┐(￣▽￣)ゞ Where Was I

We hope this project will be useful in providing a rigorous yet frictionless experience, with a tinge of fun, for biologists when they are trying to extract insights from raw microscopy data from FLIM (fluorescence lifetime imaging microscopy) and other modalities (e.g. brightfield, QPI).

The idea and vision behind it are illustrated on the journal cover and in the cover legend (Cell Reports Methods Volume 6, Issue 8).

A hopscotch court chalked on pavement maps FLIM Playground end-to-end. From square 1, the player hops through each step the software unifies: identifying fields of view; calibrating unprocessed photon decays; extracting single-cell fit, phasor, morphology, and texture features; merging datasets into a unified table; tagging each cell; and analyzing them through interactive widgets—iterative, frictionless, exploratory, and intuitive. A once-fragmented, code-heavy trek becomes one court, hopped back and forth as each biological question suggests the next. Chalk, inexpensive and open to all, echoes Zhao et al.’s paper’s advance: an open-source, code-free platform everyone can play on.

Conceptually, FLIM Playground is divided into two integrated yet independent sections:

• Data Extraction that extracts multichannel, multimodal, single-region-of-interest (e.g. single-cell) features, and
• Data Analysis that provides interactive visualization and statistical modeling on those features, or on any tabular dataset.

Key architectural innovations:

• A three-class abstraction. The multi-level complexity built into microscopy data — pixel, region of interest, channel, field of view, and experiment — collapses into three semantic classes: a single-object identifier, numerical features, and categorical features.
• One vocabulary across both sections. Those three classes are what let data extraction and data analysis fit together, because analysis of any tabular dataset can be decomposed into a single task: closing the conceptual gaps between categories with numerical measurements, at the single-object level.
• Channel-level customization. Numerical features are extracted from each channel independently: each channel is associated with an imaging modality, and each modality comes with its own built-in feature extractors.

To learn more, see:

• Open-source code and cross-platform releases
• Documentation
• Paper