Validation of MRI as a tool in RA has been undertaken in a variety of ways. These include evaluating biopsies of MRI bone erosions (obtained under ultrasound guidance), correlating the MRI bone erosions with ultrasound erosions (and radiographs), and conducting longitudinal studies to assess the long-term outcome of MRI studies.
Ultrasound and MRI modalities measure different physical characteristics of erosions, with ultrasound identifying the absence of a cortical echo and MRI detecting change in nuclear spin in response to a magnetic field. When MRI lesions are compared with ultrasound erosions, the two modalities act as a cross-reference for each other . Studies undertaking these comparisons have shown that the cortical defects observed from ultrasound are closely correlated with MRI abnormalities on T1-weighted images  with all MRI lesions correlating with a cortical defect on ultrasound.
In one study, the presence of a MRI lesion was confirmed on ultrasound and then, under ultrasound guidance, tissue was obtained from the confirmed MRI erosion site . Five patient biopsies were obtained, all of which contained bone, and three of which were also associated with cellular material. Furthermore, one patient biopsy had a distinct population of CD34+ cells. The close correlation between ultrasound and MRI-detected bone erosions thus strongly supports the conclusion that MRI is measuring genuine abnormalities.
The sensitivity of MRI for detecting bone erosions has also been compared with radiography. In one comparison study, radiographic analysis detected bone erosions in 15% of RA patients (median symptom duration, 4 months), whereas MRI analysis detected erosions in 45% of RA patients . Furthermore, longitudinal studies confirmed that lesions seen with MRI appear later as radiographic erosions. RA patient follow-up over 5 years showed that, at baseline, only 20% of MRI lesions were detected by radiographic analysis, while at 5 years 60% of MRI lesions were detected by radiographic analysis . This reflects both the lack of sensitivity of radiography and the delay in the visualization of lesions using this technique. One potential explanation for this discrepancy may be that, at presentation, the majority of lesions detected are small in size; later, as the disease progresses, the lesions increase in size. Unfortunately, of all the bone erosions detected by radiography, only approximately 10% of these identified lesions were classified as small in size [6, 9]. Thus, in addition to lacking the capacity for multi-planar imaging, radiographs do not detect early bone erosions because early RA is characterized by small erosions, which are below the threshold of radiographic detection.