Quantitative analysis of the ACL and PCL using T1rho and T2 relaxation time mapping: an exploratory, cross-sectional comparison between OA and healthy control knees.

BACKGROUND: Quantitative magnetic resonance imaging (MRI) methods such as T1rho and T2 mapping are sensitive to changes in tissue composition, however their use in cruciate ligament assessment has been limited to studies of asymptomatic populations or patients with posterior cruciate ligament tears only. The aim of this preliminary study was to compare T1rho and T2 relaxation times of the anterior cruciate ligament (ACL) and posterior cruciate ligament (PCL) between subjects with mild-to-moderate knee osteoarthritis (OA) and healthy controls. METHODS: A single knee of 15 patients with mild-to-moderate knee OA (Kellgren-Lawrence grades 2-3) and of 6 age-matched controls was imaged using a 3.0 T MRI. Three-dimensional (3D) fat-saturated spoiled gradient recalled-echo images were acquired for morphological assessment and T1ρ- and T2-prepared pseudo-steady-state 3D fast spin echo images for compositional assessment of the cruciate ligaments. Manual segmentation of whole ACL and PCL, as well as proximal / middle / distal thirds of both ligaments was carried out by two readers using ITK-SNAP and mean relaxation times were recorded. Variation between thirds of the ligament were assessed using repeated measures ANOVAs and differences in these variations between groups using a Kruskal-Wallis test. Inter- and intra-rater reliability were assessed using intraclass correlation coefficients (ICCs). RESULTS: In OA knees, both T1rho and T2 values were significantly higher in the distal ACL when compared to the rest of the ligament with the greatest differences in T1rho (e.g. distal mean = 54.5 ms, proximal = 47.0 ms, p < 0.001). The variation of T2 values within the PCL was lower in OA knees (OA: distal vs middle vs proximal mean = 28.5 ms vs 29.1 ms vs 28.7 ms, p = 0.748; Control: distal vs middle vs proximal mean = 26.4 ms vs 32.7 ms vs 33.3 ms, p = 0.009). ICCs were excellent for the majority of variables. CONCLUSION: T1rho and T2 mapping of the cruciate ligaments is feasible and reliable. Changes within ligaments associated with OA may not be homogeneous. This study is an important step forward in developing a non-invasive, radiological biomarker to assess the ligaments in diseased human populations in-vivo.

The optimisation of deep neural networks for segmenting multiple knee joint tissues from MRIs.

Automated semantic segmentation of multiple knee joint tissues is desirable to allow faster and more reliable analysis of large datasets and to enable further downstream processing e.g. automated diagnosis. In this work, we evaluate the use of conditional Generative Adversarial Networks (cGANs) as a robust and potentially improved method for semantic segmentation compared to other extensively used convolutional neural network, such as the U-Net. As cGANs have not yet been widely explored for semantic medical image segmentation, we analysed the effect of training with different objective functions and discriminator receptive field sizes on the segmentation performance of the cGAN. Additionally, we evaluated the possibility of using transfer learning to improve the segmentation accuracy. The networks were trained on i) the SKI10 dataset which comes from the MICCAI grand challenge "Segmentation of Knee Images 2010″, ii) the OAI ZIB dataset containing femoral and tibial bone and cartilage segmentations of the Osteoarthritis Initiative cohort and iii) a small locally acquired dataset (Advanced MRI of Osteoarthritis (AMROA) study) consisting of 3D fat-saturated spoiled gradient recalled-echo knee MRIs with manual segmentations of the femoral, tibial and patellar bone and cartilage, as well as the cruciate ligaments and selected peri-articular muscles. The Sørensen-Dice Similarity Coefficient (DSC), volumetric overlap error (VOE) and average surface distance (ASD) were calculated for segmentation performance evaluation. DSC ≥ 0.95 were achieved for all segmented bone structures, DSC ≥ 0.83 for cartilage and muscle tissues and DSC of ≈0.66 were achieved for cruciate ligament segmentations with both cGAN and U-Net on the in-house AMROA dataset. Reducing the receptive field size of the cGAN discriminator network improved the networks segmentation performance and resulted in segmentation accuracies equivalent to those of the U-Net. Pretraining not only increased segmentation accuracy of a few knee joint tissues of the fine-tuned dataset, but also increased the network's capacity to preserve segmentation capabilities for the pretrained dataset. cGAN machine learning can generate automated semantic maps of multiple tissues within the knee joint which could increase the accuracy and efficiency for evaluating joint health.