Embedded GPU Platform Enables Rapid Blood Profiling for POC Diagnostics

Now, scientists have developed a real-time, low-cost system that reconstructs and analyzes blood cell images using quantitative phase microscopy (QPM).

This breakthrough by scientists at Duke University’s BIOS Lab (Durham, NC, USA;) could make rapid, high-throughput blood diagnostics feasible at the point of care. QPM leverages optical holography to visualize and measure the 3D shape, thickness, and size of individual blood cells without dyes. The technique can detect diseases that alter red blood cell (RBC) morphology, such as sickle cell disease (SCD), and has the potential to process more than 100,000 cells in under three minutes.

However, the challenge lies in digitally reconstructing and analyzing this large dataset. Traditional CPU-based reconstruction can take hours, while GPU-based methods, though faster, are prohibitively expensive for routine use. To overcome these limitations, the Duke BIOS team designed a real-time processing pipeline that reconstructs and analyzes high-throughput QPM data at a rate of 1,200 cells per second using an affordable NVIDIA Jetson Orin Nano, an embedded GPU platform costing just USD 249. This innovation balances speed, cost, and accuracy, making QPM practical for clinical environments.

Integrated with a high-throughput QPM imaging system, the pipeline automates every step—from segmenting individual cell images to digital refocusing and calculating morphological parameters such as volume and projection area—without manual input. In validation tests using polystyrene beads and healthy RBC samples, the system achieved results with less than 5% average error, demonstrating precision comparable to conventional processing methods. Their findings were published in Biophotonics Discovery.

The researchers believe that their Jetson-based, AI-assisted pipeline could serve as the foundation for a portable, low-cost QPM platform capable of performing automatic, real-time blood screening. Such a system could enable earlier detection of blood disorders like SCD, reduce diagnosis time from hours to minutes, and expand access to advanced hematology diagnostics worldwide.

“QPM has long held potential to provide detailed information about biological cells. But the technique has yet to find widespread clinical use, often due to the cost or complexity in processing the imaging data,” said Professor Adam Wax, leader of the BIOS research group and coauthor of the study. “Here we have shown not only a high-throughput means for profiling thousands of cells at a time but also for rapidly processing and analyzing the information. This may be the missing step needed to bring QPM to the clinic.”

Related Links:
Duke University BIOS Lab