Cell Imaging Analysis Tool

A PyQt5 desktop application for detecting, tracking, and analyzing cells in microscopy images and video, built on Cellpose for segmentation.

Features

• Load a single image or a video (.png, .jpg, .jpeg, .tif, .tiff, .avi, .mp4, .mov)
• Detect cells per-frame or across an entire video using Cellpose
• Track detected cells across frames (simple nearest-centroid matching)
• Measure per-cell intensity statistics (mean/max/min/std) across one or more fluorescence channels
• Flag cell-cell interactions based on centroid distance
• Plot intensity over time for a selected track
• Export detection/track/intensity results to Excel or CSV

Requirements

• Python 3.9+
• A CUDA-capable GPU is optional — Cellpose will fall back to CPU automatically if none is available

Setup

pip install -r requirements.txt
python ui_main.py

Project layout

File	Responsibility
ui_main.py	PyQt5 main window, wiring, and app entry point
video_handler.py	Loading images/video, frame caching, playback
cell_detection.py	Cellpose model management and async detection
cell_tracking.py	Frame-to-frame cell tracking
intensity_analysis.py	Per-cell fluorescence intensity statistics
interaction_detection.py	Distance-based cell-cell interaction detection
visualization_utils.py	Drawing detections/tracks/plots onto frames
data_export.py	Exporting results to Excel/CSV
config.py	Central tunable parameters

Notes

• Detection parameters (diameter, flow threshold, cellprob threshold) and the interaction distance threshold can be adjusted live from the UI; defaults come from config.py.
• detect_all_frames dispatches detection for every loaded frame to a background thread pool and reports completion once every frame has actually finished processing.
• Frame caching is capped by Config.MAX_CACHE_FRAMES / Config.MAX_MEMORY_USAGE_MB to avoid loading very large videos entirely into memory.