Fluorescence Image Studio

How to use it

Split channels exports every original channel as a lossless TIFF in its original data type (16-bit stays 16-bit), plus optional coloured copies on the same intensity scale — nothing else is changed. Merge only combines separate single-colour images (for example blue DAPI, green FITC and red PE of the same field) into one merged colour image, with a colour and brightness for each — no other processing. Enhance & export restores a selected field for figures: choose a starting preset, adjust noise reduction, background subtraction and Richardson–Lucy deconvolution, compare Before/After at native pixels, set each channel’s colour and display range, crop with the mouse or coordinates, then process the full field and download figures, crops and the scientific data with a settings manifest. Research measurements detects nuclei or cell regions, lets you review and edit the outlines, and measures raw (never enhanced) intensities with background correction and quality checks. Batch processing applies one reviewed recipe to up to 250 files: figures only (rendered RGB TIFFs accepted), measurements, or lossless splitting.

What it computes

The calculations are the Python code of the Image Studio 1.5.0 desktop app running unchanged in your browser through Pyodide (Python compiled to WebAssembly) with NumPy, SciPy, scikit-image and tifffile: non-local means or total-variation denoising, white top-hat, rolling-ball or Gaussian background estimates with an explicit subtraction strength, reflection-padded Richardson–Lucy deconvolution with an estimated Gaussian or measured PSF (peaks are never clipped; restored data are saved as float32), and the display curve clip((value − min)/(max − min)) → contrast/brightness → v^(1/gamma). Raw, restored and display arrays are kept separate; display, crop and enlargement settings never change the scientific data.

Limits

Restoration improves visibility; it cannot recover detail absent from the acquisition, fix severe defocus or replace a measured 3D optical model, and noise reduction or background subtraction can remove real dim or broad signal — always compare with the original channels. Splitting a rendered RGB image gives red/green/blue components, not recovered fluorophores. A DAPI mask is a nucleus, and overlap of two colours does not prove uptake or fusion. Browser limits: up to 400 MB per file, 512 MB decoded per series, and 300 MB per download ZIP; very large stacks need the desktop app. LZW, Deflate, PackBits, Zstd and uncompressed TIFFs are supported; JPEG-compressed TIFF, JPEG 2000 and native .xpt/.imm containers are not.

Keep your work

Nothing is stored or uploaded. The first visit downloads the analysis engine (about 40 MB, then cached by your browser). Download your ZIPs before closing the page; each contains a manifest with source checksums, selections and every setting. Clear session removes loaded files and results from memory.

Privacy and local-data controls · Credits and licenses · Interpretation limits