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

Namiri, Nikan K.; Flament, Io; Astuto, Bruno; Shah, Rutwik; Tibrewala, Radhika; Calivá, Francesco; Link, Thomas M.; Pedoia, Valentina; Majumdar, Sharmila

doi:10.1148/ryai.2020190207

Cited by 42 publications

(36 citation statements)

References 30 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…The 3D CNN models were not performed well as compared to 2D CNN due to the small dataset in the work of Namiri et al [ 39 ]. The model was found over-fitting in the case of partial tears, however obtained better results with 3D CNN than with 2D.The sample of patients were not balanced among all grading and dataset split based upon the patients, which caused correlations among multiple images.…”

Section: Discussionmentioning

confidence: 99%

“…However, the burden of training the all three architectures, in a cascaded fashion, is computationally expensive and time consuming. In the study, Namiri et al [ 39 ] used 3D CNN classify hierarchical severity stages in ACL automatically, that had an accuracy 3% more than 2D CNN. The study of [ 40 ] related arthroscopy findings of MRI dataset and used DenseNet architecture upon 489 MRI samples only, in which 163 were from an ACL tear and 245 were from an intact ACL.…”

Section: Related Workmentioning

confidence: 99%

See 1 more Smart Citation

Efficient Detection of Knee Anterior Cruciate Ligament from Magnetic Resonance Imaging Using Deep Learning Approach

et al. 2021

View full text Add to dashboard Cite

The most commonly injured ligament in the human body is an anterior cruciate ligament (ACL). ACL injury is standard among the football, basketball and soccer players. The study aims to detect anterior cruciate ligament injury in an early stage via efficient and thorough automatic magnetic resonance imaging without involving radiologists, through a deep learning method. The proposed approach in this paper used a customized 14 layers ResNet-14 architecture of convolutional neural network (CNN) with six different directions by using class balancing and data augmentation. The performance was evaluated using accuracy, sensitivity, specificity, precision and F1 score of our customized ResNet-14 deep learning architecture with hybrid class balancing and real-time data augmentation after 5-fold cross-validation, with results of 0.920%, 0.916%, 0.946%, 0.916% and 0.923%, respectively. For our proposed ResNet-14 CNN the average area under curves (AUCs) for healthy tear, partial tear and fully ruptured tear had results of 0.980%, 0.970%, and 0.999%, respectively. The proposing diagnostic results indicated that our model could be used to detect automatically and evaluate ACL injuries in athletes using the proposed deep-learning approach.

show abstract

Section: Discussionmentioning

confidence: 99%

Section: Related Workmentioning

confidence: 99%

Efficient Detection of Knee Anterior Cruciate Ligament from Magnetic Resonance Imaging Using Deep Learning Approach

et al. 2021

View full text Add to dashboard Cite

show abstract

“…The sensitivity, specificity, and accuracy of MRI in the diagnosis of ACL injury was 96.78%, 90.62%, and 92.17%, respectively, and there was no great difference from the results of arthroscopy ( P > 0.05). Namiri et al [ 18 ] found that the indirect signs of ACL tear had high specificity (91%∼100%) and sensitivity in a retrospective study of the correlation between MRI imaging and arthroscopy in 100 patients. Therefore, these signs can determine whether the patient had an ACL tear, which was similar to the conclusion of this study, indicating that MRI can accurately diagnose ACL injury.…”

Section: Discussionmentioning

confidence: 99%

Deep Learning-Based Magnetic Resonance Imaging Image Features for Diagnosis of Anterior Cruciate Ligament Injury

