Rosvoll R V, Mengason A P, Smith L, Patel H J, Maynard J, Connor F
Am J Clin Pathol. 1979 Jun;71(6):695-703. doi: 10.1093/ajcp/71.6.695.
The authors compared referee (senior author) microscopic counts, microscopic counts by several technologists, and counts obtained with two pattern-recognition leukocyte classifiers, (1) Larc and (2) Hematrak, and a cytochemical automated method for leukocyte counting, (3) Hermalog D, using samples from (1) a random patient population, (2) a selected abnormal patient population, and (3) healthy individuals. All instruments showed good accuracy and flagged abnormal results for review. Variability in pattern-recognition counts was found to be due mainly to the distribution of the cells on prepared blood smears. The Larc classifier was found to be very sensitive to minor alterations in the cytoplasm or nucleus, and rejected a number of slides. The Hemalog D showed the greatest precision. The method of preparation of slides would be a major decision factor in selecting one pattern-recognition instrument over another.
作者比较了审阅者（资深作者）的显微镜计数、几位技术人员的显微镜计数，以及使用两种模式识别白细胞分类器（1）Larc和（2）Hematrak获得的计数，还有一种用于白细胞计数的细胞化学自动方法（3）Hermalog D，使用了来自（1）随机患者群体、（2）选定的异常患者群体和（3）健康个体的样本。所有仪器都显示出良好的准确性，并标记出异常结果以供复查。发现模式识别计数的变异性主要归因于制备血涂片上细胞的分布。发现Larc分类器对细胞质或细胞核的微小变化非常敏感，并拒收了一些玻片。Hemalog D显示出最高的精密度。在选择一种模式识别仪器而非另一种时，玻片的制备方法将是一个主要的决定因素。
