Castro-Mateos Isaac, Hua Rui, Pozo Jose M, Lazary Aron, Frangi Alejandro F
Department of Electronic and Electrical Engineering, The University of Sheffield, Mappin Street, Sheffield, South Yorkshire, S1 3JD, UK.
National Center for Spinal Disorders (NCSD), Budapest, Hungary.
Eur Spine J. 2016 Sep;25(9):2721-7. doi: 10.1007/s00586-016-4654-6. Epub 2016 Jul 7.
The primary goal of this article is to achieve an automatic and objective method to compute the Pfirrmann's degeneration grade of intervertebral discs (IVD) from MRI. This grading system is used in the diagnosis and management of patients with low back pain (LBP). In addition, biomechanical models, which are employed to assess the treatment on patients with LBP, require this grading value to compute proper material properties.
T2-weighted MR images of 48 patients were employed in this work. The 240 lumbar IVDs were divided into a training set (140) and a testing set (100). Three experts manually classified the whole set of IVDs using the Pfirrmann's grading system and the ground truth was selected as the most voted value among them. The developed method employs active contour models to delineate the boundaries of the IVD. Subsequently, the classification is achieved using a trained Neural Network (NN) with eight designed features that contain shape and intensity information of the IVDs.
The classification method was evaluated using the testing set, resulting in a mean specificity (95.5 %) and sensitivity (87.3 %) comparable to those of every expert with respect to the ground truth.
Our results show that the automatic method and humans perform equally well in terms of the classification accuracy. However, human annotations have inherent inter- and intra-observer variabilities, which lead to inconsistent assessments. In contrast, the proposed automatic method is objective, being only dependent on the input MRI.
本文的主要目标是实现一种自动且客观的方法，用于从磁共振成像（MRI）计算椎间盘（IVD）的 Pfirrmann 退变分级。该分级系统用于诊断和管理下腰痛（LBP）患者。此外，用于评估 LBP 患者治疗效果的生物力学模型需要此分级值来计算合适的材料属性。
本研究使用了 48 例患者的 T2 加权 MR 图像。240 个腰椎间盘被分为训练集（140 个）和测试集（100 个）。三位专家使用 Pfirrmann 分级系统对整个椎间盘集进行手动分类，并将众数作为真实值。所开发的方法采用主动轮廓模型来勾勒椎间盘的边界。随后，使用具有八个设计特征的训练神经网络（NN）进行分类，这些特征包含椎间盘的形状和强度信息。
使用测试集对分类方法进行评估后发现，其平均特异性（95.5%）和敏感性（87.3%）与每位专家相对于真实值的结果相当。
我们的结果表明，自动方法在分类准确性方面与人类表现相当。然而，人工标注存在观察者间和观察者内的固有变异性，这会导致评估不一致。相比之下，所提出的自动方法是客观的，仅依赖于输入的 MRI。
