Semi-automatic spinal cord segmentation and quantification

Delineation of objects within medical images is very hard to perform reproducibly, unless at least semi-automatic segmentation is available. This paper presents a semi-automatic process for determining the total length and area of the spinal cord, and the length and area of the spinal cord between adjacent pairs of intervertebral discs. We performed these measurements on 20 segments from C1 to L1 with three sets of user-selected landmarks. Our results show that the average spinal cord segment area was 852.3 mm2 with a standard deviation of 0.7%, and the average spinal cord segment length was 57.4 mm with a standard deviation of 0.3%. Our method has been demonstrated to be potentially insensitive to intra- and inter-user variability.