Image-Based Approach to Guide Postoperative Care of Patients with Anterior Cruciate Ligament (ACL) Injury
A set of machine learning and image processing algorithms and software that can accurately quantify ACL integrity based on MRI
Background
Approximately 400,000 anterior cruciate ligament (ACL) reconstruction surgeries are performed each year in the US. ACL injuries result in serious disabilities and can give rise to further complications such as joint degeneration. Accurate and reliable assessment of ACL integrity and tissue healing after surgery can help reduce the re-injury rate, which is currently around 40%. Accurate and detailed ACL assessment can also be useful in prescribing the types of physical activities that a patient can safely engage in after injury or reconstruction. Currently, ACL integrity is quantified by evaluating the whole knee function, inspecting lower extremity performance, or based on patient-reported outcomes. However, these methods are subjective and highly non-specific. Hence, there is a critical need for methods that can accurately assess the ACL size and the degree of tissue organization and integrity.
Technology Overview
This invention includes a set of machine learning and image processing algorithms and software that can accurately quantify ACL integrity based on MRI. The new method can determine the stage of tissue healing based on the 3D spatial pattern of ACL MRI signal. Unlike existing MRI-based methods, which are based on only the gross ACL appearance, the new method takes into account the intensity variations across ACL. The new method consists of deep learning models that take the 3D MRI of the ACL as input and compute the probability of re-injury based on the complete 3D distribution of the MRI signal.
The inventors evaluated the performance of the new method in predicting the risk of ACL re-injury following a surgery. They found that the new method could predict the risk of ACL re-injury with high accuracy (81%), sensitivity (77%), and specificity (86%). They also found that the new method could discover novel imaging biomarkers. The results of these preliminary experiments suggest that the new method has a great potential to be used as a clinical decision-making tool to inform postoperative management of patients with ACL injury.
Benefits
- Accounts for the size and shape of ACL as well as the complete 3D distribution of the MRI signal
- Independent of MRI sequence and MRI magnet, hence it can be applied to other scanners and centers
- Fully automatic and non-invasive
- Highlights regions in the 3D MRI with important features so that the clinician/radiologist can make a more informed decision
- Capable of incorporating non-imaging data such as demographic and clinical information to improve prediction accuracy.
Applications
- Predicting the risk of re-injury after a surgery for ACL and, potentially, for other joints
- Determining the time when an ACL patient can resume physical activities
- Devising postoperative rehabilitation plans for ACL patients
- Comparing the advantages and disadvantages of different surgical procedures for ACL
IP Status
- Patent application submitted
