In this manuscript we present an automated algorithm for the alignment of thoracic scans using descriptors of bone structures. Bone structures were utilized because they are expected to be less susceptible to sources of errors such as patient positioning and breath hold. The algorithm employed the positioning of ribs relative to the spinal cord along with a description of the scapula. The spinal cord centroid was detected by extracting local maxima of the distance transform followed by point tracing along consecutive slices. Ribs were segmented using adaptive thresholding followed by the watershed algorithm to detach ribs from the vertebra, and by imposing requirements of rib proximity to the lung border. The angles formed between the spinal cord centroid and segmented rib centroids were used to describe rib positioning. Additionally, the length of the scapula was extracted in each slice. A cost function incorporating the difference of features from rib positioning and scapula length between two slices was derived and used to match slices. The method was evaluated on a set of 12 pairs of full and partial CT scans acquired on the same day. Evaluation was based on whether the slices showing a nodule at its maximum diameter in each scan were matched. Full-to-partial and partial-to-full alignment were performed. Results showed that the proposed metric matched nodule slices within an average distance of 1.08 and 1.17 slices from the target for full-to-partial and partial-to-full alignment respectively. These preliminary results are encouraging for using this method as a first step in an overall process of temporally analyzing CT lung nodules.
[1]
Pierre Soille,et al.
Morphological Image Analysis: Principles and Applications
,
2003
.
[2]
Kyle J. Myers,et al.
Alignment of full and partial CT thoracic scans using bony structures
,
2006,
SPIE Medical Imaging.
[3]
Rafael Wiemker,et al.
Comparison of different follow-up lung registration methods with and without segmentation
,
2004,
SPIE Medical Imaging.
[4]
Krishna Subramanyan,et al.
Automatic matching of the pulmonary nodules in current and former CT studies: a clinical evaluation
,
2004,
CARS.
[5]
Margrit Betke,et al.
Landmark detection in the chest and registration of lung surfaces with an application to nodule registration
,
2003,
Medical Image Anal..
[6]
D. Naidich,et al.
Computer-aided diagnosis and the evaluation of lung disease.
,
2004,
Journal of thoracic imaging.
[7]
Lawrence Dougherty,et al.
Alignment of CT lung volumes with an optical flow method.
,
2003,
Academic radiology.