RESUMEN
To validate 3D methods for femoral version measurement, we asked: (1) Can a fully automated segmentation of the entire femur and 3D measurement of femoral version using a neck based method and a head-shaft based method be performed? (2) How do automatic 3D-based computed tomography (CT) measurements of femoral version compare to the most commonly used 2D-based measurements utilizing four different landmarks? Retrospective study (May 2017 to June 2018) evaluating 45 symptomatic patients (57 hips, mean age 18.7 ± 5.1 years) undergoing pelvic and femoral CT. Femoral version was assessed using four previously described methods (Lee, Reikeras, Tomczak, and Murphy). Fully-automated segmentation yielded 3D femur models used to measure femoral version via femoral neck- and head-shaft approaches. Mean femoral version with 95% confidence intervals, and intraclass correlation coefficients were calculated, and Bland-Altman analysis was performed. Automatic 3D segmentation was highly accurate, with mean dice coefficients of 0.98 ± 0.03 and 0.97 ± 0.02 for femur/pelvis, respectively. Mean difference between 3D head-shaft- (27.4 ± 16.6°) and 3D neck methods (12.9 ± 13.7°) was 14.5 ± 10.7° (p < 0.001). The 3D neck method was closer to the proximal Lee (-2.4 ± 5.9°, -4.4 to 0.5°, p = 0.009) and Reikeras (2 ± 5.6°, 95% CI: 0.2 to 3.8°, p = 0.03) methods. The 3D head-shaft method was closer to the distal Tomczak (-1.3 ± 7.5°, 95% CI: -3.8 to 1.1°, p = 0.57) and Murphy (1.5 ± 5.4°, -0.3 to 3.3°, p = 0.12) methods. Automatic 3D neck-based-/head-shaft methods yielded femoral version angles comparable to the proximal/distal 2D-based methods, when applying fully-automated segmentations.
RESUMEN
OBJECTIVES: To identify preoperative degenerative features on traction MR arthrography associated with failure after arthroscopic femoroacetabular impingement (FAI) surgery. METHODS: Retrospective study including 102 patients (107 hips) undergoing traction magnetic resonance arthrography (MRA) of the hip at 1.5 T and subsequent hip arthroscopic FAI surgery performed (01/2016 to 02/2020) with complete follow-up. Clinical outcomes were assessed using the International Hip Outcome Tool (iHOT-12) score. Clinical endpoint for failure was defined as an iHOT-12 of < 60 points or conversion to total hip arthroplasty. MR images were assessed by two radiologists for presence of 9 degenerative lesions including osseous, chondrolabral/ligamentum teres lesions. Uni- and multivariate Cox regression analysis was performed to assess the association between MRI findings and failure of FAI surgery. RESULTS: Of the 107 hips, 27 hips (25%) met at least one endpoint at a mean 3.7 ± 0.9 years follow-up. Osteophytic changes of femur or acetabulum (hazard ratio [HR] 2.5-5.0), acetabular cysts (HR 3.4) and extensive cartilage (HR 5.1) and labral damage (HR 5.5) > 2 h on the clockface were univariate risk factors (all p < 0.05) for failure. Three risk factors for failure were identified in multivariate analysis: Acetabular cartilage damage > 2 h on the clockface (HR 3.2, p = 0.01), central femoral osteophyte (HR 3.1, p = 0.02), and femoral cartilage damage with ligamentum teres damage (HR 3.0, p = 0.04). CONCLUSION: Joint damage detected by preoperative traction MRA is associated with failure 4 years following arthroscopic FAI surgery and yields promise in preoperative risk stratification. CLINICAL RELEVANCE STATEMENT: Evaluation of negative predictors on preoperative traction MR arthrography holds the potential to improve risk stratification based on the already present joint degeneration ahead of FAI surgery. KEY POINTS: ⢠Osteophytes, acetabular cysts, and extensive chondrolabral damage are risk factors for failure of FAI surgery. ⢠Extensive acetabular cartilage damage, central femoral osteophytes, and combined femoral cartilage and ligamentum teres damage represent independent negative predictors. ⢠Survival rates following hip arthroscopy progressively decrease with increasing prevalence of these three degenerative findings.
RESUMEN
(1) Background: To evaluate the performance of a deep learning model to automatically segment femoral head necrosis (FHN) based on a standard 2D MRI sequence compared to manual segmentations for 3D quantification of FHN. (2) Methods: Twenty-six patients (thirty hips) with avascular necrosis underwent preoperative MR arthrography including a coronal 2D PD-w sequence and a 3D T1 VIBE sequence. Manual ground truth segmentations of the necrotic and unaffected bone were then performed by an expert reader to train a self-configuring nnU-Net model. Testing of the network performance was performed using a 5-fold cross-validation and Dice coefficients were calculated. In addition, performance across the three segmentations were compared using six parameters: volume of necrosis, volume of unaffected bone, percent of necrotic bone volume, surface of necrotic bone, unaffected femoral head surface, and percent of necrotic femoral head surface area. (3) Results: Comparison between the manual 3D and manual 2D segmentations as well as 2D with the automatic model yielded significant, strong correlations (Rp > 0.9) across all six parameters of necrosis. Dice coefficients between manual- and automated 2D segmentations of necrotic- and unaffected bone were 75 ± 15% and 91 ± 5%, respectively. None of the six parameters of FHN differed between the manual and automated 2D segmentations and showed strong correlations (Rp > 0.9). Necrotic volume and surface area showed significant differences (all p < 0.05) between early and advanced ARCO grading as opposed to the modified Kerboul angle, which was comparable between both groups (p > 0.05). (4) Conclusions: Our deep learning model to automatically segment femoral necrosis based on a routine hip MRI was highly accurate. Coupled with improved quantification for volume and surface area, as opposed to 2D angles, staging and course of treatment can become better tailored to patients with varying degrees of AVN.
RESUMEN
Background: This study aimed to build a deep learning model to automatically segment heterogeneous clinical MRI scans by optimizing a pre-trained model built from a homogeneous research dataset with transfer learning. Methods: Conditional generative adversarial networks pretrained on the Osteoarthritis Initiative MR images was transferred to 30 sets of heterogenous MR images collected from clinical routines. Two trained radiologists manually segmented the 30 sets of clinical MR images for model training, validation and test. The model performance was compared to models trained from scratch with different datasets, as well as two radiologists. A 5-fold cross validation was performed. Results: The transfer learning model obtained an overall averaged Dice coefficient of 0.819, an averaged 95 percentile Hausdorff distance of 1.463 mm, and an averaged average symmetric surface distance of 0.350 mm on the 5 random holdout test sets. A 5-fold cross validation had a mean Dice coefficient of 0.801, mean 95 percentile Hausdorff distance of 1.746 mm, and mean average symmetric surface distance of 0.364 mm. It outperformed other models and performed similarly as the radiologists. Conclusions: A transfer learning model was able to automatically segment knee cartilage, with performance comparable to human, using heterogeneous clinical MR images with a small training data size. In addition, the model proved robust when tested through cross validation and on images from a different vendor. We found it feasible to perform fully automated cartilage segmentation of clinical knee MR images, which would facilitate the clinical application of quantitative MRI techniques and other prediction models for improved patient treatment planning.