Comprehensive evaluation of magnetic resonance imaging sequences for signal intensity based assessment of anterior cruciate ligament healing following surgical treatment.

Normalized signal intensity (SI) obtained from magnetic resonance imaging (MRI) has been used to track anterior cruciate ligament (ACL) postoperative remodeling. We aimed to assess the effect of MRI sequence (PD: proton density-weighted; T2: T2-weighted; CISS: constructive interference in steady state) on postoperative changes in healing ACLs/grafts. We hypothesized that CISS is better at detecting longitudinal SI and texture changes of the healing ACL/graft compared to the common clinical sequences (PD and T2). MR images of patients who underwent ACL surgery were evaluated and separated into groups based on surgical procedure (Bridge-Enhanced ACL Repair (BEAR; n = 50) versus ACL reconstruction (ACLR; n = 24)). CISS images showed decreasing SI across all timepoints in both the BEAR and ACLR groups (p < 0.01), PD and T2 images showed decreasing SI in the 6-to-12- and 12-to-24-month postoperative timeframes in the BEAR group (p < 0.02), and PD images additionally showed decreasing SI between 6- and 24-months postoperation in the ACLR group (p = 0.02). CISS images showed texture changes in both the BEAR and ACLR groups, showing increases in energy and decreases in entropy in the 6-to-12- and 6-to-24-month postoperative timeframes in the BEAR group (p < $\lt $ 0.04), and increases in energy, decreases in entropy, and increases in homogeneity between 6 and 24 months postoperation in the ACLR group (p < 0.04). PD images showed increases in energy and decreases in entropy between 6- and 24-months postoperation in the ACLR group (p < 0.008). Finally, CISS was estimated to require a smaller sample size than PD and T2 to detect SI differences related to postoperative remodeling.

LigaNET: A multi-modal deep learning approach to predict the risk of subsequent anterior cruciate ligament injury after surgery.

Anterior cruciate ligament (ACL) injuries are a common cause of soft tissue injuries in young active individuals, leading to a significant risk of premature joint degeneration. Postoperative management of such injuries, in particular returning patients to athletic activities, is a challenge with immediate and long-term implications including the risk of subsequent injury. In this study, we present LigaNET, a multi-modal deep learning pipeline that predicts the risk of subsequent ACL injury following surgical treatment. Postoperative MRIs (n=1,762) obtained longitudinally between 3 to 24 months after ACL surgery from a cohort of 159 patients along with 11 non-imaging outcomes were used to train and test: 1) a 3D CNN to predict subsequent ACL injury from segmented ACLs, 2) a 3D CNN to predict injury from the whole MRI, 3) a logistic regression classifier predict injury from non-imaging data, and 4) a multi-modal pipeline by fusing the predictions of each classifier. The CNN using the segmented ACL achieved an accuracy of 77.6% and AUROC of 0.84, which was significantly better than the CNN using the whole knee MRI (accuracy: 66.6%, AUROC: 0.70; P<.001) and the non-imaging classifier (accuracy: 70.1%, AUROC: 0.75; P=.039). The fusion of all three classifiers resulted in highest classification performance (accuracy: 80.6%, AUROC: 0.89), which was significantly better than each individual classifier (P<.001). The developed multi-modal approach had similar performance in predicting the risk of subsequent ACL injury from any of the imaging sequences (P>.10). Our results demonstrate that a deep learning approach can achieve high performance in identifying patients at high risk of subsequent ACL injury after surgery and may be used in clinical decision making to improve postoperative management (e.g., safe return to sports) of ACL injured patients.

Predicting anterior cruciate ligament failure load with T2* relaxometry and machine learning as a prospective imaging biomarker for revision surgery.

