Abbreviated MRI of the foot in patients with suspected osteomyelitis.

Purpose: To determine the diagnostic accuracy of an abbreviated magnetic resonance imaging (MRI) protocol of the foot for the diagnosis of osteomyelitis in patients with acute foot infection.

Methods: This retrospective study evaluated adult patients (age 18 and over) visiting an academic medical center from 1 January 2013 to 31 December 2015 who were imaged with MRI for suspected acute pedal osteomyelitis. Examinations were performed utilizing the departmental standard protocol. All examinations were retrospectively interpreted by five radiologists under two protocols: a reference standard protocol consisting of all non-contrast sequences obtained at initial acquisition and an abbreviated protocol consisting of only coronal T1-weighted and sagittal T2-weighted fast multiplanar inversion-recovery (FMPIR) sequences. Interpretation of the two imaging subsets was separated in time by at least 6 weeks for each reader. Each examination was assigned a score to represent one of four diagnostic categories: normal; soft tissue infection without bone changes or bone changes specific to a non-infectious etiology; nonspecific bone marrow changes; or bone changes specific for osteomyelitis. Diagnostic accuracy of both protocols was determined based on clinical diagnosis and treatment of osteomyelitis, and histopathology when available.

Results: One hundred and two MRI examinations met inclusion criteria; participants ranged in age from 26 to 91 years, with a mean age of 59 years. Seventy examinations were performed for male participants (69%) and 32 for female participants (31%). Thirty-five had a confirmed diagnosis of osteomyelitis, while the remainder (n = 67) did not. An average of 6 non-contrast sequences was performed during each examination. The most common protocol (53/102 examinations) was comprised of the following 6 sequences: axial T1-weighted, axial fat-saturated proton density, sagittal T1-weighted, sagittal T2-weighted FMPIR, coronal T1-weighted, and coronal fat-saturated proton density. After patient positioning, the abbreviated protocol sequences (sagittal T2-weighted FMPIR and coronal T1-weighted) were performed in an average total of 8 min. The reference standard protocol required an average of 22 min to complete 6 sequences. Averaged across all readers, the AUC for the reference standard full protocol and the abbreviated protocols were 0.843 and 0.873, respectively. The difference in AUC between protocols was not statistically significant (p = 0.1297), with the abbreviated protocol showing a non-significantly greater AUC.

Conclusions: An abbreviated MRI protocol, including only coronal T1-weighted and sagittal T2-weighted FMPIR images, is non-inferior to standard MRI protocol for the diagnosis of acute pedal osteomyelitis. It should be considered as a diagnostic alternative for reducing imaging time and improving patient access to MRI.