iSPEED: a Scalable and Distributed In-Memory Based Spatial Query System for Large and Structurally Complex 3D Data.

Area of Science:

• Digital pathology
• Computational biology
• Bioinformatics

Background:

• Digital pathology advancements enable high-resolution 3D tissue analysis for disease investigation.
• Analyzing spatial patterns in large 3D biological data is crucial but computationally challenging due to data volume and object complexity.

Purpose of the Study:

• To present iSPEED, a scalable, in-memory spatial query system designed for large-scale, structurally complex 3D pathology data.
• To address the high I/O, communication, and computational costs associated with 3D spatial queries.

Main Methods:

• Implemented an in-memory architecture with progressive compression and levels of detail for low latency.
• Utilized pre-generated global spatial indexes and on-demand run-time indexing for low computational cost.
• Applied structural indexing for complex objects across multiple query types.

Main Results:

• Demonstrated efficient 3D spatial joins, nearest neighbor, and proximity estimation queries on multiple datasets.
• Achieved minimal memory footprint during query processing through efficient indexing and on-demand decompression.
• Provided a web-based RESTful interface for data exploration and query parameter adjustment.

Conclusions:

• iSPEED offers a scalable and efficient solution for spatial querying of large, complex 3D pathology data.
• The system facilitates faster discovery and verification of spatial patterns critical for disease understanding.
• The in-memory approach with advanced indexing significantly reduces query latency and computational overhead.