Diagnostic reliability and interobserver agreement in T1-weighted phase sensitive inversion recovery sequence for detection of cervical cord demyelinating plaques

Abstract

PurposeTo evaluate the diagnostic reliability of PSIR sequence as compared to STIR in detection & counting of MS plaques in cervical cord and assess inter and intra-observer agreement. Patient and methodsA retrospective analysis of cervical MRI of 39 patients with Multiple sclerosis; Phase sensitive inversion recovery (PSIR) & short time inversion recovery (STIR) sequences were analyzed by 2 readers twice with 2 weeks interval for plaque detection, number and lesion conspicuity. ResultsMean conspicuity of lesions in PSIR and STIR was (3.4, 3.1 and 3.1, 2.8) for R1 and R2 without significant statistical difference (p = 0.18, 0.11). There was substantial inter-observer agreement between R1 and R2 regarding number of lesions in STIR and PSIR (K = 0.7, 0.72), almost perfect intra-reader observer agreement for STIR and PSIR (K = 0.85, 0.87 for R1, 0.8, 0.85 for R2). No statistical difference between number of lesions detected in STIR and PSIR by 2 readers (P = 0.5, 0.4). PSIR had higher sensitivity, specificity and accuracy compared to STIR (88.4/81.2, 98/96.1, 95/91.4) yet no statistical difference in accuracy (p value = 0.13). ConclusionPSIR sequence is accurate and reproducible in detection of MS lesions; it has higher sensitivity, specificity and accuracy than STIR sequence.