About
This skill provides critical domain-specific guidance for researchers and developers working with multiplex immunofluorescence data and the BaSiC illumination correction algorithm. Based on empirical failure analysis, it documents why caching illumination profiles across cycles leads to significant intensity errors (up to 20%)—particularly in sparse markers—and provides the necessary context to avoid these pitfalls. Instead of risky caching, the skill offers optimized implementation patterns for CuPy and GPU acceleration, including n-dimensional DCT operations and batch processing, to ensure image processing pipelines remain both fast and scientifically accurate.