Non-invasive methods to document healing anterior cruciate ligament (ACL) structural properties could potentially identify patients at risk for revision surgery. The objective was to evaluate machine learning models to predict ACL failure load from magnetic resonance images (MRI) and to determine if those predictions were related to revision surgery incidence. It was hypothesized that the optimal model would demonstrate a lower mean absolute error (MAE) than the benchmark linear regression model, and that patients with a lower estimated failure load would have higher revision incidence 2 years post-surgery. Support vector machine, random forest, AdaBoost, XGBoost, and linear regression models were trained using MRI T2* relaxometry and ACL tensile testing data from minipigs (n = 65). The lowest MAE model was used to estimate ACL failure load for surgical patients at 9 months post-surgery (n = 46) and dichotomized into low and high score groups via Youden's J statistic to compare revision incidence. Significance was set at alpha = 0.05. The random forest model decreased the failure load MAE by 55% (Wilcoxon signed-rank test: p = 0.01) versus the benchmark. The low score group had a higher revision incidence (21% vs. 5%; Chi-square test: p = 0.09). ACL structural property estimates via MRI may provide a biomarker for clinical decision making.

Quantitative MRI Biomarkers to Predict Risk of Reinjury Within 2 Years After Bridge-Enhanced ACL Restoration.

BACKGROUND: Quantitative magnetic resonance imaging (qMRI) methods were developed to establish the integrity of healing anterior cruciate ligaments (ACLs) and grafts. Whether qMRI variables predict risk of reinjury is unknown. PURPOSE: To determine if qMRI measures at 6 to 9 months after bridge-enhanced ACL restoration (BEAR) can predict the risk of revision surgery within 2 years of the index procedure. STUDY DESIGN: Cohort study; Level of evidence, 2. METHODS: Originally, 124 patients underwent ACL restoration as part of the BEAR I, BEAR II, and BEAR III prospective trials and had consented to undergo an MRI of the surgical knee 6 to 9 months after surgery. Only 1 participant was lost to follow-up, and 4 did not undergo MRI, leaving a total of 119 patients for this study. qMRI techniques were used to determine the mean cross-sectional area; normalized signal intensity; and a qMRI-based predicted failure load, which was calculated using a prespecified equation based on cross-sectional area and normalized signal intensity. Patient-reported outcomes (International Knee Documentation Committee subjective score), clinical measures (hamstring strength, quadriceps strength, and side-to-side knee laxity), and functional outcomes (single-leg hop) were also measured at 6 to 9 months after surgery. Univariate and multivariable analyses were performed to determine the odds ratios (ORs) for revision surgery based on the qMRI and non-imaging variables. Patient age and medial posterior tibial slope values were included as covariates. RESULTS: In total, 119 patients (97%), with a median age of 17.6 years, underwent MRI between 6 and 9 months postoperatively. Sixteen of 119 patients (13%) required revision ACL surgery. In univariate analyses, higher International Knee Documentation Committee subjective score at 6 to 9 months postoperatively (OR = 1.66 per 10-point increase; P = .035) and lower qMRI-based predicted failure load (OR = 0.66 per 100-N increase; P = .014) were associated with increased risk of revision surgery. In the multivariable model, when adjusted for age and posterior tibial slope, the qMRI-based predicted failure load was the only significant predictor of revision surgery (OR = 0.71 per 100 N; P = .044). CONCLUSION: Quantitative MRI-based predicted failure load of the healing ACL was a significant predictor of the risk of revision within 2 years after BEAR surgery. The current findings highlight the potential utility of early qMRI in the postoperative management of patients undergoing the BEAR procedure.

Three-dimensional magnetic resonance imaging analysis shows sex-specific patterns in changes in anterior cruciate ligament cross-sectional area along its length.

Smaller anterior cruciate ligament (ACL) size in females has been hypothesized to be a key contributor to a higher incidence of ACL tears in that population, as a lower cross-sectional area (CSA) directly corresponds to a larger stress on the ligament for a given load. Prior studies have used a mid-length CSA measurement to quantify ACL size. In this study, we used magnetic resonance imaging to quantify the CSA along the entire length of the intact ACL. We hypothesized that changes in the ACL CSA along its length would have different patterns in males and females. We also hypothesized that changes in ACL CSA along its length would be associated with body size or knee size with different associations in females and males. MR images of contralateral ACL-intact knees of 108 patients (62 females, 13-35 years) undergoing ACL surgery were used to measure the CSA along the ACL length, using a custom program. For both females and males, the largest CSA was located at 37%-39% of ACL length from the tibial insertion. Compared to females, males had a significantly larger CSA only within the distal 41% of the ACL (p < 0.001). ACL CSA was associated with patient height and weight in males (r > 0.3; p < 0.05), whereas it was associated with intercondylar notch width in females (r > 0.3; p < 0.05). These findings highlight the importance of standardizing the location of measurement of ACL CSA.

