Room: Exhibit Hall | Forum 6
Purpose: To present a feasibility study for the decomposition of dual-energy (DE) images into equivalent thickness of basis materials.
Methods: A custom-built phantom to evaluate bone suppression capabilities using a DE imaging system with an on-board imager (OBI) is used for this study. The phantom consists of 16 cm lung-equivalent material sandwiched between 2 cm tissue-equivalent slabs. Five simulated tumors (0.5 â€“ 2.5 cm) located at two different depths are encased in the lung-equivalent materials and are allowed to be overshadowed with bone-equivalent material inserts to simulate ribs. Soft tissue images were created using a) a weighted logarithmic subtraction, where the weight was selected to suppress bony anatomy; and b) through a material decomposition algorithm that uses attenuation coefficients for aluminum and acrylic to produce thickness equivalent map for each material. Soft tissue images using both methods were created for 19 different energy pairs (high: 100-140 kVp; low: 50-90 kVp with 10 kVp increments) using the OBI of a commercial linear accelerator. Contrast-to-noise (CNR) values were computed for the largest target and the two methods were compared.
Results: For all energy pairs, our analysis demonstrated that the CNR values agree for two methods within statistical error. The highest value of CNR was achieved for 140-60 kVp energy pair with 2.94 Â± 0.19 and 2.90 Â± 0.18 for weighted logarithmic subtraction and material decomposition methods, respectively.
Conclusion: We present an algorithm to successfully decompose DE images into equivalent thickness of basis materials. The CNR values of simulated tumors obtained with tissue decomposition are comparable to the weighted logarithmic subtraction method. In addition, tissue decomposition allows one to produce virtual monoenergetic (VM) images by combining the thickness equivalent maps with corresponding attenuation coefficients. Such images may eventually be used in a rotational geometry to produce VM CBCT images.
Funding Support, Disclosures, and Conflict of Interest: Supported by grant from Varian Medical Systems.