(Sunday, 7/12/2020) [Eastern Time (GMT-4)]
Room: AAPM ePoster Library
Purpose: addition to virtual monochromatic analysis, spectral imaging in photon-counting CT can be implemented via principal component analysis, due to facilitation in energy binning (spectral channelization). In this work, using the data simulated by computer and acquired by a prototype photon-counting CT, we conduct a quantitative study to compare their performance and potential towards clinical and preclinical applications.
Methods: four energy bins (spectral channels), the principal component analysis is carried in both projection and image domains, while the virtual monochromatic analysis is implemented with two- and three-material decomposition, respectively. Two phantoms (iodine; and iodine and nanoparticulated gold) and a lab mouse are used to evaluate and compare their performance, in which target and surrounding background are adequately chosen to evaluate the contrast-to-noise ratio (CNR) in the first principal component image and its comparison to its counterpart in conventional polychromatic CT image, the image corresponding to each energy bin, and the best scenario case in virtual monochromatic imaging.
Results: study shows that principal component analysis can be implemented in both projection and image domains. Moreover, the data standardization strategy that is usually used for principal component analysis in other data sciences damages the consistence in projection data that is essential for tomographic image reconstruction, but does not impact the principal component analysis in the image domain. Both phantom and animal studies show that the CNR in the first principal component image is better than its counterpart in the image of each energy bin, the entire source spectrum, and roughly comparable to the best scenario in virtual monochromatic analysis.
Conclusion: comparison to virtual monochromatic analysis, the principal component analysis is an alternative approach for spectral imaging in photon-counting CT, due to its advantages in CNR and the capability of imaging in true color that can make pathology more discernible to an observer.
Funding Support, Disclosures, and Conflict of Interest: X. Tang is a recipient of research grant from Sinovision Technologies, Inc.
Keywords
Not Applicable / None Entered.
Taxonomy
Not Applicable / None Entered.
Contact Email