Early MRI-based quantitative outcomes are associated with a positive functional performance trajectory from 6 to 24 months post-ACL surgery.

PURPOSE: Quantitative magnetic resonance imaging (qMRI) has been used to determine the failure properties of ACL grafts and native ACL repairs and/or restorations. How these properties relate to future clinical, functional, and patient-reported outcomes remain unknown. The study objective was to investigate the relationship between non-contemporaneous qMRI measures and traditional outcome measures following Bridge-Enhanced ACL Restoration (BEAR). It was hypothesized that qMRI parameters at 6 months would be associated with clinical, functional, and/or patient-reported outcomes at 6 months, 24 months, and changes from 6 to 24 months post-surgery. METHODS: Data of BEAR patients (n = 65) from a randomized control trial of BEAR versus ACL reconstruction (BEAR II Trial; NCT02664545) were utilized retrospectively for the present analysis. Images were acquired using the Constructive Interference in Steady State (CISS) sequence at 6 months post-surgery. Single-leg hop test ratios, arthrometric knee laxity values, and International Knee Documentation Committee (IKDC) subjective scores were determined at 6 and 24 months post-surgery. The associations between traditional outcomes and MRI measures of normalized signal intensity, mean cross-sectional area (CSA), volume, and estimated failure load of the healing ACL were evaluated based on bivariate correlations and multivariable regression analyses, which considered the potential effects of age, sex, and body mass index. RESULTS: CSA (r = 0.44, p = 0.01), volume (r = 0.44, p = 0.01), and estimated failure load (r = 0.48, p = 0.01) at 6 months were predictive of the change in single-leg hop ratio from 6 to 24 months in bivariate analysis. CSA (ßstandardized = 0.42, p = 0.01), volume (ßstandardized = 0.42, p = 0.01), and estimated failure load (ßstandardized = 0.48, p = 0.01) remained significant predictors when considering the demographic variables. No significant associations were observed between MRI variables and either knee laxity or IKDC when adjusting for demographic variables. Signal intensity was also not significant at any timepoint. CONCLUSION: The qMRI-based measures of CSA, volume, and estimated failure load were predictive of a positive functional outcome trajectory from 6 to 24 months post-surgery. These variables measured using qMRI at 6 months post-surgery could serve as prospective markers of the functional outcome trajectory from 6 to 24 months post-surgery, aiding in rehabilitation programming and return-to-sport decisions to improve surgical outcomes and reduce the risk of reinjury. LEVEL OF EVIDENCE: Level II.

Automated segmentation of the healed anterior cruciate ligament from T2 * relaxometry MRI scans.

Collagen organization of the anterior cruciate ligament (ACL) can be evaluated using T2 * relaxometry. However, T2 * mapping requires manual image segmentation, which is a time-consuming process and prone to inter- and intra- segmenter variability. Automating segmentation would address these challenges. A model previously trained using Constructive Interference in Steady State (CISS) scans was applied to T2 * segmentation via transfer learning. It was hypothesized that there would be no significant differences in the model's segmentation performance between T2 * and CISS, structural measures versus ground truth manual segmentation, and reliability versus independent and retest manual segmentation. Transfer learning was conducted using 54 T2 * scans of the ACL. Segmentation performance was assessed with Dice coefficient, precision, and sensitivity, and structurally with T2 * value, volume, subvolume proportions, and cross-sectional area. Model performance relative to independent manual segmentation and repeated segmentation by the ground truth segmenter (retest) were evaluated on a random subset. Segmentation performance was analyzed with Mann-Whitney U tests, structural measures with Wilcoxon signed-rank tests, and performance relative to manual segmentation with repeated-measures analysis of variance/Tukey tests (α = 0.05). T2 * segmentation performance was not significantly different from CISS on all measures (p > 0.35). No significant differences were detected in structural measures (p > 0.50). Automatic segmentation performed as well as the retest on all segmentation measures, whereas independent segmentations were lower than retest and/or automatic segmentation (p < 0.023). Structural measures were not significantly different between segmenters. The automatic segmentation model performed as well on the T2 * sequence as on CISS and outperformed independent manual segmentation while performing as well as retest segmentation.

