Anterior cruciate ligament (ACL) injuries are one of the most common knee injuries. Unfortunately, the prevention programs have not significantly reduced these traumatic injuries, and debilitating repercussions (i.e. pain, meniscal degeneration, posttraumatic osteoarthritis) occur even after reconstructive or rehabilitative treatment. Therefore, the overall objectives of this dissertation are 1) to quantitatively determine the injury mechanism for non-contact ACL injuries using bone bruises present on magnetic resonance imaging, and 2) to assess cartilage recovery time as a potential imaging biomarker of osteoarthritis in ACL injured subjects. In the first aim, two common bone bruise patterns were determined and used to quantitatively determine the injury position of the knee using numerical optimization. These findings showed that landing in extension with minimal valgus, small internal tibial rotations, and large anterior translations may be high risk for ACL ruptures. In the second aim, a deep learning autosegmentation algorithm was developed on healthy normal knee MRIs and was successfully validated and applied on ACL reconstructed (ACLR) subjects. Then, their cartilage strains and recovery times were calculated, which showed that the ACLR subjects had significantly higher strains and a much longer recovery time, suggesting that ACLR subjects’ cartilage are softer and less able to rebound quickly to their baseline thickness. In conclusion, this dissertation accomplished both goals of providing quantitative in vivo information on both preventing non-contact ACL injuries in the future as well as identifying early changes in cartilage mechanical properties noninvasively through cartilage recovery times.