首页|MRI联合前列腺健康指数对前列腺特异性抗原阳性前列腺病灶良恶性具有较高诊断价值

MRI联合前列腺健康指数对前列腺特异性抗原阳性前列腺病灶良恶性具有较高诊断价值

扫码查看
目的 分析MRI定性及定量参数联合前列腺健康指数(PHI)鉴别前列腺特异性抗原(PSA)阳性前列腺病灶良恶性的价值。方法 选取华北理工大学附属医院2022年4月~2024年6月收治的行前列腺MRI检查的346例PSA阳性患者,前列腺影像报告数据系统(PI-RADS)3~5分。以穿刺病理结果作为金标准,分为良性组(良性前列腺病变,n=118)和恶性组(前列腺癌,n=228)。比较2组的PI-RADS评分、表观扩散系数(ADC)、慢速ADC(D)、快速ADC(D*)、灌注相关体积分数(f)和PHI。采用Logistic回归和决策树分类法构建上述参数的模型,用ROC曲线、校准曲线和决策分析曲线比较各模型的诊断效能。结果 恶性组的PI-RADS 5分占比、前2肽PSA和PHI均高于良性组,ADC、D、D*和f均低于良性组(P<0。05)。模型4B(决策树分类模型,纳入PI-RADS、PHI、ADC、D、D*和f,最终保留PHI、ADC、D和D*)的ROC曲线下面积与模型3B(决策树分类模型,纳入PI-RADS、ADC、D、D*和f)差异无统计学意义(P>0。05),但高于其他模型(模型1A~4A、1B、2B,P<0。05)。模型3B和模型4B的校准曲线与对角线重合度高,其他模型的校准曲线与对角线重合度偏低。模型4B的净收益在全阈值范围内均高于其他模型。结论 由PHI、ADC、D和D*构建的决策树分类模型在鉴别PSA阳性且PI-RADS 3~5分的前列腺病灶良恶性方面表现出较高的区分度、校准度和临床应用价值。
MRI combined with the prostate health index has high diagnostic value in differentiating benign and malignant prostate-specific antigen-positive prostate lesions
Objective To evaluate the effectiveness of integrating MRI qualitative and quantitative parameters with the prostate health index(PHI)to differentiate between benign and malignant prostate lesions in patients with elevated prostate-specific antigen(PSA)levels.Methods The study involved 346 patients with elevated PSA levels who underwent prostate MRI at the North China University of Science and Technology Affiliated Hospital from April 2022 to June 2024.These patients had prostate imaging reporting and data system(PI-RADS)scores ranging from 3 to 5.Based on biopsy pathology results,which served as the gold standard,patients were categorized into benign group(n=118)and malignant group(n=228).The PI-RADS score,apparent diffusion coefficient(ADC),slow ADC(D),fast ADC(D*),perfusion-related volume fraction(f),and PHI between these groups were compared.Logistic regression and decision tree classification were employed to develop models using these parameters,and the diagnostic performance of each model was assessed through ROC curves,calibration curves and decision analysis curves.Results The malignant group exhibited higher PI-RADS 5 proportion,pro-2 PSA,and PHI,while the values of ADC,D,D*,and f were lower compared to the benign group(P<0.05).The area under the ROC curve for model 4B,a decision tree classification model initially including PI-RADS,PHI,ADC,D,D*,and f,but ultimately retaining PHI,ADC,D,and D*,was comparable to model 3B,which incorporated PI-RADS,ADC,D,D*,f(P>0.05).However,model 4B outperformed other models(models 1A-4A,1B,2B)(P<0.05).The calibration curves for models 3B and 4B closely aligned with the diagonal,indicating high calibration accuracy,whereas the other models showed greater deviation.Model 4B demonstrated the highest net benefit across all threshold ranges.Conclusion The decision tree classification model constructed with PHI,ADC,D and D* exhibited superior discriminative ability,calibration accuracy,and clinical utility in distinguishing between benign and malignant prostate lesions in patients with elevated PSA and PI-RADS scores of 3 to 5.

MRIprostate health indexprostate-specific antigenprostate cancermodel

王晶晶、赵新斌、康绍叁、郭笑颜、李亨然

展开 >

华北理工大学附属医院 影像科,河北 唐山 063000

华北理工大学附属医院 泌尿外科,河北 唐山 063000

MRI 前列腺健康指数 前列腺特异性抗原 前列腺癌 模型

2024

分子影像学杂志
南方医科大学

分子影像学杂志

CSTPCD
ISSN:1674-4500
年,卷(期):2024.47(12)