Changes in the Cross-Sectional Profile of Treated Anterior Cruciate Ligament Within 2 Years After Surgery.

Background: The cross-sectional area (CSA) of the anterior cruciate ligament (ACL) and reconstructed graft has direct implications on its strength and knee function. Little is known regarding how the CSA changes along the ligament length and how those changes vary between treated and native ligaments over time. Hypothesis: It was hypothesized that (1) the CSA of reconstructed ACLs and restored ACLs via bridge-enhanced ACL restoration (BEAR) is heterogeneous along the length. (2) Differences in CSA between treated and native ACLs decrease over time. (3) CSA of the surgically treated ACLs is correlated significantly with body size (ie, height, weight, body mass index) and knee size (ie, bicondylar and notch width). Study Design: Cohort study; Level of evidence, 2. Methods: Magnetic resonance imaging scans of treated and contralateral knees of 98 patients (n = 33 ACL reconstruction, 65 BEAR) at 6, 12, and 24 months post-operation were used to measure the ligament CSA at 1% increments along the ACL length (tibial insertion, 0%; femoral insertion, 100%). Statistical parametric mapping was used to evaluate the differences in CSA between 6 and 24 months. Correlations between body and knee size and treated ligament CSA along its length were also assessed. Results: Hamstring autografts had larger CSAs than native ACLs at all time points (P < .001), with region of difference decreasing from proximal 95% of length (6 months) to proximal 77% of length (24 months). Restored ACLs had larger CSAs than native ACLs at 6 and 12 months, with larger than native CSA only along a small midsubstance region at 24 months (P < .001). Graft CSA was correlated significantly with weight (6 and 12 months), bicondylar width (all time points), and notch width (24 months). Restored ACL CSA was significantly correlated with bicondylar width (6 months) and notch width (6 and 12 months). Conclusion: Surgically treated ACLs remodel continuously within the first 2 years after surgery, leading to ligaments/grafts with heterogeneous CSAs along the length, similar to the native ACL. While reconstructed ACLs remained significantly larger, the restored ACL had a CSA profile comparable with that of the contralateral native ACL. In addition to size and morphology differences, there were fundamental differences in factors contributing to CSA profile between the ACL reconstruction and BEAR procedures. Registration: NCT02664545 (ClinicalTrials.gov identifier).

Reproducibility and postacquisition correction methods for quantitative magnetic resonance imaging of the anterior cruciate ligament (ACL).

Quantitative magnetic resonance imaging has been used to evaluate the structural integrity of knee joint structures. However, variations in acquisition parameters between scanners pose significant challenges. Understanding the effect of small differences in acquisition parameters for quantitative sequences is vital to the validity of cross-institutional studies, and for the harmonization of large, heterogeneous datasets to train machine learning models. The study objective was to assess the reproducibility of T2 * relaxometry and the constructive interference in steady-state sequence (CISS) across scanners, with minimal hardware-necessitated changes to acquisition parameters. It was hypothesized that there would be no significant differences between scanners in anterior cruciate ligament T2 * relaxation times and CISS signal intensities (SI). Secondarily, it was hypothesized that differences could be corrected by rescaling the SI distribution to harmonize between scanners. Seven volunteers were scanned on 3T Prisma and Tim Trio scanners (Siemens). Three correction methods were evaluated for T2 *: inverse echo time scaling, z-scoring, and Nyúl histogram matching. For CISS, scans were normalized to cortical bone, scaled by the background noise ratio, and log-transformed. Before correction, significant mean differences of 6.0 ± 3.2 ms (71.8%; p = 0.02) and 0.49 ± 0.15 units (40.7%; p = 0.02) for T2 * and CISS across scanners were observed, respectively. After rescaling, T2 * differences decreased to 2.6 ± 2.7 ms (23.9%; p = 0.03), 1.3 ± 2.5 ms (10.9%; p = 0.13), and 1.27 ± 3.0 ms (19.6%; p = 0.40) for inverse echo time, z-scoring, and Nyúl, respectively, while CISS decreased to 0.01 ± 0.11 units (4.0%; p = 0.87). These findings suggest that small acquisition parameter differences may lead to large changes in T2 * and SI values that must be reconciled to compare data across magnets.

