Improved Diagnostic Performance Using Dual-Energy CT-Derived Slope Parameter Images in Crohn's Disease.

The objective of the study is to explore the image quality and diagnosis performance of the dual-energy CT-derived slope parameter images (SPI) generated by the algorithm based on the slope function in the diagnosis of Crohn's disease (CD). Seventy-six CD patients and 53 disease-free control group subjects who underwent dual-energy CT enterography were retrospectively collected. Portal venous phase 120kVp-like and virtual monoenergetic images at 40-100 keV (VMI) were reconstructed. SPIs corresponding to the spectral curve between 40 and 100 keV (SPI) were generated using Python. Signal-to-noise ratio (SNR) and contrast-to-noise ratio (CNR) of normal and abnormal intestinal walls were calculated. Image quality, noise, and contrast were independently scored by two radiologists using a 5-point scale. Four radiologists conducted CD diagnosis with three reading models (120kVp-like, 120kVp-like with optimal VMI, and 120kVp-like with SPI). The diagnostic performances of the three reading models for diagnosing CD were evaluated using receiver operating characteristic (ROC) curves. The CNR in SPI was higher than in the other images (P < 0.05). The subjective evaluation showed that there was no statistical difference between the contrast of SPI and VMI (P > 0.05), but that of the two images was higher than the other images (P < 0.05). The scoring on the overall image quality of VMI was superior to that of other images (P < 0.05). The combined model of 120kVp-like with SPI had the strongest confidence (cases with high confidence: 36, 58, 49, 47 for radiologists 1, 2, 3, 4) and the highest efficiency in diagnosing CD (areas under the ROC curve: 0.973, 0.977, 0.982, 0.991 for radiologists 1, 2, 3, 4). SPI generated by the algorithm based on the slope function exhibited good image quality. The combined model of 120kVp-like with SPI could improve radiologists' diagnostic efficiency and confidence in diagnosing CD.