Sequential Patching Lattice for Image Classification and Enquiry: Streamlining Digital Pathology Image Processing

Area of Science:

• Computational pathology
• Digital pathology
• Artificial intelligence in histopathology

Background:

• Digital pathology and AI are transforming histopathology.
• Whole-slide images (WSIs) are large and complex, posing processing challenges.
• Existing methods for WSI processing are often impractical or too expensive.

Purpose of the Study:

• To introduce Sequential Patching Lattice for Image Classification and Enquiry (SPLICE), an unsupervised patching algorithm.
• To condense histopathology WSIs into a compact set of representative patches, minimizing redundancy.
• To improve the efficiency of image retrieval, processing, and analysis in digital pathology archives.

Main Methods:

• Developed SPLICE, an unsupervised algorithm for WSI patching.
• SPLICE sequentially analyzes WSIs to select high-quality, unique, and non-redundant features.
• The algorithm creates a collage of representative patches from the WSI.

Main Results:

• SPLICE demonstrated improved accuracy in search and match applications.
• Achieved reduced computation time and storage requirements compared to existing methods.
• Reduced storage requirements for tissue image representation by 50%.

Conclusions:

• SPLICE offers an efficient method for representing histopathology WSIs.
• The algorithm's efficiency enables broader application of computational pathology tools.
• SPLICE facilitates the accelerated adoption of digital pathology.