DECODE: An open-source cloud-based platform for the noninvasive management of peripheral artery disease

Area of Science:

• Medical Imaging
• Computational Biology
• Artificial Intelligence in Medicine

Background:

• Peripheral artery disease (PAD) affects over 237 million globally, necessitating improved diagnostic and treatment planning tools.
• Current management of PAD is often limited by access to advanced imaging and real-time analytical support.
• Accurate, patient-specific planning is crucial for effective PAD intervention.

Purpose of the Study:

• To introduce DECODE, an open-source, cloud-based platform designed to enhance the noninvasive management of PAD.
• To integrate artificial intelligence, 3D visualization, and computational modeling for improved PAD care.
• To provide a scalable infrastructure for personalized vascular care.

Main Methods:

• Developed a modular platform with deep learning for artery and intima-media segmentation (U-Net, nnU-Net).
• Implemented centerline extraction for quantitative vascular geometry analysis.
• Utilized nonlinear finite element models for balloon angioplasty simulations, calibrated with experimental data.

Main Results:

• Achieved high accuracy in peripheral artery segmentation (Dice 0.91) and intima-media segmentation (Dice 0.992).
• Finite element simulations accurately reproduced mechanical interactions, identifying key pressure and stress distributions.
• The platform demonstrated excellent usability (SUS score 87.5) and high user acceptance (TAM score 4.21/5).

Conclusions:

• DECODE offers an automated, cloud-integrated solution for PAD diagnosis and intervention planning.
• The platform facilitates precise vascular analysis, real-time simulation, and clinical decision support.
• DECODE enhances image processing, segmentation accuracy, and in-silico trials for personalized vascular care.