Surface plasmon resonance imaging (SPRi) is a powerful label-free technique for biomolecular interaction analysis, yet conventional intensity-based SPRi systems suffer from limited refractive index resolution and rely on costly light sources. Here, we present a high-throughput, low-cost SPRi biosensor based on a differential intensity detection algorithm with standard LED illumination. By probing the opposite reflectance variations on the two flanks of the resonance dip at 730 and 780 nm and taking their difference, the scheme cancels common-mode noise while enhancing the sensitivity by up to 2.6-fold. Specifically, the differential signal is computed pixel-wise as the difference between the two reflectance images acquired under the 730 and 780 nm LEDs, I_D = I₇₃₀ − I₇₈₀, so that the opposite-signed responses on the two flanks are added while common-mode intensity fluctuations are subtracted. The system employs only three standard LEDs, a six-channel flow-through cell, and a single CMOS camera, with a light-source cost below 200 USD. A sensitivity of 1.03 × 10⁴ grayscale values per refractive index unit (RIU) with excellent linearity (R² = 0.9989) and a resolution of 1.9 × 10⁻⁶ RIU were achieved. As a proof of concept, the biosensor was applied to label-free monitoring of autophagy-related LC3B accumulation in A549 lung cancer cell lysates using the small-molecule inhibitor DC-LC3in-D5 as the probe. The SPRi responses showed a dose-dependent trend fully consistent with flow cytometry and Western blotting. Benefiting from its high-throughput capacity, low cost, and antibody-free readout, this platform holds potential for evaluating autophagy-mediated therapeutic resistance upon anticancer drug treatment.