Harkin P J, Kelly S A, Jack A S
University Department of Pathology, University of Leeds.
J Clin Pathol. 1990 Oct;43(10):840-3. doi: 10.1136/jcp.43.10.840.
With increasing numbers of reagents the problem of selecting appropriate antibodies to solve problems in the diagnosis of lymphoma is becoming more complex. One approach is to use a computer program to optimise the selection process. Such a program was devised, incorporating data from an extensive literature search. When presented with a differential diagnosis it selects the most appropriate antibody panel and when given the results evaluates the relative likelihood of each possible diagnosis. In a retrospective study 81% of the tests used had been non-discriminatory, but using the results of the remaining 19% of the tests, the computer was able to select the "correct" diagnosis with a high degree of certainty. The development and use of this system illustrated several problems in the application of computer assisted diagnostic techniques in histopathology. These problems include incomplete data and lack of understanding of the process of histopathological diagnosis.
随着试剂数量的增加,选择合适的抗体来解决淋巴瘤诊断问题变得越来越复杂。一种方法是使用计算机程序来优化选择过程。设计了这样一个程序,纳入了广泛文献检索的数据。当给出鉴别诊断时,它会选择最合适的抗体组合,并且在给出结果时评估每种可能诊断的相对可能性。在一项回顾性研究中,81%的检测没有鉴别意义,但利用其余19%检测的结果,计算机能够高度确定地选择“正确”诊断。该系统的开发和使用说明了计算机辅助诊断技术在组织病理学应用中的几个问题。这些问题包括数据不完整以及对组织病理学诊断过程缺乏了解。