Enhancing Privacy-Preserving Cancer Classification with Convolutional Neural Networks

Date: 2025-01-04

The paper presents OGHE, a CNN-based method for privacy-preserving cancer classification using Homomorphic Encryption (HE) to process encrypted genomic data securely. It introduces VarScout, a feature selection technique that preserves spatial patterns through clustering and occurrence analysis. OGHE is optimized for HE efficiency via a novel data-packing strategy, achieving high accuracy and reduced latency on the iDash 2020 dataset.

DOI: https://doi.org/10.1142/9789819807024_0040