From High-SNR Radar Signal to ECG: A Transfer Learning Model with Cardio-Focusing Algorithm for Scenarios with Limited Data

TL;DR

This work tackles radar-based ECG recovery in data-scarce scenarios by introducing a cardio-focused signal acquisition and transfer-learning pipeline. The CFT algorithm localizes and tracks the cardiac region to extract high-SNR radar signals, while RFcardi employs self-supervised pre-training on radar data with a sparse-recovery objective to learn cardio representations, enabling ECG recovery with only a few radar-ECG pairs for fine-tuning. Experimental results across indoor scenarios demonstrate improved SNR, robust performance under range and posture changes, and superior transfer-learning performance compared with SSL baselines. The framework promises practical deployment for remote health monitoring, with future work addressing real-time processing, energy efficiency, and multi-target monitoring.

Abstract

Electrocardiogram (ECG), as a crucial find-grained cardiac feature, has been successfully recovered from radar signals in the literature, but the performance heavily relies on the high-quality radar signal and numerous radar-ECG pairs for training, restricting the applications in new scenarios due to data scarcity. Therefore, this work will focus on radar-based ECG recovery in new scenarios with limited data and propose a cardio-focusing and -tracking (CFT) algorithm to precisely track the cardiac location to ensure an efficient acquisition of high-quality radar signals. Furthermore, a transfer learning model (RFcardi) is proposed to extract cardio-related information from the radar signal without ECG ground truth based on the intrinsic sparsity of cardiac features, and only a few synchronous radar-ECG pairs are required to fine-tune the pre-trained model for the ECG recovery. The experimental results reveal that the proposed CFT can dynamically identify the cardiac location, and the RFcardi model can effectively generate faithful ECG recoveries after using a small number of radar-ECG pairs for training. The code and dataset are available after the publication.

Figures (13)

• Figure 1: Challenges for radar-based ECG recovery: (a) and (b) Radar signals with high and low SNR for adjacent points with a distance of $0.03$m; (c) and (d) Inference results of a well-trained deep learning model in original and new scenarios.
• Figure 2: Overview of the CFT-RFcardi framework: (a) Rough localization of human body; (b) Use CFT to find CF point and extract high-SNR radar signals; (c) Transfer learning with pre-text task training and fine-tuning stages.
• Figure 3: Procedures for obtaining RA map: (a) Range FFT for chirps along fast time; (b) Angle FFT along virtual channels.
• Figure 4: Template for assessing SNR: (a) High-SNR radar signal; (b) Extracted signal envelope with the synthetic template; (c) (a) Low-SNR radar signal, (d) Extracted signal envelope with the synthetic template.
• Figure 5: Illustration of the CFT algorithm with bold line wrapping the search region $S_k$: (a) Equality between $\gamma$ and $\Gamma$ (same as in CS algorithm); (b) Large $\Gamma_k$ with refined $\gamma_k$, providing more potential points to be evaluated; (c) Jump out of the local minimum by adjusting $\Gamma_k$ and $\gamma_k$.