Notes on investigation of CT number accuracy and consistency. Questions and comments should be sent to:phil@philipfjudy.com
Two active applications:
1. Distinguish solid lesions from cysts in abdomen CT
The motivation for this activity is detailed in the introduction of the 2009 SPIE Medical Imaging Proceeding Paper. The motivation was that CT number of solid lesions in abdomen images from 64-channel CT scanners were so much lower as compared to CT number in images from 4-channel CT scanners that these potentially malignant lesions might be confused with benign cysts.
2. Quantitative evaluation of lung disease, particularly COPD
9-11-2008 COPDGene Background.ppt 11-6-2008 COPDGene Investigators Meeting - QA Phantom.ppt 11-18-2008 Truncation CT Number Air.ppt 12-5-2008 Compare Sensation 64 and Definition.ppt QIBA RSNA 2010FinalC.pdf 2011 QIBA/COPD Poster_RSNA-NIST.pdf 3. Background of CT Measurement of Lung Density The publications of Kemerink have been particularly useful. 4. Truncation Truncation Materials 5. Background and Articles of Possible Interest | Background and notes These are two papers that develop a theoretical basis for prediction of CT numbers from x-ray spectra. Derivation of the prediction : Effects of source X-ray filtration on dose and image performance of CT scanners Medical Physics 1979 Validation of prediction : Comparison of equivalent photon energy calibration methods in computed tomograpy Medical Physics 1980 These two previous papers deal with how x-ray energy spectrum effect the CT numbers. In 2009, I realized that many investigators believed that reconstruction filters may effect the CT numbers in a similar manner. I have always realized that the CT numbers were altered near edges, however I did not realize that "small" regions of interest "near " edge have been used. For the most part, the investigators felt that they were using "large" regions of interest "far" from edges. My interest in reconstruction filter variations was in the perceptual effects of reconstruction kernels of noise power spectrum (NPS) of images. The NPS is altered by the reconstruction filter. To be more specific I started using "kernel" rather that "filter" to describe my interests, because CT vendors may incorporate other processed into the label: "reconstruction filter." |
Home page >