Dataset Information

Deep Learning for Hierarchical Severity Staging of Anterior Cruciate Ligament Injuries from MRI.

ABSTRACT: Purpose:To evaluate the diagnostic utility of two convolutional neural networks (CNNs) for severity staging of anterior cruciate ligament (ACL) injuries. Materials and Methods:In this retrospective study, 1243 knee MR images (1008 intact, 18 partially torn, 77 fully torn, and 140 reconstructed ACLs) from 224 patients (mean age, 47 years ± 14 [standard deviation]; 54% women) were analyzed. The MRI examinations were performed between 2011 and 2014. A modified scoring metric was used. Classification of ACL injuries using deep learning involved use of two types of CNN, one with three-dimensional (3D) and the other with two-dimensional (2D) convolutional kernels. Performance metrics included sensitivity, specificity, weighted Cohen κ, and overall accuracy, and the McNemar test was used to compare the performance of the CNNs. Results:The overall accuracies for ACL injury classification using the 3D CNN and 2D CNN were 89% (225 of 254) and 92% (233 of 254), respectively (P = .27), and both CNNs had a weighted Cohen κ of 0.83. The 2D CNN and 3D CNN performed similarly in classifying intact ACLs (2D CNN, sensitivity of 93% [188 of 203] and specificity of 90% [46 of 51] vs 3D CNN, sensitivity of 89% [180 of 203] and specificity of 88% [45 of 51]). Classification of full tears by both networks was also comparable (2D CNN, sensitivity of 82% [14 of 17] and specificity of 94% [222 of 237] vs 3D CNN, sensitivity of 76% [13 of 17] and specificity of 100% [236 of 237]). The 2D CNN classified all reconstructed ACLs correctly. Conclusion:Two-dimensional and 3D CNNs applied to ACL lesion classification had high sensitivity and specificity, suggesting that these networks could be used to help nonexperts grade ACL injuries. Supplemental material is available for this article. © RSNA, 2020.

SUBMITTER: Namiri NK

PROVIDER: S-EPMC7392061 | biostudies-literature |

REPOSITORIES: biostudies-literature

ACCESS DATA

Similar Datasets

Project description:PurposeTo assess failure rate, outcomes, and patient satisfaction in patients who underwent anterior cruciate ligament (ACL) repair with suture augmentation for clinical instability and proximal avulsion of the ACL.MethodsWe retrospectively reviewed consecutive suture-augmented ACL repairs performed by a single surgeon between January 2014 and June 2016 for proximal ACL avulsion. Patients were included if they were at least 24 months postoperative from repair surgery. Patients were excluded from the study if they underwent primary ACL reconstruction instead of repair or if they had a concomitant multiligamentous knee injury. Knee Injury and Osteoarthritis Outcome Score (KOOS), visual analog scale (VAS), Veterans RAND-12 (VR-12), Marx Activity, and Single Assessment Numeric Evaluation data were collected.ResultsOf 172 patients who underwent ACL surgery between January 2014 and June 2016, 28 (16%) with Sherman type I or II ACL tears or high-grade partial avulsion with clinical instability underwent ACL repair with suture augmentation. One patient was not available for follow-up. The 27 patients were age 27.4 ± 8.6 years, 18 males (66.7%), and 2.8 ± 0.7 years follow-up (range, 2.0-3.8 years). Of these 27 patients, 4 recurrent ACL injuries (14.8%) required revision to reconstruction. The remaining 23 patients had successful ACL repair with no clinical instability and no subjective complaints at final follow-up. Final scores were KOOS 83.7 ± 12.8, Marx 8.6 ± 4.0, VAS 1.1 ± 1.8, physical VR-12 53.6 ± 5.2, mental VR-12 53.1 ± 8.1, and Single Assessment Numeric Evaluation 83.0 ± 12.9. In the 11 patients with baseline data, significant improvements were observed in composite KOOS (50.4 ± 11.5 to 85.7 ± 8.4; P < .001; VAS: 3.9 ± 2.6 to 0.8 ± 0.8; P = .002; and physical VR-12: 39.9 ± 6.5 to 55.5 ± 3.3; P < .001). All 11 patients (100%) met or exceeded the KOOS composite minimum clinically important difference (mean 34.0 increase).ConclusionsIn patients with proximal ACL avulsion, arthroscopic primary ACL repair with suture augmentation demonstrated high functional outcome and improved patient-reported outcomes at 2-year follow-up. The rate of graft failure was 15%.Level of evidenceLevel IV, therapeutic case series.

