SAM-dPCR: Accurate and Generalist Nuclei Acid Quantification Leveraging the Zero-Shot Segment Anything Model.

Area of Science:

• Biotechnology
• Molecular Diagnostics
• Bioanalysis

Background:

• Digital PCR (dPCR) enables absolute quantification of nucleic acids but faces challenges with cost, complexity, and speed.
• Existing dPCR detection methods often require specialized equipment and extensive training, limiting accessibility.

Purpose of the Study:

• To introduce SAM-dPCR, a novel, training-free, open-source bioanalysis paradigm for rapid and precise absolute quantification.
• To demonstrate the accuracy and broad applicability of SAM-dPCR across various sample types and dPCR platforms.

Main Methods:

• Leveraged the Segment Anything Model (SAM) for automated image analysis in dPCR.
• Developed a training-free approach for dPCR data processing, reducing reliance on "ground truth" data.
• Validated SAM-dPCR using standard fluorescence microscopes with both droplet and microwell dPCR formats.

Main Results:

• Achieved processing times under 4 seconds with accuracy exceeding 97.10%.
• Demonstrated accurate quantification across a wide dynamic range (0.154 copies µL⁻¹ to 3.629 × 10³ copies µL⁻¹).
• Observed a strong linear correlation (r² > 0.96) between expected and measured nucleic acid concentrations.

Conclusions:

• SAM-dPCR offers a highly accurate, accessible, and rapid solution for absolute nucleic acid quantification.
• The open-source and training-free nature of SAM-dPCR enhances its utility, particularly in resource-limited settings.
• This method significantly improves upon traditional dPCR detection limitations, broadening its diagnostic potential.