Privacy-Preserving SAM Quantization for Efficient Edge Intelligence in Healthcare

TL;DR

A data-free quantization framework for SAM is proposed, called DFQ-SAM, which learns and calibrates quantization parameters without any original data, thus effectively preserving data privacy during model compression and facilitating the pervasive application of artificial intelligence in worldwide healthcare.

Abstract

The disparity in healthcare personnel expertise and medical resources across different regions of the world is a pressing social issue. Artificial intelligence technology offers new opportunities to alleviate this issue. Segment Anything Model (SAM), which excels in intelligent image segmentation, has demonstrated exceptional performance in medical monitoring and assisted diagnosis. Unfortunately, the huge computational and storage overhead of SAM poses significant challenges for deployment on resource-limited edge devices. Quantization is an effective solution for model compression; however, traditional methods rely heavily on original data for calibration, which raises widespread concerns about medical data privacy and security. In this paper, we propose a data-free quantization framework for SAM, called DFQ-SAM, which learns and calibrates quantization parameters without any original data, thus effectively preserving data privacy during model compression. Specifically, we propose pseudo-positive label evolution for segmentation, combined with patch similarity, to fully leverage the semantic and distribution priors in pre-trained models, which facilitates high-quality data synthesis as a substitute for real data. Furthermore, we introduce scale reparameterization to ensure the accuracy of low-bit quantization. We perform extensive segmentation experiments on various datasets, and DFQ-SAM consistently provides significant performance on low-bit quantization. DFQ-SAM eliminates the need for data transfer in cloud-edge collaboration, thereby protecting sensitive data from potential attacks. It enables secure, fast, and personalized healthcare services at the edge, which enhances system efficiency and optimizes resource allocation, and thus facilitating the pervasive application of artificial intelligence in worldwide healthcare.

Figures (11)

• Figure 1: Illustration of data-free quantization. It synthesizes data based on information extracted from the pre-trained model, which is then used for quantization calibration. Since no original data is utilized, it effectively protects data privacy and security.
• Figure 2: Overview of the proposed DFQ-SAM. It retrieves the pre-trained model solely from the cloud, protecting data privacy by avoiding access to any original data. The model is then compressed using data-free quantization for deployment on various edge devices. It enables low-cost and reliable intelligent healthcare at the edge, allowing underdeveloped regions to equally benefit from artificial intelligence technology, thereby helping to alleviate the disparity in healthcare resources.
• Figure 3: Visualization of segmentation results on different datasets (256$\times$256 pixels). The proposed DFQ-SAM, which utilizes synthesized data for 4-bit quantization, consistently performs comparable to full-precision baseline. See Fig. \ref{['fig:app_result_vis']} for more samples.
• Figure 4: Illustration of pseudo-label evolution and synthesized image updating (256$\times$256 pixels). Label evolution and image updating are conducted alternately: The label continuously discovers new regions based on the image outputs, which in turn further drives the image optimization. Ultimately, both converge at a high semantic level. See Fig. \ref{['fig:app_label_vis']} for more samples.
• Figure 5: Visualization of synthesized images (256$\times$256 pixels). The first two rows are grayscale images and the third row is color images, both reflecting medical semantic contents.