Zebrafish Image Classifier (SVM + CNN)

Description

This project is a simple image classification pipeline for thrombosis research, built using scikit-learn and PyTorch. It uses a Support Vector Machine (SVM) model and Convolutional Neural Network (CNN) model.

The SVM model resizes all images to 15×15 pixels and flattens them into feature vectors. The dataset is then split into training and testing sets, after which an SVM classifier is trained. Model performance is evaluated using accuracy, sensitivity, specificity, balanced accuracy, and ROC AUC.

The CNN model resizes all images to 224x224 pixel image and applies a green fluorescence filter which then trains a CNN after being split into training and testing datasets. This model is then tested on accuracy, sensitivity, specificity, balanced accuracy, and ROC AUC.

Purpose: The approach is intended to examine clotting pattern differences between different zebrafish thrombosis models at 5 days-post-fertilization (dpf). It distinguishes between an acquired (estrogen-induced) model exhibiting speckled patterns of thrombus distribution, versus a genetic (spontaneous; protein C deficient) model exhibiting sprouting patterns with denser fluorescence signals.

How to setup and run

1. Clone the repository

Clone the repo into your local directory, then navigate into it.

git clone https://github.com/ymnmurat/Zebrafish-Imaging-Classification.git
cd Zebrafish-Imaging-Classification

2. (Optional) Create a virtual environment

This is recommended to use so that dependencies in svm.py do not conflict with your local system packages.

python -m venv venv

Activate on Windows:

venv\Scripts\activate

Activate on macOS/Linux:

source venv/bin/activate

3. Install dependencies

Run pip install -r requirements.txt to install dependencies.

4. Prepare dataset

Place your images inside the images/ folder, each corresponding to its respective subfolder. For example:

images/
├── class_0/
│ ├── img1.png
│ ├── img2.png
├── class_1/
│ ├── img3.png
│ ├── img4.png

Note: The images included in this repository are a small subset of data. The full dataset is available upon request.

The labeling of the subfolders in /images corresponds to the 2 groups of experimental images used for the SVM and CNN models, and are included as subfolders in the full dataset:
class_0: '20220626_ProC-MePS-12'
class_1: '20220704_FGBA3-MCE-MEPS-VALID-A-2-6'

5. Run model

To run the pipeline, use the following command:

SVM Model

python svm.py

CNN Model

python cnn.py