Diagnostic Performance of Articular Ultrasound Versus Magnetic Resonance Imaging in the Approach to Rotator Cuff Lesions at a Referral Hospital

Introduction Rotator cuff injury is a leading cause of musculoskeletal pain and the most common shoulder pathology. It is characterized by pain and limited mobility, particularly during overhead activities, and is associated with factors such as age, trauma, occupation, limb dominance, and cardiovascular risk. Diagnosis requires clinical evaluation and imaging, with ultrasound being a cost-effective and accessible option, whereas MRI is preferred for persistent or ambiguous cases. However, MRI access may be restricted by geographical and economic constraints. Methodology A retrospective census was conducted at a fourth-level hospital in Medellín, Antioquia (2015-2020), following research committee approval. Patients over 18 years old with suspected rotator cuff injury who underwent arthroscopy and had MRI and ultrasound reports within six months were included. Imaging and surgical data were retrieved from Xen RIS HIRUKO and Matrix systems. Ultrasound was performed using a General Electric Logiq S8 (15 MHz transducer) (General Electric Healthcare, Chicago, IL) and a Toshiba Aplio 400 (18 MHz transducer) (Toshiba Medical Systems Corporation, Tokyo, Japan). MRI was conducted on a Philips Achieva 1.5 T (Philips Healthcare, Eindhoven, Netherlands) with standard musculoskeletal protocols. Data were collected independently, processed in Excel (Microsoft Corporation, Redmond, WA), and analyzed using STATA 14 (StataCorp LLC, College Station, TX). Quantitative variables were reported as means and standard deviations, with normality assessed using the Kolmogorov-Smirnov test, while qualitative variables were expressed as frequencies. Results Twenty-four patients (62.4 ± 12.2 years) had 27 rotator cuff injuries; eight (33.3%) were traumatic (all men), and 16 (66.7%) were degenerative. Right-sided lesions occurred in 12 (50%), left-sided in nine (37.5%), and bilateral in three (12.5%). The most common arthroscopic finding was a complete supraspinatus rupture in 16 shoulders (59%), followed by subscapularis partial ruptures not exceeding 50% in 10 shoulders (37%) and supraspinatus partial ruptures exceeding 50% in seven shoulders (26%). For supraspinatus evaluation, ultrasound detected 79% of lesions and MRI detected 75%, both achieving 93.7% sensitivity for complete ruptures. For partial ruptures exceeding 50%, both modalities demonstrated good specificity and predictive values but low sensitivity (42.8%), with an area under the curve (AUC) of 0.78 for complete ruptures. Regarding the subscapularis, MRI detected 54.5% of lesions compared to 26.3% detected by ultrasound, demonstrating higher specificity. In the infraspinatus, both modalities detected only one of two complete ruptures, with several false positives. Agreement with arthroscopy was moderate for the supraspinatus (kappa: ultrasound = 0.59; MRI = 0.448) and low for the subscapularis (kappa: ultrasound = 0.25; MRI = 0.56), with ultrasound performing better in traumatic cases. Conclusions This study provides a direct comparison between ultrasound and MRI in the evaluation of rotator cuff injuries, with arthroscopy as the reference standard. While both modalities were effective in detecting complete supraspinatus tears, MRI demonstrated greater accuracy for subscapularis tendon assessment. These findings support the continued use of ultrasound as a practical and reliable tool, particularly in resource-limited settings.