Filament Fillies

A web-based microscopy analysis tool available at www.filament-tracker.co.uk for detecting and tracking filamentous structures in time-lapse fluorescence microscopy images. The system provides a pipeline for segmentation, classification, and visualization of filaments, with a modern React frontend and Python backend.

Use of LLMs:

Claude was extensively used to generate code for website UI. ChatGPT assisted with CellPose coding.

Features

Upload multi-channel TIF stacks via a browser UI Automated channel splitting, segmentation (Cellpose), and Frangi-based filament detection Filament classification (length, lifecycle, etc.) Interactive results visualization: playback, per-track inspection, and metrics dashboards Hyperparameter tuning and metrics dashboard Synthetic data generation and standard dataset support

System Architecture

Frontend: React (located in frontend) Backend: Python (Flask API, main code in biohack) Data: Input/output in data, including synthetic and live datasets Deployment: Nginx serves the frontend and proxies API/data requests to Flask

Quickstart

git clone https://github.com/briandaniel14/biohack
cd biohack

python3 -m venv .venv
source .venv/bin/activate
pip install -r requirements.txt

To start the frontend (from frontend/):

cd frontend
npm install
npm run dev

Usage

Upload TIF stacks via the web UI Run the detection pipeline and inspect results Tune hyperparameters and view metrics

Built With

• flask
• python
• react
• scipy