Fast Spot Locating for Low-Density DNA Microarray.

Low-density DNA microarrays are crucial in molecular diagnostics due to their cost-effectiveness and high sensitivity. However, reliable spot localization remains challenging due to positional variations and image artifacts. Traditional intensity-based methods often struggle with weak fluorescence signals. To address this, we propose a rapid spot localization method that combines template matching with point pattern matching, enhanced through vectorized programming and square (box) templates. Vectorized programming accelerated the most time-consuming calculation by 82 times on a PC and was 6000 times faster on a Raspberry Pi compared to a for-loop implementation. While this improvement applies to the vectorized square calculation alone, substantial performance gains were still achieved in the overall process. Additionally, replacing circular templates with square templates resulted in a fourfold reduction in processing time without compromising detection performance. The proposed method effectively reduces computational overhead, making it suitable for high-throughput and resource-constrained applications. The method was validated using HPV genotyping images from commercial DNA microarrays, demonstrating its practical applicability and robust performance in clinical settings.