Deep-En-Chroma: mining the spectral fingerprints in single-kV CT acquisitions using energy integration detectors

CT imaging is one of the primary diagnostic medical imaging modalities. However, there is a long-standing technical limitation associated with conventional CT imaging: anatomical structures with different material compositions may have the same CT number, thereby limiting the ability to differentiate and classify different tissue types and contrast agents. To address this limitation, the currently available strategy is to modify the hardware acquisition systems such that dual energy CT (DECT) data acquisition scheme can be accommodated. In this work, we show that the elemental composition of a material can be directly extracted from a conventional single-kV CT acquisition without invoking DECT acquisition scheme.