Hadamard-encoded compressive principal component regression for rapid wavelength-scanning surface plasmon resonance imaging.

Surface plasmon resonance imaging (SPRi) enables label-free, high-throughput biomolecular analysis, yet conventional wavelength-scanning configurations rely on costly dispersive optics and broadband sources, suffering from prolonged acquisition, spectral broadening, and poor noise immunity at the resonance dip. We report a Hadamard-encoded compressive principal component regression (HC-PCR) strategy integrated with a compact LED-array-based SPRi system. Six narrowband LEDs are encoded through a Hadamard-style binary encoding matrix, allowing full spectral reconstruction from only six compressive intensity measurements per 162 ms cycle. A regularized least-squares inversion recovers the reflectance spectrum while avoiding the noise-vulnerable resonance dip. The PCR model then maps the reconstructed spectra directly to relative refractive index estimates within a low-dimensional principal-component subspace. Importantly, pixel-to-pixel spectral diversity induced by fabrication non-uniformities is harnessed to enrich the training set rather than suppressed as error. The system was calibrated using NaCl solutions and cross-chip validated across an extended concentration range with independently aligned resonance angles, achieving excellent quantitative agreement ( R 2 > 0.99 ). The HC-PCR mode achieves a detection resolution of 9.73 × 10 - 7 RIU and suppresses noise by approximately one order of magnitude relative to conventional resonance wavelength tracking. Furthermore, using a well-characterized rabbit IgG/goat anti-rabbit IgG binding system, the HC-PCR mode demonstrates superior quantitative linearity ( R 2 = 0.99276 ) compared to the resonance wavelength mode ( R 2 = 0.65554 ), and the extracted binding kinetics agree well with previously reported SPR values, validating the reliability of the system in real biomolecular interaction detection. This work establishes a cost-effective, rapid, and accurate approach for high-throughput SPRi refractive index mapping and biomolecular interaction analysis.