Change assessment for CT spine imaging

Abstract

Efficient change quantification is a key issue in follow-up studies. For the support of spine follow up examinations in the context of osteoporosis, we successfully apply in this paper a model based vertebra assessment framework. Automated vertebra labeling and segmentation is accompanied by registration of each vertebra-pair in base-line and follow-up CT scan. The latter is achieved by a point-wise registration of the individualized model meshes. A surface displacement distance measure between pairs of vertebra is introduced and applied to detect interval changes caused by osteoporotic compression fractures of the vertebral body. The associated ROC curve is characterized by an area-under-curve value of 0.97, indicating a high classification power.