Background Bone erosions are a characteristic of inflammatory joint disease although they are also found in the healthy elderly population. In order to correctly discriminate pathological bone erosions from normal physiological bone changes, it is important to use a high resolution imaging techniques. Hr-pQCT offers higher spatial resolution but whole body CT scanners are more widely available and offer faster scans times, larger anatomic coverage and improved patient comfort. So far a golden standard has not been established.
Objectives To compare the performance of two imaging modalities – whole body clinical CT scanners (wbCT) versus high resolution peripheral QCT (hr-pQCT) to detect bone erosions and to quantify erosion size.
Methods Left and right hands of 12 human cadavers (7 females, age 65-90y, mean 89y, cause of death diseases related to internal medicine or oncology) were scanned twice (with repositioning) with wbCT (Siemens Somatom Flash) using an extremity protocol (120 kV, 120 mAs, wrist UHR, U70, 0.5 mm slice thickness, FoV 135mm) and once with hr-pQCT (XtremeCT, Scanco Medical, Switzerland, isotropic voxel size of 80μm). The 300-400 slices of the hr-pQCT scan covered the metacarpal-phalangeal joints. Erosion detection by an expert rheumatologist was performed independently for both imaging modalities. Volume of the erosions detected in the hr-pQCT data sets was measured using MIAF-Finger. In short, first periosteal bone surface and then the volume of interest (VOI) covered by an erosion was segmented in the hr-pQCT data. In a second step the periosteal bone surface was rigidly registered to the wbCT scans. Erosions at corresponding locations were segmented independently in the wbCT scans to compare erosion volume between both modalities.
Results 3 hr-pQCT scans were excluded from analysis due to inadequate scan range. In 11 cadavers a total of 93 erosions were found in the HR-pQCT scans. Nine of these could not be segmented in at least one of the two wbCT scans due to small volume or small cortical break. For the remaining 85 erosions correlation of volume between the repeat wbCT scans and between wbCT and hr-pQCT is shown in the figure.
Conclusions It is unknown which erosions were related to inflammation. Correlation of erosion volume between the two modalities was high, although as expected wbCT underestimated volume by about 30%. Large discrepancies in size were detected for example if two erosions close together but still separated appeared as one larger erosion in the lower resolution wbCT images. Detectability of erosions still has to be compared between the two modalities. The preliminary results however are promising that by applying high-resolution extremity protocol, wbCT may be used for erosion detection, if HR-pQCT is not available.
Disclosure of Interest None declared