Center for Medical Device Evaluation, CFDA, Beijing, China, 100044.
Department of Radiology, the First Affiliated Hospital with Nanjing Medical University, 300, Guangzhou Road, Nanjing, Jiangsu Province, China, 210009.
Eur Radiol. 2017 Oct;27(10):4082-4090. doi: 10.1007/s00330-017-4800-5. Epub 2017 Apr 3.
To investigate whether machine learning-based analysis of MR radiomics can help improve the performance PI-RADS v2 in clinically relevant prostate cancer (PCa).
This IRB-approved study included 54 patients with PCa undergoing multi-parametric (mp) MRI before prostatectomy. Imaging analysis was performed on 54 tumours, 47 normal peripheral (PZ) and 48 normal transitional (TZ) zone based on histological-radiological correlation. Mp-MRI was scored via PI-RADS, and quantified by measuring radiomic features. Predictive model was developed using a novel support vector machine trained with: (i) radiomics, (ii) PI-RADS scores, (iii) radiomics and PI-RADS scores. Paired comparison was made via ROC analysis.
For PCa versus normal TZ, the model trained with radiomics had a significantly higher area under the ROC curve (Az) (0.955 [95% CI 0.923-0.976]) than PI-RADS (Az: 0.878 [0.834-0.914], p < 0.001). The Az between them was insignificant for PCa versus PZ (0.972 [0.945-0.988] vs. 0.940 [0.905-0.965], p = 0.097). When radiomics was added, performance of PI-RADS was significantly improved for PCa versus PZ (Az: 0.983 [0.960-0.995]) and PCa versus TZ (Az: 0.968 [0.940-0.985]).
Machine learning analysis of MR radiomics can help improve the performance of PI-RADS in clinically relevant PCa.
• Machine-based analysis of MR radiomics outperformed in TZ cancer against PI-RADS. • Adding MR radiomics significantly improved the performance of PI-RADS. • DKI-derived Dapp and Kapp were two strong markers for the diagnosis of PCa.
研究基于机器学习的 MR 放射组学分析是否有助于提高 PI-RADS v2 在临床相关前列腺癌（PCa）中的性能。
这项经 IRB 批准的研究纳入了 54 例接受前列腺切除术前行多参数（mp）MRI 检查的 PCa 患者。基于组织学-影像学相关性，对 54 个肿瘤、47 个正常外周区（PZ）和 48 个正常移行区（TZ）进行了成像分析。通过 PI-RADS 对 mp-MRI 进行评分，并通过测量放射组学特征进行定量分析。通过使用一种新的支持向量机，利用（i）放射组学、（ii）PI-RADS 评分、（iii）放射组学和 PI-RADS 评分来开发预测模型。通过 ROC 分析进行配对比较。
对于 PCa 与正常 TZ 相比，基于放射组学训练的模型具有显著更高的 ROC 曲线下面积（Az）（0.955 [95%CI 0.923-0.976]），高于 PI-RADS（Az：0.878 [0.834-0.914]，p < 0.001）。对于 PCa 与 PZ 相比，两者之间的 Az 无显著性差异（0.972 [0.945-0.988] vs. 0.940 [0.905-0.965]，p = 0.097）。当加入放射组学后，PI-RADS 的性能对于 PCa 与 PZ（Az：0.983 [0.960-0.995]）和 PCa 与 TZ（Az：0.968 [0.940-0.985]）的比较均有显著提高。
基于 MRI 放射组学的机器学习分析有助于提高 PI-RADS 在临床相关 PCa 中的性能。
• 基于机器学习的 MR 放射组学分析在 TZ 癌中的表现优于 PI-RADS。
• 加入 MR 放射组学显著提高了 PI-RADS 的性能。
• DKI 衍生的 Dapp 和 Kapp 是诊断 PCa 的两个强有力的标志物。
