Data privacy is a major concern when accessing and processing sensitive medical data. A promising approach among privacy-preserving techniques is homomorphic encryption (HE), which allows for computations to be performed on encrypted data. Currently, HE still faces practical limitations related to high computational complexity, noise accumulation, and sole applicability the at bit or small integer values level. We propose herein an encoding method that enables typical HE schemes to operate on real-valued numbers of arbitrary precision and size. The approach is evaluated on two real-world scenarios relying on EEG signals: seizure detection and prediction of predisposition to alcoholism. A supervised machine learning-based approach is formulated, and training is performed using a direct (non-iterative) fitting method that requires a fixed and deterministic number of steps. Experiments on synthetic data of varying size and complexity are performed to determine the impact on runtime and error accumulation. The computational time for training the models increases but remains manageable, while the inference time remains in the order of milliseconds. The prediction performance of the models operating on encoded and encrypted data is comparable to that of standard models operating on plaintext data.

Deep learning (DL)-based algorithms have demonstrated remarkable results in potentially improving the performance and the efficiency of healthcare applications. Since the data typically needs to leave the healthcare facility for performing model training and inference, e.g., in a cloud based solution, privacy concerns have been raised. As a result, the demand for privacy-preserving techniques that enable DL model training and inference on secured data has significantly grown. We propose an image obfuscation algorithm that combines a variational autoencoder (VAE) with random non-bijective pixel intensity mapping to protect the content of medical images, which are subsequently employed in the development of DL-based solutions. A binary classifier is trained on secured coronary angiographic frames to evaluate the utility of obfuscated images in the context of model training. Two possible attack configurations are considered to assess the security level against artificial intelligence (AI)-based reconstruction attempts. Similarity metrics are employed to quantify the security against human perception (structural similarity index measure and peak signal-to-noise-ratio). Furthermore, expert readers performed a visual assessment to determine to what extent the reconstructed images are protected against human perception. The proposed algorithm successfully enables DL model training on obfuscated images with no significant computational overhead while ensuring protection against human eye perception and AI-based reconstruction attacks. Regardless of the threat actor’s prior knowledge of the target content, the coronary vessels cannot be entirely recovered through an AI-based attack. Although a drop in accuracy can be observed when the classifier is trained on obfuscated images, the performance is deemed satisfactory in the context of a privacy–accuracy trade-off.

In recent years, computational fluid dynamics (CFD) has become a valuable tool for investigating hemodynamics in cerebral aneurysms. CFD provides flow-related quantities, which have been shown to have a potential impact on aneurysm growth and risk of rupture. However, the adoption of CFD tools in clinical settings is currently limited by the high computational cost and the engineering expertise required for employing these tools, e.g., for mesh generation, appropriate choice of spatial and temporal resolution, and of boundary conditions. Herein, we address these challenges by introducing a practical and robust methodology, focusing on computational performance and minimizing user interaction through automated parameter selection. We propose a fully automated pipeline that covers the steps from a patient-specific anatomical model to results, based on a fast, graphics processing unit- (GPU-) accelerated CFD solver and a parameter selection methodology. We use a reduced order model to compute the initial estimates of the spatial and temporal resolutions and an iterative approach that further adjusts the resolution during the simulation without user interaction. The pipeline and the solver are validated based on previously published results, and by comparing the results obtained for 20 cerebral aneurysm cases with those generated by a state-of-the-art commercial solver (Ansys CFX, Canonsburg PA). The automatically selected spatial and temporal resolutions lead to results which closely agree with the state-of-the-art, with an average relative difference of only 2%. Due to the GPU-based parallelization, simulations are computationally efficient, with a median computation time of 40 minutes per simulation.