Department of Radiology and Nuclear Medicine, Amsterdam UMC, location VUMC, Amsterdam, The Netherlands.
Imaging and Biomarkers, Cancer Center Amsterdam, Amsterdam, The Netherlands.
J Neuroimaging. 2024 Nov-Dec;34(6):673-693. doi: 10.1111/jon.13233. Epub 2024 Sep 19.
To develop and test a decision tree for predicting contrast enhancement quality and shape using precontrast magnetic resonance imaging (MRI) sequences in a large adult-type diffuse glioma cohort.
Preoperative MRI scans (development/optimization/test sets: n = 31/38/303, male = 17/22/189, mean age = 52/59/56.7 years, high-grade glioma = 22/33/249) were retrospectively evaluated, including pre- and postcontrast T1-weighted, T2-weighted, fluid-attenuated inversion recovery, and diffusion-weighted imaging sequences. Enhancement prediction decision tree (EPDT) was developed using development and optimization sets, incorporating four imaging features: necrosis, diffusion restriction, T2 inhomogeneity, and nonenhancing tumor margins. EPDT accuracy was assessed on a test set by three raters of variable experience. True enhancement features (gold standard) were evaluated using pre- and postcontrast T1-weighted images. Statistical analysis used confusion matrices, Cohen's/Fleiss' kappa, and Kendall's W. Significance threshold was p < .05.
Raters 1, 2, and 3 achieved overall accuracies of .86 (95% confidence interval [CI]: .81-.90), .89 (95% CI: .85-.92), and .92 (95% CI: .89-.95), respectively, in predicting enhancement quality (marked, mild, or no enhancement). Regarding shape, defined as the thickness of enhancing margin (solid, rim, or no enhancement), accuracies were .84 (95% CI: .79-.88), .88 (95% CI: .84-.92), and .89 (95% CI: .85-.92). Intrarater intergroup agreement comparing predicted and true enhancement features consistently reached substantial levels (≥.68 [95% CI: .61-.75]). Interrater comparison showed at least moderate agreement (group: ≥.42 [95% CI: .36-.48], pairwise: ≥.61 [95% CI: .50-.72]). Among the imaging features in the EPDT, necrosis assessment displayed the highest intra- and interrater consistency (≥.80 [95% CI: .73-.88]).
The proposed EPDT has high accuracy in predicting enhancement patterns of gliomas irrespective of rater experience.
为了在大型成人弥漫性胶质瘤队列中使用对比前磁共振成像（MRI）序列开发和测试一种预测对比增强质量和形状的决策树。
回顾性评估了术前 MRI 扫描（开发/优化/测试集：n=31/38/303，男性=17/22/189，平均年龄=52/59/56.7 岁，高级别胶质瘤=22/33/249），包括对比前和对比后 T1 加权、T2 加权、液体衰减反转恢复和扩散加权成像序列。使用开发和优化集开发了增强预测决策树（EPDT），其中包括四个成像特征：坏死、弥散受限、T2 不均匀性和非增强肿瘤边界。三位经验不同的评分者在测试集上评估了 EPDT 的准确性。使用预对比和后对比 T1 加权图像评估真实增强特征（金标准）。统计分析使用混淆矩阵、Cohen's/Fleiss' kappa 和 Kendall's W。显著性阈值为 p<0.05。
评分者 1、2 和 3 分别在预测增强质量（标记、轻度或无增强）方面达到了 0.86（95%置信区间[CI]：0.81-0.90）、0.89（95% CI：0.85-0.92）和 0.92（95% CI：0.89-0.95）的总体准确性。关于形状，定义为增强边界的厚度（固体、边缘或无增强），准确性分别为 0.84（95% CI：0.79-0.88）、0.88（95% CI：0.84-0.92）和 0.89（95% CI：0.85-0.92）。在预测和真实增强特征之间的组内组间比较中，一致性达到了较高水平（≥0.68[95% CI：0.61-0.75]）。组间比较显示出至少中等水平的一致性（组间：≥0.42[95% CI：0.36-0.48]，两两比较：≥0.61[95% CI：0.50-0.72]）。在 EPDT 中的成像特征中，坏死评估具有最高的组内和组间一致性（≥0.80[95% CI：0.73-0.88]）。
无论评分者的经验如何，所提出的 EPDT 在预测胶质瘤的增强模式方面都具有很高的准确性。