Ren

Zhou

et al. 2021

Journal of Healthcare Engineering

View full text Add to dashboard Cite

To study and explore the adoption value of magnetic resonance imaging (MRI) in the diagnosis of anterior cruciate ligament (ACL) injuries, a multimodal feature fusion model based on deep learning was proposed for MRI diagnosis. After the related performance of the proposed algorithm was evaluated, it was utilized in the diagnosis of knee joint injuries. Thirty patients with knee joint injuries who came to our hospital for treatment were selected, and all patients were diagnosed with MRI based on deep learning multimodal feature fusion model (MRI group) and arthroscopy (arthroscopy group). The results showed that deep learning-based MRI sagittal plane detection had a great advantage and a high accuracy of 96.28% in the prediction task of ACL tearing. The sensitivity, specificity, and accuracy of MRI in the diagnosis of ACL injury was 96.78%, 90.62%, and 92.17%, respectively, and there was no considerable difference in contrast to the results obtained through arthroscopy ( P > 0.05 ). The positive rate of acute ACL patients with bone contusion and medial collateral ligament injury was substantially superior to that of chronic injury. Moreover, the incidence of chronic injury ACL injury with meniscus tear and cartilage injury was notably higher than that of acute injury, with remarkable differences ( P < 0.05 ). In summary, MRI images based on deep learning improved the sensitivity, specificity, and accuracy of ACL injury diagnosis and can accurately determined the type of ACL injury. In addition, it can provide reference information for clinical treatment plan selection and surgery and can be applied and promoted in clinical diagnosis.

show abstract

“…Right lower cut-out boxes represent a magnification of the left upper area (dashed box). In A and B: dark gray = Liu et al [3]; yellow = Germann et al [4]; blue = Namiri et al [14]; green = Bien et al [11]. In C: gray = Oei et al [18]; orange = Smith et al [20]; light blue = Phelan et al [21]; red = Crawford et al [19].…”

Section: Meniscus Tearsmentioning

confidence: 99%

“…The CNN achieved a sensitivity of 97-100% and a specificity of 100% for identifying ACL grafts. For intact ACL, the CNN sensitivities were 89-93% and specificities were 88-90%, whereas the CNN achieved sensitivities of 76-82% and specificities of 94-100% for full-thickness ACL tears [14]. The ground truth was based on radiological assessments, and comparisons to human readers were not reported.…”

Section: Introductionmentioning

confidence: 99%

Artificial intelligence for MRI diagnosis of joints: a scoping review of the current state-of-the-art of deep learning-based approaches

Fritz

2021

Skeletal Radiol

View full text Add to dashboard Cite

Deep learning-based MRI diagnosis of internal joint derangement is an emerging field of artificial intelligence, which offers many exciting possibilities for musculoskeletal radiology. A variety of investigational deep learning algorithms have been developed to detect anterior cruciate ligament tears, meniscus tears, and rotator cuff disorders. Additional deep learning-based MRI algorithms have been investigated to detect Achilles tendon tears, recurrence prediction of musculoskeletal neoplasms, and complex segmentation of nerves, bones, and muscles. Proof-of-concept studies suggest that deep learning algorithms may achieve similar diagnostic performances when compared to human readers in meta-analyses; however, musculoskeletal radiologists outperformed most deep learning algorithms in studies including a direct comparison. Earlier investigations and developments of deep learning algorithms focused on the binary classification of the presence or absence of an abnormality, whereas more advanced deep learning algorithms start to include features for characterization and severity grading. While many studies have focused on comparing deep learning algorithms against human readers, there is a paucity of data on the performance differences of radiologists interpreting musculoskeletal MRI studies without and with artificial intelligence support. Similarly, studies demonstrating the generalizability and clinical applicability of deep learning algorithms using realistic clinical settings with workflow-integrated deep learning algorithms are sparse. Contingent upon future studies showing the clinical utility of deep learning algorithms, artificial intelligence may eventually translate into clinical practice to assist detection and characterization of various conditions on musculoskeletal MRI exams.

show abstract

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

Abstract: This deep learning pipeline may lend towards diagnostic worklist prioritization, standardization, and generalizability in assessing anterior cruciate ligament lesions, in addition to point-of-care communication with patients by non-experts.

Cited by 42 publications

References 30 publications

Efficient Detection of Knee Anterior Cruciate Ligament from Magnetic Resonance Imaging Using Deep Learning Approach

Efficient Detection of Knee Anterior Cruciate Ligament from Magnetic Resonance Imaging Using Deep Learning Approach

Deep Learning-Based Magnetic Resonance Imaging Image Features for Diagnosis of Anterior Cruciate Ligament Injury

Artificial intelligence for MRI diagnosis of joints: a scoping review of the current state-of-the-art of deep learning-based approaches

Contact Info

Product

Resources

About