Real-Time Peripheral Revascularization Planning in Chronic Limb Threatening Ischemia Using HarVI: A Digital Twin Approach.
Authors
Affiliations (3)
Affiliations (3)
- Department of Biomedical Engineering, Duke University, 534 Research Dr., Durham, NC, 27705, USA.
- Department of Surgery, Duke University, 2301 Erwin Rd., Durham, NC, 27707, USA.
- Department of Biomedical Engineering, Duke University, 534 Research Dr., Durham, NC, 27705, USA. [email protected].
Abstract
Peripheral artery disease (PAD) is a leading cause of limb loss and morbidity worldwide, with chronic limb-threatening ischemia (CLTI) representing its most severe presentation. Although image-guided endovascular interventions are routinely performed, clinicians currently lack tools that provide real-time, patient-specific predictions of hemodynamic outcomes to guide revascularization decisions. Existing computational fluid dynamics (CFD) approaches can recover pre-operative hemodynamics but are typically too slow or insufficiently integrated into clinical workflows to support interactive, intraoperative planning. We extend HarVI (HARVEY Virtual Intervention), a previously established digital twin framework, to the peripheral circulation and evaluate its use for real-time prediction of postoperative blood flow in patients with superficial femoral artery (SFA) lesions. HarVI integrates one-dimensional CFD with machine learning to enable rapid assessment of patient-specific revascularization strategies. Key components include: (1) automated boundary condition tuning using patient-averaged and optimization-based approaches; (2) simulation of a wide range of endovascular interventions via a machine-learned surrogate model; and (3) validation of predicted postoperative hemodynamics against clinical duplex ultrasound measurements. Performance was evaluated retrospectively in a cohort of seven patients with SFA disease. HarVI accurately predicted postoperative peak systolic velocities and reproduced full 1D CFD results across a synthetic revascularization landscape. Surrogate model predictions closely matched high-fidelity simulations while enabling rapid exploration of intervention scenarios, supporting near-real-time evaluation of treatment options. These results establish HarVI as a promising digital twin platform for real-time, patient-specific intervention planning in PAD. By enabling rapid, data-driven prediction of postoperative hemodynamics, HarVI opens the door to interactive intraoperative decision support with the potential to improve revascularization outcomes in patients with CLTI.