Clinical utility of artificial intelligence-assisted computed tomography volumetry and the digital vital capacity index in pulmonary hypertension risk stratification

Background Conventional spirometry revolves around pulmonary function but often provides a misleading representation of lung anatomy in patients with pulmonary hypertension (PH). We hypothesized that artificial intelligence (AI)-computer tomography (CT) volumetry provides a more accurate anatomical reference than spirometry, and we aimed to validate the Digital Vital Capacity Index (dVCI) as a prognostic marker. Methods We evaluated 90 PH patients (including idiopathic pulmonary arterial hypertension - IPAH and chronic thromboembolic pulmonary hypertension - CTEPH subgroups) using a multimodal approach: right heart catheterization, 6-minute walking test, and chest CT with automated AI-segmentation (Siemens Syngo.via). The dVCI was calculated as an integrated metric of exercise tolerance, heart rate response, and spirometric data. Results Automated CT volumetry revealed a profound overestimation of lung capacity by standard spirometry (p<0.001). Specifically, anatomical total lung capacity was lower than spirometric values by 48.0% in IPAH and 52.5% in CTEPH. Also, predictive accuracy for the volume-anatomy mismatch showed significant individual variance in CTEPH cases. In IPAH, traditional vital capacity (%) showed limited correlation with clinical status, but the dVCI exhibited strong (p<0.05) associations with key prognostic markers, such as NT-proBNP (r = -0.531), TAPSE (r = 0.427), mPAP (r = -0.423), highlighting dVCI as a potential non-invasive tool for longitudinal monitoring of IPAH progression. Conclusions In PH management, AI-driven CT volumetry exposes the limitations of traditional spirometry. The dVCI serves as a superior surrogate marker, reflecting both haemodynamic severity and right heart strain in IPAH more accurately than standard pulmonary function parameters.Figure1_Bland_Altman_VC.