本文選題：乳腺疾病 + 超聲檢查　； 參考：《安徽醫(yī)科大學(xué)》2017年碩士論文

【摘要】：目的:應(yīng)用Logistic回歸模型探討乳腺影像報告數(shù)據(jù)系統(tǒng)(BI-RADS)中的超聲診斷指標(biāo)在乳腺腫塊良、惡性鑒別中的應(yīng)用價值,基于此建立乳腺超聲影像報告數(shù)據(jù)系統(tǒng)分類(3~5類)評分系統(tǒng),并對各乳腺腫塊評分并分類,旨在為超聲醫(yī)師提供客觀有效的評估分類標(biāo)準(zhǔn)。方法:回顧分析在安徽醫(yī)科大學(xué)第二附屬醫(yī)院行乳腺超聲檢查的401例乳腺腫塊超聲圖像特征,結(jié)合手術(shù)或穿刺活檢病理結(jié)果,以第5版乳腺影像報告與數(shù)據(jù)系統(tǒng)病灶超聲特征并加入年齡因素,建立Logistic回歸模型;依據(jù)回歸模型篩選結(jié)果及各因素的權(quán)重提出BI-RADS 3~5類評分分類標(biāo)準(zhǔn)。另納入243例乳腺腫塊作為測試組,對所有研究病例分別進(jìn)行評分分類,以病理結(jié)果為金標(biāo)準(zhǔn),計算每類的陽性預(yù)測值,與BI-RADS各類別的理論風(fēng)險范圍相比較,觀察兩者的一致性,評估該評分分類系統(tǒng)的診斷價值。結(jié)果:1、多因素回歸分析顯示最后進(jìn)入模型的因素共6個:分別為年齡(≥40歲)、方位(不平行)、形態(tài)(不規(guī)則)、內(nèi)部回聲(不均勻)、邊緣(不光整)、腫塊內(nèi)微鈣化。2、依據(jù)回歸模型篩選結(jié)果及各因素對腫塊良、惡性貢獻(xiàn)率不同,制定評分分類系統(tǒng),BI-RADS 3、4a、4b、4c、5類相對應(yīng)的分值為6分、7~8分、9~15分,16~22分,≥23分。3、模型病例綜合評分的BI-RADS 3~5類的陽性預(yù)測值分別為0%、2.67%、16.18%、90.74%、100%;測試病例綜合評分的BI-RADS 3~5類的陽性預(yù)測值分別為0%、4.17%、21.43%、84.85%、100%;所有研究病例綜合評分的BI-RADS 3~5類的陽性預(yù)測值為0%、3.25%、17.70%、88.51%、100%。4、測試病例ROC曲線下分析,曲線下面積為0.948。以14分作為乳腺腫塊良惡性的Cut-off值(診斷界點),其靈敏度為90.10%、特異度90.14%、約登指數(shù)0.80、準(zhǔn)確率90.12%。結(jié)論:利用多因素建立的乳腺腫塊Logistic回歸模型并以此制定的BI-RADS評分分類系統(tǒng),能夠?qū)θ橄倌[塊BI-RADS系統(tǒng)進(jìn)行客觀的分類,為臨床評價乳腺腫塊的良、惡性提供有效參考依據(jù)。
[Abstract]:Objective: to explore the value of ultrasound diagnostic indexes in breast image reporting data system (BI-RADS) in differentiating benign and malignant breast masses with Logistic regression model. The purpose of this paper is to provide an objective and effective criteria for the evaluation and classification of breast masses. Methods: the ultrasonographic features of 401 cases of breast masses performed in the second affiliated Hospital of Anhui Medical University were analyzed retrospectively, combined with the pathological results of surgical or puncture biopsy. A Logistic regression model was established based on the ultrasonic features of breast image report and data system in the 5th edition and the age factors were added, and the BI-RADS 3 / 5 classification criteria were proposed according to the results of the regression model and the weight of each factor. In addition, 243 cases of breast masses were included as test group. All the cases were graded and classified separately. The positive predictive value of each group was calculated according to the pathological results as gold standard, and compared with BI-RADS, the consistency of the two groups was observed. To evaluate the diagnostic value of the scoring classification system. Results the multivariate regression analysis showed that there were six factors that entered the model: age (鈮，

本文編號：2093745