nexusx12 / dental-x-ray-analysis

Automating dental X-ray analysis using computer vision. Scripts classify dental conditions.

Dental X-ray Analysis

• Dependencies
• Dataset structure
• Abstract
• Introduction
• Code Basis
• Installation
• Usage
• Future Work
• Authors
• References
• Conclusion
• License

Dependencies

Download data

Dataset structure

The dataset should be arranged in a structured fashion, as the following:

unlabeled_images/
	- 0001_jpg.rf.30a42966fb9c51553f6949b70234218d.jpg
	- 0001_jpg.rf.57229a11e925669019e179341e22c97a.jpg
	- 0001_jpg.rf.f94abcb7858bb419a7202ef60ef95bd6.jpg
	- ......
	- 1000_jpg.rf.c9dd4b2cf2f9c945b8d1921990531852.jpg
_annotations.csv
train.py
tooth_segmentation.py

Abstract

This project aims to develop a tool for dental X-ray analysis using computer vision techniques. The primary objective is to segment and identify teeth in panoramic X-ray images, providing a foundation for future applications in dental diagnosis and treatment planning. The project is currently under development, and this document outlines the basis of the code, its underlying ideas, and how to use the tool.

Introduction

Dental X-rays are essential for diagnosing and treating various dental conditions. Automating the analysis of X-ray images can significantly aid dental professionals in their decision-making processes. This project leverages computer vision, specifically image segmentation, to identify and mark teeth in panoramic X-ray images.

Code Basis

The project consists of two main scripts:

Script 1: Dental X-ray Classification

• Objective: Train a machine learning model to classify dental conditions based on annotated X-ray images.
• Implementation: Utilizes a Random Forest classifier to predict dental conditions (e.g., Fillings, Impacted Tooth, Implant).
• Usage:
  • Modify the threshold value in create_teeth_mask function for optimal segmentation.
  • Train the model using annotated data in the _annotations.csv file.
  • Save the trained model as a pickle file using train.py.

Script 2: Tooth Segmentation and Boundary Drawing

• Objective: Segment individual teeth and draw boundaries around them in a given X-ray image.
• Implementation: Applies adaptive thresholding and contour detection to identify teeth regions and draw rectangles around them.
• Usage:
  • Adjust the threshold_value parameter in the draw_tooth_boundaries function for optimal segmentation.
  • Provide an input X-ray image to visualize tooth boundaries.

Installation

To run the scripts, follow these installation steps:

1. Clone the repository:

   git clone https://github.com/NexusX12/Dental-X-ray-Analysis

2. cd Dental-X-ray-Analysis:

   cd Dental-X-ray-Analysis

3. Install dependencies:

   pip install -r requirements.txt

Usage

Script 1: Dental X-ray Classification

 python train.py

This script will generate trained_model.pkl in current directory.

Script 2: Tooth Segmentation and Boundary Drawing

Make sure the trained_model.pkl is in the same directory.

 python tooth_segmentation.py

Future Work

This project is currently under development, and future work will focus on:

• Refining segmentation techniques for improved accuracy.
• Integrating deep learning approaches for more sophisticated tooth detection.
• Developing a user-friendly interface for easy adoption by dental professionals.

Authors

• Tharun Birla

References

1. scikit-learn: Machine Learning in Python. https://scikit-learn.org/
2. OpenCV: Open Source Computer Vision Library. https://opencv.org/
3. NumPy: The fundamental package for scientific computing with Python. https://numpy.org/
4. Pandas: Powerful data structures for data analysis. https://pandas.pydata.org/

Conclusion

Dental X-ray analysis through computer vision holds great promise for enhancing dental diagnosis and treatment planning. This project provides a foundation for further research and development in this domain.

Note: This project is under active development, and contributions are welcome. Please check the repository for updates and improvements.

License

• The MIT License (MIT) Copyright (c) 2018 École Polytechnique, Université de Montréal
• Dataset - CC BY-SA 4.0