A transfer learning approach for automatic segmentation of the surgically treated anterior cruciate ligament.

Quantitative magnetic resonance imaging enables quantitative assessment of the healing anterior cruciate ligament or graft post-surgery, but its use is constrained by the need for time consuming manual image segmentation. The goal of this study was to validate a deep learning model for automatic segmentation of repaired and reconstructed anterior cruciate ligaments. We hypothesized that (1) a deep learning model would segment repaired ligaments and grafts with comparable anatomical similarity to intact ligaments, and (2) automatically derived quantitative features (i.e., signal intensity and volume) would not be significantly different from those obtained by manual segmentation. Constructive Interference in Steady State sequences were acquired of ACL repairs (n = 238) and grafts (n = 120). A previously validated model for intact ACLs was retrained on both surgical groups using transfer learning. Anatomical performance was measured with Dice coefficient, sensitivity, and precision. Quantitative features were compared to ground truth manual segmentation. Automatic segmentation of both surgical groups resulted in decreased anatomical performance compared to intact ACL automatic segmentation (repairs/grafts: Dice coefficient = .80/.78, precision = .79/.78, sensitivity = .82/.80), but neither decrease was statistically significant (Kruskal-Wallis: Dice coefficient p = .02, precision p = .09, sensitivity p = .17; Dunn post-hoc test for Dice coefficient: repairs/grafts p = .054/.051). There were no significant differences in quantitative features between the ground truth and automatic segmentation of repairs/grafts (0.82/2.7% signal intensity difference, p = .57/.26; 1.7/2.7% volume difference, p = .68/.72). The anatomical similarity performance and statistical similarities of quantitative features supports the use of this automated segmentation model in quantitative magnetic resonance imaging pipelines, which will accelerate research and provide a step towards clinical applicability.

Regional Differences in Anterior Cruciate Ligament Signal Intensity After Surgical Treatment.

BACKGROUND: Magnetic resonance-based measurements of signal intensity have been used to track healing of surgically treated anterior cruciate ligaments (ACLs). However, it is unknown how the signal intensity values in different regions of the ligament or graft change during healing. HYPOTHESES: (1) Normalized signal intensity of the healing graft or repaired ACL is heterogeneous; (2) temporal changes in normalized signal intensity values differ among the tibial, middle, and femoral regions; and (3) there are no differences in regional normalized signal intensity values 2 years postoperatively among grafts, repaired ACLs, and contralateral native ACLs. STUDY DESIGN: Cohort study; Level of evidence, 2. METHODS: Magnetic resonance imaging scans were analyzed from patients in a trial comparing ACL reconstruction (n = 35) with bridge-enhanced ACL repair (n = 65). The ACLs were segmented from images acquired at 6, 12, and 24 months postoperatively and were partitioned into 3 sections along the longitudinal axis (femoral, middle, and tibial). Linear mixed modeling was used to compare location-specific differences in normalized ligament signal intensity among time points (6, 12, and 24 months) and groups (ACL reconstruction, repair, and contralateral native ACL). RESULTS: For grafts, the middle region had a higher mean normalized signal intensity when compared with the femoral region at all time points (P < .01) but compared with the tibial region only at 6 months (P < .01). For repaired ACLs, the middle region had a higher mean normalized signal intensity versus the femoral region at all time points (P < .01) but versus the tibial region only at 6 and 12 months (P < .04). From 6 to 24 months, the grafts showed the greatest reduction in normalized signal intensity in the femoral and middle regions (vs tibial regions; P < .01), while there were no regional differences in repaired ACLs. At 2 years after surgery, repaired ACLs had a lower normalized signal intensity in the tibial region as compared with reconstructed grafts and contralateral native ACLs (P < .01). CONCLUSION: The results suggest that graft remodeling is location specific. Repaired ACLs were more homogeneous, with lower or comparable normalized signal intensity values at 2 years as compared with the contralateral native ACL and reconstructed grafts.

ACL Size, but Not Signal Intensity, Is Influenced by Sex, Body Size, and Knee Anatomy.

