Recent advances in technology and artificial intelligence have transformed radiology. Artificial intelligence, which uses machine learning, deep learning algorithms, and cellular neural networks, is now applicable to image analysis, enabling the analysis of large volumes of data and the early identification of lesion features, such as tumors. Integrating artificial intelligence into clinical workflows increases the efficiency of medical report generation and enables efficient prioritization of complex or urgent cases. Predictive analytics based on artificial intelligence uses extensive patient data, including demographics, medical history, tests, and imaging, to build more accurate predictive models. Among the areas where artificial intelligence is successfully implemented in radiology is the automated detection of lung nodules. In the field of neurology, artificial intelligence helps track progressive lesions through volumetric analysis of the brain and surveillance of demyelinating diseases. In oncologic radiology, artificial intelligence is used for automated image segmentation, lesion review, and standardized report generation. Therefore, in this ever-evolving landscape, radiologists must embrace emerging technological advances in order to occupy a unique position at the intersection of different diagnostic fields.
Background: Rotator cuff tendinopathy is most often described as a continuum between the normal cuff and rotator cuff tears with calcific tendinitis having its place along this continuum. Although many studies have focused on the role of magnetic resonance imagining (MRI) in diagnosing the extent of rotator cuff tears and their associated findings with good interobserver reliability, the same cannot be stated about MRI tendinopathy findings. Because of this discrepancy in diagnostic reliability, tendinopathy tends to be overtreated with injections when associated with symptoms, thus potentially increasing the risk of calcific tendinitis and progression toward rotator cuff tears. This study aims to assess whether diagnosing shoulder MRI tendinopathy patterns through dichotomization can accelerate clinical progress toward consensus. Methods: This study is a large retrospective cohort of 184 patients that underwent a 1.5T shoulder MRI for shoulder pain. Inclusion criteria were acromioclavicular arthrosis diagnosed in patients of any age. Exclusion criteria were partial or complete rotator cuff tears. Tendinopathy was considered the dependent variable and registered as a dichotomous variable while acromioclavicular joint arthrosis together with gender was categorical and age was the continuous variable. An attempt was made to generate a clinically significant binary logistic regression to assess the odds ratio of diagnosing tendinopathy based on age, gender, and acromioclavicular joint arthrosis status. Results: An overwhelming proportion of patients was positive for tendinopathy findings (95.11%). 64.12% of patients were within the active age group with patients within the 50-59 group being diagnosed the most with rotator cuff tendinopathy. Conclusions: Due to the high variability of MRI findings that can be considered positive for rotator cuff tendinopathy, an overwhelming skew toward a positive diagnosis was observed, thus dichotomizing tendinopathy diagnosis is not appropriate for clinically relevant conclusion-making.
