全数字化乳腺X线成像计算机辅助检测结果的初步临床评估

Evaluation on the use and results of computer-aided detection for full-field digital manunography

目的初步评估计算机辅助检测(CAD)技术在中国女性乳腺X线检查中的意义。方法沈阳地区某单位内40岁以上女职工934例接受了全数字化平板乳腺X线成像的检查,首先由放射科医师对乳腺图像给予影像学诊断,然后应用R2公司的计算机辅助诊断系统(Image CheckerTM)对所有图像予以辅助诊断的提示,对比分析两者的结果。结果CAD对于基本正常的1734个乳腺图像有799个阳性提示,假阳性提示率为46．1％。对存在结节及微钙化的图像提示符合率(正确率)较高(70．5％及75．0％),对于恶性肿瘤的正确提示率为100％。结论CAD对数字化乳腺X线图像的辅助提示具有较重要的临床应用价值,但仍存在一定的假阳性及假阴性结果。

Objective To evaluate the using and results of computer-aided detection for full-field digital mammography in Chinese women. Methods 934 individuals more than 40 years old from a specific unit in Shenyang city were screened by full-field digital mammography （ Senographe 2000D, GE）. The examination images were diagnosed firstly by the experienced radiologists and then were analyzed by the computer-aided detection system （R2＇ Image Checker^TM）, respectively. The results from these two assays were compared. Results The prompts were generated in 799 out of 1734 normal images, false-positive rate of computer-aided detection （CAD） was 46. 1% in the cases with normal breast image. The correct prompts rate was 70. 5 % in the cases with breast mass, 75. 0 % in the cases with breast microcalcification and 100. 0 % in the cases with breast cancer. Conclusion Although the CAD system has showed clinical values in practice, there are still some false-positive and false-negative results.