Centre d'Imagerie de Fribourg, Groupe 3R, Rue du Centre 10, 1752 Villars-sur-Glâne, Switzerland.
Institut de Radiologie de Sion, Groupe 3R, Rue du Scex 2, Sion, Switzerland.
Phys Med. 2021 Mar;83:64-71. doi: 10.1016/j.ejmp.2021.02.010. Epub 2021 Mar 11.
Evaluation of a deep learning approach for the detection of meniscal tears and their characterization (presence/absence of migrated meniscal fragment).
A large annotated adult knee MRI database was built combining medical expertise of radiologists and data scientists' tools. Coronal and sagittal proton density fat suppressed-weighted images of 11,353 knee MRI examinations (10,401 individual patients) paired with their standardized structured reports were retrospectively collected. After database curation, deep learning models were trained and validated on a subset of 8058 examinations. Algorithm performance was evaluated on a test set of 299 examinations reviewed by 5 musculoskeletal specialists and compared to general radiologists' reports. External validation was performed using the publicly available MRNet database. Receiver Operating Characteristic (ROC) curves results and Area Under the Curve (AUC) values were obtained on internal and external databases.
A combined architecture of meniscal localization and lesion classification 3D convolutional neural networks reached AUC values of 0.93 (95% CI 0.82, 0.95) for medial and 0.84 (95% CI 0.78, 0.89) for lateral meniscal tear detection, and 0.91 (95% CI 0.87, 0.94) for medial and 0.95 (95% CI 0.92, 0.97) for lateral meniscal tear migration detection. External validation of the combined medial and lateral meniscal tear detection models resulted in an AUC of 0.83 (95% CI 0.75, 0.90) without further training and 0.89 (95% CI 0.82, 0.95) with fine tuning.
Our deep learning algorithm demonstrated high performance in knee menisci lesion detection and characterization, validated on an external database.
评估一种用于检测半月板撕裂及其特征（存在/不存在半月板碎片迁移）的深度学习方法。
结合放射科医生的专业知识和数据科学家的工具，构建了一个大型注释成人膝关节 MRI 数据库。回顾性收集了 11353 次膝关节 MRI 检查（10401 名个体患者）的冠状位和矢状位质子密度脂肪抑制加权图像，并与其标准化的结构化报告配对。数据库整理后，在 8058 次检查的子集中训练和验证深度学习模型。由 5 名肌肉骨骼专家审查的 299 次检查测试集上评估算法性能，并与普通放射科医生的报告进行比较。使用公开可用的 MRNet 数据库进行外部验证。在内部和外部数据库上获得了接收器操作特征（ROC）曲线结果和曲线下面积（AUC）值。
一种半月板定位和病变分类的联合 3D 卷积神经网络架构，内侧半月板撕裂检测的 AUC 值为 0.93（95%CI 0.82，0.95），外侧半月板撕裂检测的 AUC 值为 0.84（95%CI 0.78，0.89），内侧半月板撕裂迁移检测的 AUC 值为 0.91（95%CI 0.87，0.94），外侧半月板撕裂迁移检测的 AUC 值为 0.95（95%CI 0.92，0.97）。未进一步训练的联合内侧和外侧半月板撕裂检测模型的外部验证结果为 AUC 为 0.83（95%CI 0.75，0.90），微调后的 AUC 为 0.89（95%CI 0.82，0.95）。
我们的深度学习算法在膝关节半月板病变检测和特征描述方面表现出了很高的性能，在外部数据库上得到了验证。
