Skip to main content Skip to main menu
UF Health University of Florida
Computational Microscopy Imaging Laboratory Section of Quantitative Health in the Department of Medicine
join us

FUSION NAVIGATOR

Functional Unit State Identification for WSI INTEGRATING MOLECULAR AND STRUCTURAL OMICS FOR BIOLOGICAL DISCOVERY AND PRECISION MEDICINE

Fusion Logo

AI-driven visualization and quantification of spatial OMICS

EXPLORE THE KIDNEY
fusi-gif-no-background

 

schematic diagram showing a man, two lines coming out of kidney region to shows zoomed in kidney WSI where glomerulus, tubules and other compartments are segmented. There is fusion logo on the top that reads "FUSION, wsi navigator". There are other omages of DNA, cells, cells, showing the platform works with multimodal data.

WHAT IS FUSION

FUSION is an interactive interface to view spatial transcriptomics data integrated with histopathology.  This is driven by artificial intelligence (AI).  AI segments Functional Tissue Units (FTU) and links these with gene expression data that informs on major and minor healthy and injured cell types directly mapped on histological features on the FTUs in the WSI.

HOW FUSION WORKS

State of the art machine learning algorithms in a user-friendly data input interface

A detectron2 pipeline diagram showing a crop of whole slide image, convolutional blocks in series reducing in size to show reduction in dimension of feature maps, reaching a minimum point and then second series of blocks increasing in size leading to another segmented WSI. This shows a U-Net style architecture.
there are two images, one segmented WSI with selected ROI. A blue arrow points from there to another image that contains a pie chart and a bar diagram with cell state proportions.

Highly interactive frontend design to drive hypothesis development and testing

Cloud-hosted data storage made possible through Digital Slide Archive (DSA)

Data flows from scattered elements into a cloud, then onward to other elements. A green arrow indicates the first flow.
Illustration of two people interacting with a laptop displaying data visualization. One person is seated, the other is standing and pointing.

We want people to test FUSION 

Bring your own data!

Woman in lab coat carrying data cylinders toward a funnel labeled 'Fusion' and 'DSA'.

Visit FUSION 1.0 GitHub Repo

Visit FUSION 2.0 GitHub Repo

FUSION ++

To streamline user experience and broaden accessibility, FUSION now includes a fully integrated JupyterHub environment as a core component of its open-source offering. This pre-configured deployment removes the technical barriers to entry, enabling diverse range of users—from students and researchers to data scientists—to instantly create and run notebooks with full Fusion++ functionality, eliminating the need for complex manual installation or environment configuration.

By embedding JupyterHub directly into the FUSION++ ecosystem, we provide more than just a coding interface; we offer a modular, high-performance platform optimized for the analysis of spatial omics and histology data. This integration leverages robust compute resources and a secure enclave, ensuring that user data remains private while benefiting from a standardized Python environment. Within this workspace, researchers can execute a complete, reproducible workflow: fetching large-scale datasets (such as those from the HuBMAP consortium), running complex spatial analyses, and performing interactive visualizations using the integrated HitomicUI.

Beyond these built-in tools, the platform provides immense flexibility for advanced users to develop custom workflows and generate bespoke visualizations programmatically. While HitomicUI offers powerful interactive capabilities, the notebook environment allows researchers to extend FUSION++’s functionality to create specialized plots using Fusion++. As we move toward FUSION 2.0, this ecosystem will continue to evolve, incorporating agentic AI for molecular discovery and expanding support for single-cell spatial omics, further cementing its role as a cross-disciplinary tool for the global research community.


FUSION interface in FUSION++

screenshot of jupyter notebook

Custom Plots in FUSION++ (N/A in FUSION)

a volcano plot and hierarchical plot

FUSION++ Resources

  • FUSION++ GitHub repository
  • Documentation: a beginner-friendly manual for running the FUSION++ pipeline

KEY Developers

PRINCIPAL INVESTIGATOR

Pinaki Sarder

The PI

prof

Bioinformatician III

Heather R Kates

Expertise: Biomedical Research, Bioinformatics, Spatial Transcriptomics

Headshot of a woman with brown hair and light brown eyes, smiling gently.

Kitware inc

David Manthey

Software and Production

Headshot of a man with short brown hair, glasses, and a light blue checkered shirt, smiling at the camera.

MS student, ELECTRICAL AND COMPUTER ENGINEERING

Srisharanya Injarapu

Expertise: Full-Stack Software Development

Close-up of a woman with long, dark wavy hair smiling at the camera, wearing a graduation stole.

ASSISTANT SCIENTIST

Anindya Sankar Paul

Expertise: Deep Learning, Federated Learning, Secure and Privacy Preserving AI, Signal Processing, Medical Imaging

Ap
Samuel Border PH.D. STUDENT, 2020-2025
Nicholas Lucarelli PH.D. STUDENT, BIOMEDICAL ENGINEERING
Jessica Kirwan Project Manager
Suraj Ramadugu Image Analysis User Support
Suhas Katari Chaluva Kumar System Analyst
Ujwala Guttikonda Data Management Analyst
Sumanth Devarasetty Web Developer
Praveen Kumar Dande System Analyst
Nikhil Yerra Data Management Analyst
Haitham Magdy Data Management Analyst
Annika Holmstrom Usability Coordinator
Ishaq Ansari Data Management Analyst OPS

contact details

Contact Pinaki Sarder <pinaki.sarder@ufl.edu> with queries, request for testing, and suggestions for enhancing FUSION. 

©Copyright 2023 University of Florida Research Foundation, Inc. All Rights Reserved.