Evaluation of performance of the machine vision based on microscope / 国际检验医学杂志

ABSTRACT

Objective To evaluate performance of the AVE‐766 automated urine sediment analyzer(AVE‐766) based on the ma‐chine vision for detecting erythrocytes(RBCs) ,leukocytes(WBCs) ,epithelial cells(ECs)and CASTs in urine specimen .Methods The within‐run variable coefficients(CVs) ,linearities and carryover rates for RBC ,WBC ,EC ,and CAST in urine specimens detected by using the AVE‐766 were analyzed ,the results of RBC ,WBC ,EC ,and CAST count in urine specimens detected by using AVE‐766 and Fast‐Read102 counting plate(Fast‐Read102) were compared .Results The within‐run CVs for RBC ,WBC ,EC ,and CAST detected by using AVE‐766 and Fast‐Read102 were increased by decreases of concentration of urine sediment .Good linearities (R2 >0 .97 ,P<0 .05) were observed for RBC(in the range of 60 -1 255/μL) ,WBC(in the range of 68 -2 718/μL) ,EC(in the range of 28-296/μL) and CAST(in the range of 5-86/μL) detected by using the AVE‐766 .The carryover rates for RBC ,WBC , EC ,and CAST detected by using AVE‐766 was 0 .9% or less .The values detected by using AVE‐766 were correlated well with those detected by using Fast‐Read102 for RBC ,WBC ,and EC in urine specimens(0 .67< r<0 .75) .However ,for CAST ,the values detected by using AVE‐766 were poorly correlated with those detected by using Fast‐Read102(r=0 .183) .There were statistically significant differences between manual and automated urinalysis for RBC ,WBC ,EC ,and CAST in urine specimens(P<0 .05) .Con‐clusion The AVE‐766 could not take over microscopic examination ,only is suitable for the first screening to detect RBC ,WBC , EC ,and CAST in urine specimen .