Objective: To compare accuracy of ultrasound and MRI for detecting focal peripheral nerve pathology, excluding idiopathic carpal or cubital tunnel syndromes. Methods: We performed a retrospective review of patients referred for neuromuscular ultrasound to identify patients who had ultrasound and MRI of the same limb for suspected brachial plexopathy or mononeuropathies, excluding carpal/cubital tunnel syndromes. Ultrasound and MRI results were compared to diagnoses determined by surgical or, if not performed, clinical/ electrodiagnostic evaluation. Results: We identified 53 patients who had both ultrasound and MRI of whom 46 (87%) had nerve pathology diagnosed by surgical (n = 39) or clinical/electrodiagnostic (n = 14) evaluation. Ultrasound detected the diagnosed nerve pathology (true positive)more often than MRI (43/46 vs 31/46, p<0.001). Nerve pathology was correctly excluded (true negative) with equal frequency by MRI and ultrasound (both 6/7). In 25%(13/53), ultrasound was accurate (true positive or true negative) when MRI was not. These pathologies were typically (10/13) long (>2 cm) and only occasionally (2/13) outside the MRI field of view. MRI missed multifocal pathology identified with ultrasound in 6 of 7 patients, often (5/7) because pathology was outside the MRI field of view. Conclusions: Imaging frequently detects peripheral nerve pathology and contributes to the differential diagnosis in patients with mononeuropathies and brachial plexopathies. Ultrasound is more sensitive than MRI (93% vs 67%), has equivalent specificity (86%), and better identifies multifocal lesions than MRI. In sonographically accessible regions ultrasound is the preferred initial imaging modality for anatomic evaluation of suspected peripheral nervous system lesions.

Original languageEnglish
Pages (from-to)1634-1640
Number of pages7
Issue number18
StatePublished - Apr 30 2013


Dive into the research topics of 'Detection of peripheral nerve pathology: Comparison of ultrasound and MRI'. Together they form a unique fingerprint.

Cite this