Project Summary/Abstract The objective of the project is to develop auto-calibrating quantitative dual-energy CT (AC-QDECT) through algorithm development for accurate, simultaneous reconstruction of images and spectra on cone-beam CT (CBCT). Quantitative DECT (QDECT) is a recognized technique of significant clinical potential for improving disease diagnosis and management, and is performed currently only on advanced diagnostic (Dx) CT in which spectra must be estimated as a prior from separate measurements. There is little effort yet in developing non- Dx QDECT systems such as CBCT for use in, e.g., surgery and radiation therapy (RT); a leading reason for this is that accurate spectra cannot be measured/estimated readily in experiments in current CBCT imaging. In CBCT with AC-QDECT capability proposed, both images and spectra are treated as unknowns on an equal footing, and we develop a non-convex primal-dual (NCPD) algorithm simultaneously to yield images and spectra only from data. We will also use the NCPD algorithm to enable AC-QDECT capability on CBCT with innovative partial scanning configurations of clinical application significance. Finally, we will evaluate AC- QDECT capability enabled on CBCT in extensive simulated- and real-data studies. The project hypothesis is that CBCT with AC-QDECT capability can be enabled by use of the NCPD algorithm for yielding quantitatively accurate images without spectra as a priori estimated from separate measurements. The specific aims of the project are (1) to develop the NCPD algorithm for enabling AC-QDECT capability on CBCT and (2) to prototype and evaluate the AC-QDECT capability proposed on CBCT. The project is built upon our success research on the development of QDECT and CBCT tailored to applications growing rapidly in surgery, RT, and orthopedics. The project outcome is the establishment of the feasibility of algorithm-enabled AC-QDECT capability on CBCT, substantially expanding the application domain and utility of CBCT.