Introducing GPGPUs to smartphone-based digital holographic microscope for 3D imaging

2025-03-17

AI Summary

To address the poor real-time performance and limited portability of smartphone-based digital holographic microscopy (DHM), this work proposes the first mobile-oriented, general-purpose GPU (GPGPU)-accelerated real-time reconstruction framework. The system integrates a 3D-printed miniature optical module, a smartphone CMOS sensor, and a laser source, and implements an Android-optimized GPU-parallelized phase retrieval algorithm based on the angular spectrum method—enabling single-shot, non-contact, three-dimensional imaging of transparent specimens. Compared to CPU-based reconstruction, the GPGPU acceleration achieves a 1.65× improvement in frame rate, supporting real-time amplitude and phase video streaming. This is the first demonstration of deep GPGPU integration into an on-device holographic microscope, achieving a compact form factor (<15 cm³), low cost (<$200), and field-deployability. The framework provides a practical, portable 3D microscopic solution for telemedicine and point-of-care biological analysis.

Technology Category

• Computer Vision: 3D Computer Vision
• Machine Learning: Hardware-aware ML
• Search and Optimization: Distributed Search

Application Category

• User Modeling, Personalization and Recommendation: On-Device user modeling, personalization, and recommendation
• Systems and Infrastructure for Web, Mobile and WoT: Applied ML and AI for Web-based mobile applications
• Security and Privacy: Data transparency and provenance

Abstract

Digital holography (DH) enables non-contact, noninvasive 3D imaging of transparent and moving microscopic samples by capturing amplitude and phase information in a single shot. In this work, we present a compact, low-cost, real-time smartphone-based DHM system accelerated by GPUs. The system comprises a 3D-printed optical system using readily available image sensors and lasers, coupled with an Android app for hologram reconstruction, extracting amplitude and phase information. Results show a frame rate improvement of approximately 1.65x compared to a CPU-only system. This inexpensive, compact DHM, combining 3D-printed optics and smartphone-based reconstruction, offers a novel approach compared to existing systems and holds promise for fieldwork and remote diagnostics.

Problem

• Develops a smartphone-based digital holographic microscope for 3D imaging.
• Utilizes GPGPUs to enhance real-time hologram reconstruction speed.
• Creates a low-cost, compact system for fieldwork and remote diagnostics.

Innovation

• GPGPU-accelerated smartphone-based digital holographic microscope
• 3D-printed optics with readily available components
• Real-time hologram reconstruction via Android app