Research On LED Based Lens-less In-line Digital Hologram For Red Blood Cell Physiological Parameters

Blood Routine Examination is a common clinical detection item in medical diagnosis and therapy,which may reflect abundant human physiological and pathological information.Nevertheless,current clinical devices in use are normally based on Kurt principle or flow cytometry and involve several inherent problems,for example,damage toward sample cells by reagent,complexity of light-path or flow-path,costliness,difficulty in miniaturization and daily maintenance as well.Fortunately,lens-less in-line digital holography,with advantages of full-field,large target surface,contactless with sample,and high resolution etc.,even could image transparent objects clearly,which means it possesses intrinsic merit in microscopic imaging toward bio-sample.Therefore a lens-less in-line digital holography platform was used in this project to image blood sample and the algorithms for the holographic reconstruction and the cell counting analysis were developed with the aim to process the obtained blood cell hologram and extract the red blood cell(RBC)-relevant physiological information.In more details,since convolutional based algorithm is limited in direct reconstruction with large-scale hologram,overlapped window moving pooling strategy was proposed to reconstruct the whole hologram sub-block by sub-block.As for each sub-block,Tenengrad(TEN)auto focus functio,which depends on edge gradient,was applied to seek best reconstruction distance.Then all sub-blocks were integrated into a whole picture.This was followed by the tile pooling,and forthputting logarithmic normal distribution function to fit gray level distribution curve in each sub-block so that gray threshold shall be determined in order to extract cell area.Furthermore,morphology distance was used to recognize and locate adhered ones.So far,global RBC-relevant information collected would support to estimate final index.A series of experiments made up of two parts divided into three steps were carried out.Firstly,the standard particles board made of melamine resin with known size and shape,exhibiting the similar index of refraction to RBC was used as imaging object in the initial experiments,whose size and shape is known,to test algorithm effectivity.Secondly,the algorithms were applied to the same RBC sample with distinct dilution ratio to explore and explain the regular pattern between dilution ratio and estimation effect.Finally,with the determined sample conditions,the replicated experiment was executed to test stability and accuracy of the algorithms.By comparing the experimental results with the distribution status of standard particles under microscope,it can be found that applied algorithm is capable to find best reconstruction distance of each sub-blocks and to extract target objects effectively,which can not only reflect size and shape,but also recognize adhered particle and locate them.Besides,as the dilution ratio increases,each physiological indicator is more consistent with the corresponding clinical result.Simultaneously,the results of replicated experiments show that the accuracy of the three indicators(RBC,MCV,HCT)is higher.,but there are significant fluctuations in that of RDW-SD and RDW-CV.These experimental results reveal that the research group has proposed a statistic and analysis algorithm with potential engineering value for calculating RBC related indicators in blood routine.Benefiting from adaptation of lens-less digital holography in high throughput cell analysis,the holistic indicators well match the clinical results.In future study,the phase information is urgent to be applied so as to increase the accuracy of individual indicators.