Iodine and bismuth dual-contrast imaging on photon-counting CT with noise-optimized virtual monoenergetic imaging:...

Dual-contrast imaging of iodine and bismuth has shown promise in previous phantom and animal studies utilizing spectral CT. However, these studies often employed Pepto-Bismol containing bismuth subsalicylate with dosages exceeding the recommended limits for human consumption, limiting their clinical applicability.

The purpose of this study was to improve iodine-bismuth dual-contrast imaging on a clinical photon-counting CT (PCCT) system using a noise-optimized virtual monoenergetic imaging (VMI) approach.

Four bismuth samples (0.6, 1.3, 2.5, and 5.1 mg/mL) and three iodine rods (1.0, 2.0, and 5.0 mg/mL) were inserted into a multi-energy CT phantom and scanned on a PCCT system at 140 kV. VMIs were generated from 40 to 190 keV in 10 keV increments. Iodine and bismuth maps were computed for all possible keV pairs using a generic image-based three-material decomposition method and compared with maps generated from the original low-/high-energy PCCT images. Mass concentrations, noise levels, and individual as well as joint root-mean-square error (RMSE) for iodine and bismuth were measured to determine the optimal keV pair. The contrast-to-noise ratios (CNRs) of low concentration samples (bismuth: 0.6 and 1.3 mg/mL; iodine: 1.0 mg/mL) were calculated at the optimal keV pair and compared to the values determined with the original PCCT images. In an IRB-approved feasibility study, a research patient was administered two doses of Pepto-Bismol prior to a routine abdomen-pelvis CT scan. Material decomposition was performed using both the original low-/high-energy images and the optimal VMI pair identified in the phantom study. Bismuth signals in the stomach and proximal small bowel were assessed qualitatively and quantitatively.

The VMI pair of 110 and 190 keV was determined to be optimal, achieving the lowest joint RMSE of 0.49 mg/mL, compared to 1.42 mg/mL for the original low-/high-energy pair. It also yielded the highest CNR, representing a 190-318% improvement over the original low/high-energy decomposition. Qualitatively, both phantom and patient studies demonstrated improved iodine-bismuth separation and reduced noise using the optimized VMI pair.

These findings suggest that noise-optimized VMI selection can enhance dual-contrast material decomposition performance on clinical PCCT, supporting its potential for translation into gastrointestinal imaging applications.