论文部分内容阅读
目的评价通过T2加权、扩散加权和动态对比增强多参数MR成像获得的大量参数用以计算机辅助诊断(CAD)前列腺癌并评价其侵袭性的潜在应用价值。材料与方法本研究经机构审查委员会批准并符合HIPAA标准。通过直肠内线圈获得48例前列腺癌病人(年龄44～73岁,中位数62.5岁)接受前列腺切除术前多参数MR影像。一名放射医师和一名病理医师以组织学与MR表现相对照鉴定104个感兴趣区(ROI)(614个癌性ROI,43个正常ROI)。通过线性判别性分析法单独及联合分析第10百分位数、平均表观扩散系数(ADC)值、T2加权信号强度偏态直方图、Tofts Ktrans,以受试者操作特征曲线分析曲线下面积(AUC)作为区分癌性灶与正常灶的参考值。计算癌性灶Gleason分级和影像特征间Spearman秩相关系数(r)。结果 AUC(最大似然估计±标准误)值鉴别前列腺癌与正常组织在第10百分位数ADC、平均ADC、T2加权偏态及Ktrans分别为0.92±0.03、0.8±0.03、0.86±0.04和0.69±0.04。联合第10百分位数ADC、平均ADC和T2加权偏态在鉴别前列腺癌与正常组织的AUC值为0.95±0.02。Gleason评分(GS)与第10百分位数ADC(ρ=20.34,P=0.008)、平均ADC(ρ=20.30,P=0.02)和Ktrans(ρ=0.38,P=0.004)中度相关。结论联合第10百分位数ADC、平均ADC和T2加权偏态应用CAD有望鉴别前列腺癌与正常组织。ADC影像特征和Ktrans与GS中度相关。
Objective To evaluate the potential value of using T2-weighted, diffusion-weighted, and dynamic contrast-enhanced multiparametric MR imaging for computer-assisted diagnosis of prostate cancer (CAD) and evaluate its invasiveness. Materials and Methods The study was approved by the Institutional Review Board and complied with HIPAA standards. Forty-eight prostate cancer patients (aged 44-73 years, median 62.5 years) received multi-parametric MR imaging of prostatectomy through the intrarectal coil. A radiologist and a pathologist identified 104 ROIs (614 cancerous ROI, 43 normal ROI) histologically and MR findings. The tenth percentile, mean apparent diffusion coefficient (ADC) value, histogram of T2-weighted signal intensity skewness, and Tofts Ktrans were analyzed individually and jointly by linear discriminant analysis. The area under the curve (AUC) as a reference value to distinguish between cancerous lesions and normal lesions. Spearman rank correlation coefficients (r) were calculated between the Gleason grading and image features of the cancerous lesions. Results The AUC (maximum likelihood estimation ± standard error) value discriminated between prostate cancer and normal tissue in the 10th percentile ADC, mean ADC, T2-weighted skewness and Ktrans were 0.92 ± 0.03,0.8 ± 0.03,0.86 ± 0.04 and 0.69 ± 0.04. In combination with the 10th percentile ADC, mean ADC and T2-weighted skewness, the AUC of differentiating between prostate cancer and normal tissue was 0.95 ± 0.02. The Gleason score (GS) was moderately correlated with the 10th percentile ADC (ρ = 20.34, P = 0.008), mean ADC (ρ = 20.30, P = 0.02) and Ktrans (ρ = 0.38, P = 0.004). Conclusions CAD in combination with the 10th percentile ADC, mean ADC and T2-weighted skewedness is expected to distinguish between prostate cancer and normal tissue. ADC image features and Ktrans are moderately related to GS.