Magnetic resonance in the identification and assessment of extent and severity of bone contusion damage: comparison of several types of imaging (SE/T1, GE/T2 and SE/T1) with fat suppression

INTRODUCTION: Bone marrow contusion is a common cause of chronic joint pain in the post-traumatic knee; it is usually associated with meniscal and/or ligament damage but it may also be the only abnormal sign. In this condition MRI plays a fundamental role because it is the only imaging technique which can diagnose it and accurately assess its extent and any cortical bone and/or cartilage involvement. We investigated the diagnostic potentials of 3 MR sequences in identifying and assessing the extent of bone contusion. MATERIAL AND METHODS: January, 1994, to December, 1997, we performed 539 MR examinations of the knee on 478 patients with chronic joint pain following trauma. We used T1-weighted SE, T2*-weighted GE and T1-weighted SE sequences with fat saturation (fs), paying great attention to technical parameter setting. The images were retrospectively reviewed by groups of radiologists using 2 semi-objective evaluation systems: a qualitative assessment to grade the ease of detection of the contusion with each sequence and a quantitative assessment of contusion extent and of osteochondral involvement, if any, again for each sequence. RESULTS: The qualitative analysis showed that GE and SE-fs sequences were superior to the conventional ones in identifying bone contusion foci; GE sequences tended to overestimate the number of poorly depictable lesions relative to SE-fs ones. The quantitative analysis showed that the conventional sequences tend to underestimate the foci extent, while GE and SE-fs sequences performed nearly the same in the assessment of high-grade lesions. The number of low-midgrade lesions was overestimated by GE sequences relative to SE-fs ones. DISCUSSION AND CONCLUSIONS: GE and SE-fs sequences permit accurate studies of bone contusions. Diagnostic efficacy is increased by enhancing the signal from non-fatty structures within a tissue rich in fat: the quantitative reduction of the non-fatty structural component and/or its replacement with abnormal components is thus best shown. Fat-suppression sequences were more accurate in defining lesion size, which is also related to low susceptibility to the artifacts from extrinsic and especially intrinsic structural inhomogeneities. In contrast, GE sequences demonstrated high susceptibility, which caused poorer definition of the contusion borders and, consequently, overestimation of the lesion size.