Introduction: Carpal tunnel syndrome is a common and debilitating condition of the upper extremity. Failure of surgical treatment and persistence of symptoms is reported in 10-20% of patients, with 12% of patients require re-operation. Incomplete release of the flexor retinaculum is a frequent finding at re-operation. The management of recurrent carpal tunnel syndrome remains a clinical dilemma due in part to the lack of a gold standard for assessing the integrity of the flexor retinaculum after previous release.

Purpose: To assess the accuracy of non-invasive diagnostic imaging in determining the integrity of the flexor retinaculum after surgical release of the carpal tunnel.

Methods: In part I of the study, 15 cadaver hands underwent three sequential operations: palmar skin incision, partial division of the flexor retinaculum and complete division of the flexor retinaculum. T1 & T2 weighted MR images were obtained of each specimen after each procedure, and the pooled images assessed by two blinded radiologists. Sensitivity, specificity, intra-rater and inter-rater reliability were calculated for the diagnosis of complete or incomplete release using Spearman Rank Correlation and k statistic. Part two of the study consisted of a retrospective clinical review of patients with recurrent carpal tunnel syndrome who presented to a single Hand Surgeon's practice over a three year period. Only patients who underwent Ultrasound or MR imaging of the wrist followed by surgical exploration were included. The radiological diagnoses were compared to intra-operative findings to estimate sensitivity and specificity.

Results: Correct identification of retinacular integrity in the cadaver was obtained in 93% of specimens when based on combined T1 & T2 weighted images (sensitivity 94% and specificity 93%). Inter-rater reliability, measured by kappa statistic, was very good (k=0.71), while intra-rater reliability was excellent (1.0 & 0.8 for radiologist 1 and 2, respectively). Diagnoses based on T1 & T2 weighted images were significantly more accurate and reliable than those based on T1 images alone. Ten patients were included in the clinical pilot cohort. The accuracy of non-invasive imaging when compared with intra-operative findings was 70% overall (sensitivity=86%, specificity=33%).

Conclusions: MRI is sensitive, specific and reliable in the diagnosis of partially released flexor retinaculum in a cadaver model. Clinically, although the sample size was small, our pilot study suggests potential high diagnostic value of non-invasive imaging in the clinical diagnosis of incomplete release of the flexor retinaculum in cases of recurrent carpal tunnel syndrome.