计算机辅助检测在乳腺癌X线检测中的价值

The Value of Computer-Aided Detection for the Diagnosis of Breast Cancer

  • 摘要: 目的: 评价计算机辅助检测(computer-aided detection,CAD)软件在全数字化乳腺X线摄影中对乳腺癌检出的价值。 方法: 收集226例2004年12月~2005年5月期间经病理及手术证实乳腺癌的全数字化乳腺X摄影图像,采用CAD软件独立检测、初级放射科医生应用CAD软件前、后分别检测及1名有经验放射科医生独立检测四种方法分别进行检测。采用χ2检验,进行统计学分析。 结果: 1)CAD软件和初级放射科医生独立对乳腺癌、肿块(masses,MA)及微钙化(microcalcifications,MC)的检出率差别均无统计学意义(P=0.3061,P=0.5063,P=0.4215)。2)CAD独立检测与有经验放射科医生独立检测对乳腺癌、MA及MC的检出率差别均有明显统计学差异(P=0.0000,P=0.0000,P=0.0002)。3)联合应用CAD软件后,初级放射科医生对乳腺癌、MA及MC的检出率均明显提高,差别有统计学意义(P=0.0011,P=0.0000,P=0.0001),且MC的检出率提高比MA检出率提高明显(MC,由80.23%提高到98.84%;MA,由85.65%提高到96.76%)。4)初级放射科医生应用CAD软件后与1名有经验放射科医生独立对乳腺癌、MA及MC检出率差别无统计学意义(P=0.1536,P=0.0921,P=0.3159)。 结论: CAD软件可以协助初级放射科医生提高乳腺癌检出率。

     

    Abstract: Objective: To assess the value of a computer-aided detection (CAD) software in detection of breast cancer. Methods: A total of 226 patients with breast cancer conformed histologically from December 2004 to May 2005 were collected.All of them underwent full-field digital mammography (FFDM). Four methods were used in detecting breast cancer: CAD software alone, inexperienced radiologists with and without the CAD software and experienced radiologists alone. Chi-square test was applied to analyze the data. Results: (1) No statistical significance was found in detecting breast cancer,masses (MA) and microcalcifications (MC) between CAD software and the inexperienced radiologists(P=0.3061, P=0.5063,P=0.4215). (2) There were significant differences in the detection of breast cancer, MA and MC between the CAD and the experienced radiologists (P=0.0000, P=0.0000, P=0.0002). (3) The detection rates of breast cancer, MA, MC were signifi-cantly increased after implementation, respectively (P=0.001 1, P=0.0000, P=0.0001). The detection rates of MC were in-creased more than that of MA (MC, from 80.23% to 98.84%; MA, from 85.65% to 96.76%). (4) There were no significant differences in the detection of MA and MC between the inexperienced radiologist with the CAD software and the experi-enced radiologists(P=0.153 6, P=0.092 1, P=0.315 9). Conclusion: CAD software showed a potential value to assist the in-experienced radiologists in detecting breast cancer.

     

/

返回文章
返回