WHIDE--a web tool for visual data mining colocation patterns in multivariate bioimages.

Area of Science:

• Bioimaging
• Computational Biology
• Data Visualization

Background:

• Bioimaging techniques are advancing, generating high-dimensional multivariate bioimages (MBIs) that require sophisticated analysis for spatial and molecular data.
• Analyzing MBIs involves understanding both sample morphology and molecular colocation or interaction patterns.

Purpose of the Study:

• To introduce WHIDE (Web-based Hyperbolic Image Data Explorer), a free web application for analyzing multivariate bioimages.
• To combine computational learning, dimension reduction, and visualization principles to address MBI analysis challenges.

Main Methods:

• WHIDE utilizes computational learning and dimension reduction techniques.
• It provides a multimodal visualization combining topology-preserving pseudocoloring with information visualization.

Main Results:

• WHIDE was applied to multivariate bioimages from a colon cancer study using the Toponome Imaging System.
• The tool successfully reduced data complexity by clustering pixels into Molecular Co-Expression Phenotypes, aiding in the comparison of normal and tumor tissues.
• Demonstrated applicability across toponome imaging, high-content screens, and MALDI imaging.

Conclusions:

• WHIDE offers an efficient approach to analyze complex bioimage data.
• The tool facilitates sophisticated visualization and aids in understanding tissue characteristics and molecular interactions.