BACKGROUND: Little is known about sex-based differences in anterior cruciate ligament (ACL) tissue quality in vivo or the association of ACL size (ie, volume) and tissue quality (ie, normalized signal intensity on magnetic resonance imaging [MRI]) with knee anatomy. HYPOTHESIS: We hypothesized that (1) women have smaller ACLs and greater ACL normalized signal intensity compared with men, and (2) ACL size and normalized signal intensity are associated with age, activity levels, body mass index (BMI), bicondylar width, intercondylar notch width, and posterior slope of the lateral tibial plateau. STUDY DESIGN: Cross-sectional study; Level of evidence, 3. METHODS: Knee MRI scans of 108 unique ACL-intact knees (19.7 ± 5.5 years, 62 women) were used to quantify the ACL signal intensity (normalized to cortical bone), ligament volume, mean cross-sectional area, and length. Independent t tests were used to compare the MRI-based ACL parameters between sexes. Univariate and multivariate linear regression analyses were used to investigate the associations between normalized signal intensity and size with age, activity levels, BMI, bicondylar width, notch width, and posterior slope of the lateral tibial plateau. RESULTS: Compared with men, women had significantly smaller mean ACL volume (men vs women: 2028 ± 472 vs 1591 ± 405 mm3), cross-sectional area (49.4 ± 9.6 vs 41.5 ± 8.6 mm2), and length (40.8 ± 2.8 vs 38.1 ± 3.1 mm) (P < .001 for all), even after adjusting for BMI and bicondylar width. There was no difference in MRI signal intensity between men and women (1.15 ± 0.24 vs 1.12 ± 0.24, respectively; P = .555). BMI, bicondylar width, and intercondylar notch width were independently associated with a larger ACL (R 2 > 0.16, P < .001). Younger age and steeper lateral tibial slope were independently associated with shorter ACL length (R 2 > 0.03, P < .04). The combination of BMI and bicondylar width was predictive of ACL volume and mean cross-sectional area (R 2 < 0.3). The combination of BMI, bicondylar width, and lateral tibial slope was predictive of ACL length (R 2 = 0.39). Neither quantified patient characteristics nor anatomic variables were associated with signal intensity. CONCLUSION: Men had larger ACLs compared with women even after adjusting for BMI and knee size (bicondylar width). No sex difference was observed in signal intensity, suggesting no difference in tissue quality. The association of the intercondylar notch width and lateral tibial slope with ACL size suggests that the influence of these anatomic features on ACL injury risk may be partially explained by their effect on ACL size. REGISTRATION: NCT02292004 and NCT02664545 (ClinicalTrials.gov identifier).

Automated magnetic resonance image segmentation of the anterior cruciate ligament.

The objective of this study was to develop an automated segmentation method for the anterior cruciate ligament that is capable of facilitating quantitative assessments of the ligament in clinical and research settings. A modified U-Net fully convolutional network model was trained, validated, and tested on 246 Constructive Interference in Steady State magnetic resonance images of intact anterior cruciate ligaments. Overall model performance was assessed on the image set relative to an experienced (>5 years) "ground truth" segmenter in two domains: anatomical similarity and the accuracy of quantitative measurements (i.e., signal intensity and volume) obtained from the automated segmentation. To establish model reliability relative to manual segmentation, a subset of the imaging data was resegmented by the ground truth segmenter and two additional segmenters (A, 6 months and B, 2 years of experience), with their performance evaluated relative to the ground truth. The final model scored well on anatomical performance metrics (Dice coefficient = 0.84, precision = 0.82, and sensitivity = 0.85). The median signal intensities and volumes of the automated segmentations were not significantly different from ground truth (0.3% difference, p = .9; 2.3% difference, p = .08, respectively). When the model results were compared with the independent segmenters, the model predictions demonstrated greater median Dice coefficient (A = 0.73, p = .001; B = 0.77, p = NS) and sensitivity (A = 0.68, p = .001; B = 0.72, p = .003). The model performed equivalently well to retest segmentation by the ground truth segmenter on all measures. The quantitative measures extracted from the automated segmentation model did not differ from those of manual segmentation, enabling their use in quantitative magnetic resonance imaging pipelines to evaluate the anterior cruciate ligament.