论文部分内容阅读
目的采用一组低发病率队列,评价一种高级“预先读片者”工作流程的计算机辅助诊断的CT结肠成像(CTC)对≥6mm结直肠腺瘤的显示能力。方法一名放射科医师采用“预先读片者”工作流程的计算机辅助诊断技术,回顾性分析616例经结肠镜证实的CT结肠成像数据。计算机辅助诊断自动生成病变的影像,包括全部数据的二维(2D)和三维(3D)影像、病灶部位的交互式3D影像和2D多平面重组影像。每例病人的数据首先进行计算机辅助诊断,而后采用快速2D浏览方式评价结肠区的情况。在每例病人、每个息肉和每个腺瘤水平上对≥6mm的病灶计算其敏感性。统计学方法采用Fisher确切概率检验和McNemar检验。结果根据参考标准,在91例病人(91/616)中共检出131个≥6mm的息肉(92个腺瘤,39个非腺瘤)和2个癌灶。采用“预先读片者”工作流程的计算机辅助诊断,放射医生检测出了所有≥10mm的腺瘤(34/34)和癌灶。每例病人、每个息肉水平上对于≥6mm病灶的敏感度分别为84.3%(75/89)和83.2%(109/131),对于腺瘤的敏感度分别为89.1%(57/64)和85.9%(79/92)。总体的特异度为95.6%(504/527)。对每例病人的平均读片时间为3.1min。结论采用基于影像工作站的“预先读片者”工作流程的计算机辅助诊断算法的CT结肠成像,能够在较低发病率人群中明显缩短读片时间情况下准确发现结直肠腺瘤。
Objective To evaluate the ability of computer-assisted diagnostic CT colonography (CTC) to display ≥6 mm colorectal adenomas using a low-prevalence cohort of high-level prediscovery workers. Methods A radiologist used a computer-assisted diagnostic technique of a “preliminary reader” workflow to retrospectively analyze 616 colon colonoscopy-confirmed CT colonography data. Computer-aided diagnosis automatically generates images of lesions, including two-dimensional (2D) and three-dimensional (3D) images of all data, interactive 3D images of lesion sites, and 2D multiplanar reconstructed images. Each patient’s data was first computer-aided diagnosed, and then the condition of the colon was evaluated using rapid 2D viewing. Sensitivity was calculated for lesions ≥ 6 mm at each patient, at each polyp, and at each adenoma level. Statistical methods used Fisher’s exact test and McNemar test. RESULTS: According to the reference standard, a total of 131 ≥6 mm polyps (92 adenomas, 39 non-adenomas) and 2 foci were detected in 91 patients (91/616). Using a computer-aided diagnosis of the “pre-reader” workflow, the radiologist detected all adenomas (34/34) and foci that were > 10 mm. The sensitivity of ≥6 mm lesions was 84.3% (75/89) and 83.2% (109/131) respectively for each patient and each polyp. The sensitivity for adenomas was 89.1% (57/64) and 85.9% (79/92). The overall specificity was 95.6% (504/527). The average reading time for each patient was 3.1 min. Conclusions CT colonography based on computer-aided diagnosis algorithm based on image workstation’s “pre-reader” workflow can significantly reduce colorectal adenomas in the low-incidence population.