Lung iodine mapping by subtraction with image registration allowing for tissue sliding

Brian Mohr; Monique Brink; Luuk J. Oostveen; Joanne D. Schuijf; Mathias Prokop

doi:10.1117/12.2216262

21 March 2016 Lung iodine mapping by subtraction with image registration allowing for tissue sliding

Brian Mohr, Monique Brink, Luuk J. Oostveen, Joanne D. Schuijf, Mathias Prokop

Proceedings Volume 9784, Medical Imaging 2016: Image Processing; 978442 (2016) https://doi.org/10.1117/12.2216262
Event: SPIE Medical Imaging, 2016, San Diego, California, United States

Abstract

Pulmonary embolism is a fairly common and serious entity, so rapid diagnosis and treatment has a significant impact on morbidity and mortality rates. Iodine maps representing tissue perfusion enhancement are commonly generated by dual-energy CT acquisitions to provide information about the effect of the embolism on pulmonary perfusion. Alternatively, the iodine map can be generated by subtracting pre- from post-contrast CT scans as previously reported. Although accurate image registration is essential, subtraction has the advantage of a higher signal-to-noise ratio and suppression of bone. This paper presents an improvement over the previously reported registration algorithm. Significantly, allowance for sliding motion at tissue boundaries is included in the regularization. Pre- and post-contrast helical CT scans were acquired for thirty subjects using a Toshiba Aquilion ONE scanner. Ten of these subjects were designated for algorithm development, while the remaining twenty were reserved for qualitative clinical evaluation. Quantitative evaluation was performed against the previously reported method and using publicly available data for comparison against other methods. Comparison of 100 landmarks in seven datasets shows no change in the mean Euclidean error of 0.48 mm, compared to the previous method. Evaluation in the publicly available DIR-Lab data with 300 annotations results in a mean Euclidean error of 1.17 mm in the ten 4DCT cases and 3.37 mm in the ten COPDGene cases. Clinical evaluation on a sliding scale from 1 (excellent) to 5 (non-diagnostic) indicates a slight, but non-significant, improvement in registration adequacy from 3.1 to 2.9.

Citation Download Citation

Brian Mohr, Monique Brink, Luuk J. Oostveen, Joanne D. Schuijf, and Mathias Prokop "Lung iodine mapping by subtraction with image registration allowing for tissue sliding", Proc. SPIE 9784, Medical Imaging 2016: Image Processing, 978442 (21 March 2016); https://doi.org/10.1117/12.2216262

ACCESS THE FULL ARTICLE