Securing Clinical AI Pipelines: From DICOM Ingestion to Model Inference
[03 Parts]A 3-part technical monograph examining end-to-end threat vectors in hospital AI deployments: defending 3D volumetric vision models against adversarial perturbations, sandboxing untrusted inference runtimes with eBPF, and preventing gradient leakage in federated training.
Adversarial Robustness in 3D Volumetric Medical Imaging: Defending Vision Transformers Against Imperceptible Perturbations
Volumetric 3D Swin-UNETR segmentation models achieve clinician-level Dice scores on multi-modal brain MRI, yet remain brittle under L-infinity perturbations below sensor thermal noise. We derive Lipschitz-constrained volumetric attention and evaluate certified smoothing bounds.
Memory-Safe Sandboxing & eBPF Runtime Monitoring for Untrusted Clinical AI & Agent Execution
Clinical AI pipelines and autonomous security agents routinely parse untrusted binary formats (DICOM, NIfTI, ONNX, safetensors) and invoke native C/C++ shared libraries. We design a zero-trust Linux sandbox combining Rust Landlock LSM with Linux Security Module (LSM) eBPF ring-buffer enforcement.
Privacy-Preserving Federated Learning in Hospital Networks: Mitigating Gradient Inversion Attacks
Sharing raw parameter gradients in multi-hospital federated learning does not guarantee patient privacy: optimization-based gradient inversion can reconstruct diagnostic chest X-rays and volumetric slices. We derive tight Rényi Differential Privacy (RDP) guarantees with adaptive gradient clipping.