RodSizer

RodSizer is a local, scientific desktop tool for measuring gold nanorods (and similar nanoparticles) in electron-microscopy images. You drop in TEM / STEM / SEM images and it detects every particle, separates touching ones, measures each, and produces size distributions and exportable reports — all running on your own machine, with no data leaving your computer.

What it does

• Detects & segments particles with K-means thresholding, then separates touching / overlapping rods using a marker-controlled watershed on the Euclidean distance transform — so dense clumps are split into individual rods instead of being counted as one large blob.
• Measures each particle from a rotated bounding box: length, width, aspect ratio, area, volume (hemispherically-capped cylinder), orientation, plus shape descriptors (solidity, circularity, eccentricity, convexity).
• Calibrates automatically: reads pixel size from embedded metadata (Gatan DM3/DM4, Velox/EMD, OME- / ImageJ / FEI TIFF). For camera exports that only carry a burned-in scale bar or a "Pixel size / Fov" footer, it reads the value with on-image OCR. Manual draw-a-line calibration is always available.
• Lets you curate results on an interactive analysis page: click a particle to keep/exclude it, sort the table by area (or L / W / AR / ID), bulk select/deselect everything above or below a row, auto-remove statistical outliers (Tukey 1.5×IQR), and toggle the on-image ID labels.
• Exports per-image and per-folder statistics, histograms, and CSV / Excel / PDF reports.

Typical workflow

1. Create a folder and upload images (you can drop in many at once).
2. Open each image to review detection. Fix the scale if needed, then deselect clumps / misreads — click them on the image, or use the sort, bulk-select, and "Deselect outliers" tools in the selection panel.
3. Press Update Particles to generate statistics and histograms, then download the report.
4. Use View Analysis to aggregate the whole folder.

Methodology

Segmentation follows the AutoDetect-mNP approach (JACS Au 2021, 1, 316−327), extended in RodSizer with distance-transform watershed splitting for dense clumps and OCR scale-bar calibration for un-tagged camera exports.

Contributors

• Shi Chen — Murphy Group, UIUC
• Arda Turk — Murphy Group, UIUC
• Built with assistance from Claude and ChatGPT.

Group website: https://murphy-group.chemistry.illinois.edu/

Instruction for Launching

1. Click on "<>code" in Github page.
2. Click on "Download ZIP".
3. Open Finder (MacOS) or Files (Windows).
4. Click on the RodSizer.zip to unzip.
5. Double-click on RodSizer_Launcher_MacOS.command or RodSizer_Launcher_Windows.bat.
6. Wait for the environment to be set up (first time only) and the local website to be opened.

Notes:

• DO NOT double-click on RodSizer_CLEANER_MacOS.command unless you are sure that you want to clean ALL local history of data and reports.
• First-time launching may take some time, like 5-10 mins. Most of that is installing the deep-learning packages (tensorflow + stardist, ~1 GB). These are kept for possible future ML-based detection but are not used by the current pipeline (K-means + watershed), so the wait is one-time setup only — later launches are fast. See backend/requirements.txt to remove them if you want a leaner install.
• Try ask a coding agent if there's an issue with environment setup.
• MacOS is more recommended.
• (Windows) If Windows Defender asks, click More Info -> Run Anyway.
• (Windows) Users should ensure Add Python to PATH is checked during installation.
• Automatic scale-bar OCR uses Tesseract, which the launcher installs for you (Homebrew on macOS, winget on Windows). If it cannot be installed, RodSizer still runs — just calibrate manually.

Cleaner Script (Mac)

• RodSizer_CLEANER_MacOS.command is a cleanup tool for clearing local history data.
• It permanently removes:
  • uploaded files in uploads/
  • generated results in results/
  • folder analysis cache in .analysis_cache
  • backend/server.log
  • Python cache files (__pycache__ and *.pyc)
• It does not remove your Python virtual environment or installed packages in backend/.venv.
• Use it only when you want to reset the app's local working history and cached outputs.

Parameter Guide

• PARAMETER_GUIDE.md is a reference document for the current detection and measurement parameters used by RodSizer.
• It is mainly intended for developers or advanced users who want to understand or adjust processing behavior.
• The actual implementation is in backend/processing.py.

Requirements

• macOS or Windows
• Python 3 installed (standard on most Macs, or downloadable from python.org)