前列腺影像报告和数据系统综合评分方法的探讨

The integrated scoring method for prostate imaging reporting and data system

目的探讨前列腺影像报告和数据系统（PI-RADS）最合理的综合评分方法及诊断临界值。方法 99例患者因前列腺疾病行T2WI、DWI及动态增强扫描（DCE-MRI）,共118个前列腺周围带病灶,其中良性病变55个,前列腺癌63个。以PI-RADS标准分别对病灶T2WI、DWI及DCE-MRI序列评分,计算总评分、校正总评分,以Logistic回归建立预测概率模型,比较三者差异。以总评分8、9、10分为诊断临界值计算敏感度、特异度,并分别计算总评分8-9分的病灶前列腺癌在其中的构成比。结果总评分、校正总评分及Logistic模型三者诊断前列腺癌ROC曲线下面积两两比较差异均无统计学意义（P〉0.05）。以总评分8分为诊断临界值,诊断前列腺癌的敏感度为100%,特异度为66.7%,与9、10分诊断临界值的敏感度相比较,差异均有统计学意义（P〈0.05）。8-9、8和9分的病灶,前列腺癌占该分总病灶数的构成比分别为32.0%、37.5%和22.2%,差异均无统计学意义（P〉0.05）。结论对初诊的前列腺周围带病变患者,PI-RADS综合评分不需要对序列或序列中的极值进行权重,各序列评分相加并以8作诊断临界值是最合理的方法。

Objective To explore the most reasonable integrated scoring method and the diagnostic critical value for prostate imaging reporting and data system（ PI-RADS）. Methods The T2 WI,DWI and dynamic contrast-enhanced MRI（ DCE-MRI） were used to examine 99 patients because of prostate disease. There were 118 lesions in prostate peripheral zone including 55 benign lesions and 63 prostate cancers. With the criterion of PI-RADS,a sequence score was applied to lesions discovered by T2 WI,DWI and DCEMRI,respectively. Then,the total scores were calculated and further weighted. Prediction probability model was established by using logistic regression and differences were compared between the three types of scores. The total scores 8,9 and 10 were regarded as diagnostic critical values to calculate the sensitivity and specificity. The constituent ratios of lesions in prostate cancer in the total score 8 and 9were also calculated. Results There was no statistical difference in the areas under ROC curves of the total score,the corrected total score and the logistic regression model in pairwise comparisons. The sensitivity and specificity of diagnosis of prostate cancer was 100%and 66. 7%,respectively when total score 8 was divided into diagnostic threshold. There was a significant difference in the sensitivity compared to total scores 9 or 10 as diagnostic thresholds（ P〈0. 05）. Lesions for scores between 8 and 9 or score 8 or score 9,the constituent ratios of prostate cancer were 32. 0%,37. 5% and 22. 2% respectively. The differences had no statistical significance. Conclusion For newly diagnosed patients with lesions in prostate peripheral zones,PI-RADS integrated scoring does not need to weight the sequences or extremums in the sequences. The most reasonable way is adding scores of all sequences and regarding 8 as the diagnostic critical value.