Advanced photon counting CT imaging pipeline for cardiac phenotyping of apolipoprotein E mouse models

Abstract

Background Cardiovascular disease (CVD) is associated with the apolipoprotein E (APOE) gene and lipid metabolism. This study aimed to develop an imaging-based pipeline to comprehensively assess cardiac structure and function in mouse models expressing different APOE genotypes using photon-counting computed tomography (PCCT). Methods 123 mice grouped based on APOE genotype (APOE2, APOE3, APOE4, APOE knockout (KO)), gender, human NOS2 factor, and diet (control or high fat) were used in this study. The pipeline included PCCT imaging on a custom-built system with contrast-enhanced in vivo imaging and intrinsic cardiac gating, spectral and temporal iterative reconstruction, spectral decomposition, and deep learning cardiac segmentation. Statistical analysis evaluated genotype, diet, sex, and body weight effects on cardiac measurements. Results Our results showed that PCCT offered high quality imaging with reduced noise. Material decomposition enabled separation of calcified plaques from iodine enhanced blood in APOE KO mice. Deep learning-based segmentation showed good performance with Dice scores of 0.91 for CT-based segmentation and 0.89 for iodine map-based segmentation. Genotype-specific differences were observed in left ventricular volumes, heart rate, stroke volume, ejection fraction, and cardiac index. Statistically significant differences were found between control and high fat diets for APOE2 and APOE4 genotypes in heart rate and stroke volume. Sex and weight were also significant predictors of cardiac measurements. The inclusion of the human NOS2 gene modulated these effects. Conclusions This study demonstrates the potential of PCCT in assessing cardiac structure and function in mouse models of CVD which can help in understanding the interplay between genetic factors, diet, and cardiovascular health.

Deep learning models for rapid denoising of 5D cardiac photon-counting micro-CT images

“…The radiation dose was ~118 mGy. Projections were sorted into 10 different cardiac phases by a semi-automated, projection-domain intrinsic gating procedure [1]. All projections were reconstructed with the Multi-Channel Reconstruction (MCR) Toolkit [14].…”

“…Non-invasive, in vivo cardiac imaging of mice is critical for cardiovascular disease (CVD) research. Multi-energy (spectral) CT enables separation between materials, which is valuable for myocardial perfusion or atherosclerotic plaque imaging [1]. However, current use of energy-integrating detectors (EIDs) limits the performance of spectral CT. Photoncounting detectors (PCDs) can improve spectral CT. PCDs simultaneously acquire projections at two or more energies by sorting individual x-ray photons into bins based on energy thresholds [2].…”