Miller Fern S, Nagel Lynn E, Kenny-Moynihan Mary B
Metropolitan Pathologists, PC, 7444 W. Alaska Drive, Lakewood, CO 80226, USA.
Diagn Cytopathol. 2007 Apr;35(4):213-7. doi: 10.1002/dc.20627.
The Papanicolaou test has proven to be the most effective cancer screening test ever developed. However, with a declining number of skilled cytotechnologists, there is an increased need for computer assistance in cervical cancer screening. The ThinPrep Imaging System (Cytyc Corporation, Marlborough, MA) is a unique system that combines computer imaging technology and human interpretive expertise in the review of ThinPrep Pap test slides. The purpose of this study is to report on the introduction and validation of this technology and present data related to the performance and productivity in our laboratory. Following completion of the ThinPrep Imaging System validation protocol, all imaged ThinPrep Pap test results were tracked and compared with year-2003 manually screened results to identify whether the Imaging System was effective in aiding human interpretive skills. Cases rescreened in the 10% random quality control (QC) program from the negative population that showed abnormal cells consistent with low-grade squamous intraepithelial lesion (LSIL) and above were compared with imaged versus non-imaged cases to establish an estimated laboratory false-negative (F/N) rate. The study compared results of 82,063 manually screened ThinPrep Pap tests in 2003 with 84,473 imaged ThinPrep Pap tests in 2004. Results demonstrated a significant increase in LSIL (37%) and high-grade squamous intraepithelial lesion (HSIL) (42%) detection on the Imager cohort. The F/N rate was reduced by half. The evaluation period after validation of the Imager showed a significant increase in LSIL and HSIL detection with the ThinPrep Imaging System compared to manual screening. These results demonstrate that the Imager has the potential to allow the cytotechnologists to detect more disease and reduce the false-negative rate for the laboratory. Although not evaluated in this study, cytotechnologists reported increased job satisfaction.
巴氏试验已被证明是有史以来最有效的癌症筛查测试。然而，随着熟练的细胞技术人员数量的减少，宫颈癌筛查对计算机辅助的需求日益增加。ThinPrep成像系统（赛迪科公司，马萨诸塞州马尔伯勒）是一种独特的系统，它在ThinPrep巴氏试验玻片的检查中结合了计算机成像技术和人工解读专业知识。本研究的目的是报告该技术的引入和验证情况，并展示与我们实验室的性能和生产率相关的数据。在完成ThinPrep成像系统验证方案后，对所有成像的ThinPrep巴氏试验结果进行跟踪，并与2003年的手动筛查结果进行比较，以确定成像系统在辅助人工解读技能方面是否有效。将在10%随机质量控制（QC）程序中从显示与低级别鳞状上皮内病变（LSIL）及以上一致的异常细胞的阴性人群中重新筛查的病例与成像病例和未成像病例进行比较，以确定实验室估计的假阴性（F/N）率。该研究将2003年82,063次手动筛查的ThinPrep巴氏试验结果与2004年84,473次成像的ThinPrep巴氏试验结果进行了比较。结果表明，成像队列中LSIL（37%）和高级别鳞状上皮内病变（HSIL）（42%）的检测率显著提高。F/N率降低了一半。成像系统验证后的评估期显示，与手动筛查相比，ThinPrep成像系统对LSIL和HSIL的检测率显著提高。这些结果表明，成像系统有可能使细胞技术人员检测到更多疾病，并降低实验室的假阴性率。尽管本研究未进行评估，但细胞技术人员报告工作满意度有所提高。
