Abstract Introduction Computed Tomography (CT) pulmonary angiography is central to the evaluation, prognostication and management of chronic thromboembolic pulmonary hypertension (CTEPH). Increasingly, automated methods are available to quantify structures from CT imaging to develop panels of individualized markers that may be integrated into clinical practice, particularly using AI powered modeling. In this study we examined the relationship between volumes of heart chamber and the main pulmonary artery and hemodynamics in a cohort of patients with CTEPH. integral part of both the diagnosis and treatment of patients with thromboembolic disease. Methods Patients diagnosed with chronic thromboembolic pulmonary hypertension (CTEPH) who underwent pulmonary thromboendarterectomy (PTE) at our institution between 3/2016 and 11/2024 were included. Clinical and hemodynamic data, pre-operatively was collected. The most recent CT pulmonary angiogram prior to PTE was utilized. Cardiac structures and pulmonary artery were segmented using the TotalSegmentator, a publicly available validated software for automated segmentation of CT imaging. Right ventricular volumes (RVV), right atrial volumes (RAV), left atrial volumes (LAV), and total pulmonary arterial volumes (TPA) were computed. Additionally, RAV and RVV were divided by LAV as a normalization. Pearson correlation coefficients were calculated and implemented in R 4.5.2. Results 54 patients were used in this preliminary analysis. RA dilation was associated with increased baseline mean pulmonary arterial pressure (mPAP, R = 0.31, p = 0.002), a correlation that improved when normalizing by left atrial volume (RA/LA, R = 0.42, p = 0.002). Right ventricular volume (RVV) was also correlated with mPAP (R = 0.39, p = 0.005), though normalizing by LVV did not improve this correlation (R = 0.31, p = 0.03). However RVV/LAV showed a stronger correlation (R = 0.47, p = 0.0006). Similar pattern of findings showed the strongest correlation between pulmonary vascular resistance (PVR) and RVV/LAV (R = 0.55, p 0.0001) as well as between RVV/LAV and transpulmonary gradient (R = 0.50, p = 0.0002). Of note, Left atrial volume itself was negatively correlated with PVR (R=-0.34, p = 0.02). Discussion Automated segmentation tools can yield quantitative segmentations that can be generated in real-time for clinical evaluation. In the assessment of CTEPH we found that right ventricular volume normalized by left atrial volume showed the strongest hemodynamic correlation. This is likely due to the interplay of RV dilation with disease severity as well as poor left atrial filling due perhaps to lack of flow or compression by the right atrium, making this ratio a potentially strong prognostic marker. This abstract is funded by: R01HL164717
Rahaghi et al. (Fri,) studied this question.