Volumetric reconstruction: matching and merging in DICOM data

Evolving protocols in Whole Body MRI (WB-MRI) present new challenges in the application of automated image processing and analysis procedures. We address two issues relating to sectionally acquired coronal WBMRI: greyscale non-uniformity correction and accurate data merging. The body is scanned in a number of slightly overlapping coronal sections, each containing 32 eight mm thick slices. Adjacent coronal sub-section often demonstrate a significant greyscale mismatch resulting in sharp discontinuities within homogeneous tissue regions in the body. This presents problems for automatic segmentation and classification approaches. Similarly a simplistic reconstruction procedure can lead to inaccuracies in the relative alignment of the data from each DICOM image in the reconstructed volume, resulting in the dislocation of small structures