CAJ | 학술논문

目的基于乳腺超声图像特征建立BI-RADS评估分类评分系统,旨在为超声医师提供乳腺超声评估分类的客观依据.方法回顾分析2011年1-9月及2012年1月行乳腺超声检查并经病理证实的1 938例女性病例,以2011年病例为模型病例,通过单因素分析,获得各个超声征象诊断乳腺癌的权重,依据各特征的权重给予相应的赋值,建立评估分类评分标准模型,以乳腺影像报告与数据系统第4版为标准,提出BI-RADS评估分类3类、4类、5类及4类亚类的分类评分标准;最后以2012年病例为测试病例,验证该评分标准的有效性.结果纳入24个BI-RADS征象术语,其中18项差异有统计学意义(P＜0.05).依据OR值大小,将征象分为大征象、中征象、小征象3类,各类赋值分别为3分、2分、1分,建立评分模型如下:Score=3*(X1+X2+…+X6)+2*(X7 ++ X8 ++…+X12)+(X9+ K10+…+X18),其中X1-X18依次为Cooper韧带受牵拉、强回声晕、皮肤增厚、腋窝淋巴结、结构扭曲、边缘毛刺征、不规则形、模糊、水肿、边界不完整、阻力指数≥0.7、肿块内微钙化、成角、流速≥20 cm/s、后方回声衰减、方位不平行、微小分叶、血流分级为Ⅱ级或Ⅲ级.依据BI-RADS标准,假设＜2分为3类,2≤Score＜6分为4A类,6≤Score＜10分为4B类,10≤Score＜ 16分为4C类,≥16分为5类.模型病例与测试病例3类、4A类、4B类、4C类及5类阳性预测值分别为1.5％/1.4％、6.9％/3.4％、22.1％/21.1％、62.5％/69.4％、96.1％/92.7％.结论数据评分系统能够客观评估乳腺肿块BI-RADS标准分类,便于医生间交流和预测乳腺癌危险度.
목적 기우유선초성도상특정건립BI-RADS평고분류평분계통,지재위초성의사제공유선초성평고분류적객관의거.방법 회고분석2011년1-9월급2012년1월행유선초성검사병경병리증실적1 938례녀성병례,이2011년병례위모형병례,통과단인소분석,획득각개초성정상진단유선암적권중,의거각특정적권중급여상응적부치,건립평고분류평분표준모형,이유선영상보고여수거계통제4판위표준,제출BI-RADS평고분류3류、4류、5류급4류아류적분류평분표준;최후이2012년병례위측시병례,험증해평분표준적유효성.결과 납입24개BI-RADS정상술어,기중18항차이유통계학의의(P＜0.05).의거OR치대소,장정상분위대정상、중정상、소정상3류,각류부치분별위3분、2분、1분,건립평분모형여하:Score=3*(X1+X2+…+X6)+2*(X7 ++ X8 ++…+X12)+(X9+ K10+…+X18),기중X1-X18의차위Cooper인대수견랍、강회성훈、피부증후、액와림파결、결구뉴곡、변연모자정、불규칙형、모호、수종、변계불완정、조력지수≥0.7、종괴내미개화、성각、류속≥20 cm/s、후방회성쇠감、방위불평행、미소분협、혈류분급위Ⅱ급혹Ⅲ급.의거BI-RADS표준,가설＜2분위3류,2≤Score＜6분위4A류,6≤Score＜10분위4B류,10≤Score＜ 16분위4C류,≥16분위5류.모형병례여측시병례3류、4A류、4B류、4C류급5류양성예측치분별위1.5％/1.4％、6.9％/3.4％、22.1％/21.1％、62.5％/69.4％、96.1％/92.7％.결론 수거평분계통능구객관평고유선종괴BI-RADS표준분류,편우의생간교류화예측유선암위험도.
Objective To set up an objective criterion for BI-RADS assessment.Methods The breast sonograms of 1 938 female cases were retrospectively studied which had been confirmed by pathology from January to September 2011 and 2012 January.1 660 cases during 2011 were model cases,and the importance of each single feature in distinguishing between benign and malignant was obtained with Chi square and OR value.A detailed criterion of BI-RADS assessment category was set up based on US-BI-RADS.Results Eighteen of 24 lexicons were statistically significant distinguished between benign and malignant breast masses(P ＜ 0.05).The lexicons were divided into major suspicious signs,middle ones and minor ones assigned 3 points,2 points and 1 point,and a scoring model was established as follows:Score =3 * (X1 + X2 + … + X6) + 2 * (X7 + X8 + … + X12) + (X9 + X10 + … + X18).Based on BI-RADS,the positive predictive value of model cases was 1.5％,6.9％,22.1％,62.5％,96.1％ followed by category 3,4A,4B,4C and 5,and it was 1.4％,3.4％,21.1％,69.4％,92.7％ in test cases.Conclusions The scoring model could be useful for BI-RADS final assessment more objectively,and could make it more convenient to predict the risk of breast cancer.