Digital image analysis for biothreat detection via rapid centrifugal microfluidic orthogonal flow immunocapture

We report clear proof-of-principle for centrifugally-driven, multiplexed, paper-based orthogonal flow sandwich-style immunocapture (cOFI) and colorimetric detection of Zaire Ebola virus-like particles. Capture antibodies are immobilized onto nanoporous nitrocellulose membranes that are then laminated into polymeric microfluidic discs to yield ready-to-use analytical devices. Fluid flow is controlled solely by rotational speed, obviating the need for complex pneumatic pumping systems, and providing more precise flow control than with the capillary-driven flow used in traditional lateral flow immunoassays (LFIs). Samples containing the antigen of interest and gold nanoparticle-labeled detection antibodies are pumped centrifugally through the embedded, prefunctionalized membrane where they are subsequently captured to generate a positive, colorimetric signal. When compared to the equivalent LFI counterparts, this cOFI approach generated immunochromatographic colorimetric responses that are objectively darker (saturation), more intense (grayscale), and less variable regarding total area of the color response. We also describe an image analysis approach that enables access to rich color data and area statistics without the need for a commercial 'strip reader' or custom-written image analysis algorithms. Instead, our analytical method exploits inexpensive equipment (e.g., smart phone, flatbed scanner, etc.) and freely available software (Fiji distribution of ImageJ) to permit characterization of immunochromatographic responses that includes multiple color metrics, offering insights beyond typical grayscale analysis. The findings reported here stand as clear proof-of-principle for the feasibility of disc-based, centrifugally driven orthogonal flow through a membrane with immunocapture (cOFI) and colorimetric readout of a sandwich-type immunoassay in less than 15 minutes. Once fully developed, this cOFI platform could render a faster, more accurate diagnosis, while processing multiple samples simul-taneously.