Performance of a rapid two-sequence screening protocol for osteomyelitis of the foot.

Abstract

Compare a two sequence protocol to a standard protocol in the detection of pedal osteomyelitis (OM) and abscesses and to identify patients that benefit from a full protocol. One hundred thirty-two foot MRIs ordered to assess for OM were enrolled, and the following items were extracted from the clinical reports: use of IV contrast, the presence of OM, reactive osteitis, and a soft tissue abscess. Using only one T1 nonfat-suppressed and one fluid sensitive fat-suppressed sequences, two experienced musculoskeletal radiologists reviewed each case for the presence of OM, reactive osteitis, or an abscess. A Kappa test was calculated to assess for interobserver agreement, and diagnostic performance was determined. The McNemar test was used to assess for the effect of contrast. Agreement between both observers and the clinical report on the presence of osteomyelitis was substantial ( k= 0.63 and 0.72, p < 0.001), while the agreement for abscess was fair (k = 0.29 and 0.38, p < 0.001). For osteomyelitis, both observers showed good accuracy (0.85 and 0.86). When screening bone for a normal versus abnormal case, this method was highly sensitive (0.97-0.98), but was less sensitive for abscess (0.63-0.75). Fifty-one percent of exams used contrast, and it did impact the diagnosis of abscess for one observer. This rapid protocol is accurate in making the diagnosis of OM, and its high sensitivity makes it useful to screen for patients that would benefit from a full protocol.