethyl methacrylate(PMMA)material and K9 glass material were used to make the axial length measurement calibrators,and the length and refractive index of the axial length calibrator which are key parameters of calibrator were calibrated.The reference variance and adjustment factor based wave atoms image speckle reduction method proposed in this thesis could effectively reduce the speckle noise in the human eye OCT image,and make the small objects and structure edges clear.It provides a new method for speckle reduction in human eye OCT image The ATVR based NAS-RIF blind human eye OCT image restoration method proposed in this thesis could effectively realize the blind human eye OCT image restoration,improve the quality of human eye OCT image,and provide effective blind image restoration method for human eye OCT image restoration.The RM Multi-stage Mask R-CNN human eye OCT image segmentation method for axial length measurement constructed in this thesis can accurately segment the cornea,lens,and retina in the human eye OCT image,realizing accurate measurement of axial length and providing an accurate image segmentation method for SS-OCT based axial length measurement.The developed axial length measurement calibrator could realize calibration of axial length measuring equipment.The calibrator had a wide range of applications in the field of axial length measuring equipment calibration prospect.