Project description:PurposeTo investigate the feasibility of using a deep learning-based approach to detect an anterior cruciate ligament (ACL) tear within the knee joint at MRI by using arthroscopy as the reference standard.Materials and methodsA fully automated deep learning-based diagnosis system was developed by using two deep convolutional neural networks (CNNs) to isolate the ACL on MR images followed by a classification CNN to detect structural abnormalities within the isolated ligament. With institutional review board approval, sagittal proton density-weighted and fat-suppressed T2-weighted fast spin-echo MR images of the knee in 175 subjects with a full-thickness ACL tear (98 male subjects and 77 female subjects; average age, 27.5 years) and 175 subjects with an intact ACL (100 male subjects and 75 female subjects; average age, 39.4 years) were retrospectively analyzed by using the deep learning approach. Sensitivity and specificity of the ACL tear detection system and five clinical radiologists for detecting an ACL tear were determined by using arthroscopic results as the reference standard. Receiver operating characteristic (ROC) analysis and two-sided exact binomial tests were used to further assess diagnostic performance.ResultsThe sensitivity and specificity of the ACL tear detection system at the optimal threshold were 0.96 and 0.96, respectively. In comparison, the sensitivity of the clinical radiologists ranged between 0.96 and 0.98, while the specificity ranged between 0.90 and 0.98. There was no statistically significant difference in diagnostic performance between the ACL tear detection system and clinical radiologists at P < .05. The area under the ROC curve for the ACL tear detection system was 0.98, indicating high overall diagnostic accuracy.ConclusionThere was no significant difference between the diagnostic performance of the ACL tear detection system and clinical radiologists for determining the presence or absence of an ACL tear at MRI.© RSNA, 2019Supplemental material is available for this article.

Project description:PURPOSE:To evaluate differences in repair and nonoperatively managed grade III medial collateral ligament (MCL) injuries during anterior cruciate ligament (ACL) reconstruction. METHODS:Patients enrolled in a multicenter prospective longitudinal group who underwent unilateral primary ACL reconstruction between 2002 and 2008 were evaluated. Patients with concomitant grade III MCL injuries treated either operatively or nonoperatively were identified. Concurrent injuries, subsequent surgeries, surgical chronicity, and MCL tear location were analyzed. Patient-reported outcomes were measured at time of ACL reconstruction and 2-year follow-up. RESULTS:Initially, 3,028 patients were identified to have undergone primary ACL reconstruction during the time frame; 2,586 patients completed 2-year follow-up (85%). Grade III MCL tears were documented in 1.1% (27 of 2,586): 16 operatively managed patients and 11 nonoperatively treated MCLs during ACL reconstruction. The baseline Knee Injury and Osteoarthritis Outcome Score (KOOS) and International Knee Documentation Committee scores were lower in patients who underwent operative MCL treatment. Reoperation rates for arthrofibrosis were 19% after repair and 9% after conservative management (P = .48). At 2 years, both groups significantly improved; however, the nonoperative MCL group maintained superior patient-reported outcomes in terms of minimal clinically important differences, but these differences did not reach statistical significance (KOOS sports/recreation [88.2 vs 74.4, P = .10], KOOS knee-related quality of life [81.3 vs 68.4, P = .13], and International Knee Documentation Committee [87.6 vs 76.0, P = .14]). Tibial-sided MCL injuries were associated with clinically inferior baseline scores compared with femoral-sided MCL (KOOS knee-related quality of life, 34.4 vs 18.5, P = .09), but these differences resolved by 2 years. Surgical chronicity did not influence 2-year outcome. CONCLUSIONS:Both operative and nonoperative management of MCL tears in our patient group demonstrated clinical improvements between study enrollment and 2-year follow-up. MCL surgery during ACL reconstruction was assigned to patients with worse symptoms at enrollment and was associated with worse outcomes at 2 years. A subset of patients with severe combined ACL and medial knee injuries may benefit from operative management; however, that population has yet to be defined. LEVEL OF EVIDENCE:Level III, retrospective cohort.

Dataset Information

Deep Learning for Hierarchical Severity Staging of Anterior Cruciate Ligament Injuries from MRI.

Similar Datasets

OmicsDI is part of the ELIXIR infrastructure

Tweets