click here to download project base paper
click here to download project abstract
– ABSTRACT
Tuberculosis is one of the most ancient diseases and still it is one of the top 10 causes of
death across the world. Most people who get infected with tuberculosis can be saved with
proper treatment and their Life can be saved but due to lack of medical support to detect
tuberculosis in most parts of World still mortality rate due to tuberculosis is high. This
project helps to detect tuberculosis by using image processing techniques over chest x-
rays.
Our objective is to prepare a model which classifies the chest x-ray. This model contains
two classes normal and abnormal (infected with TB) we need to classify between these
two classes and also to achieve high accuracy while classifying.
In our proposed approach we will be improving the accuracy of model by using deep
neural networks to train the model and that model helps us in classifying new chest
x-ray given as input to the model thus meeting our objective.
To achieve good accuracy, we need to pre-process the images first we pre-processed the
images we took images from both datasets Shenzhen and Montgomery and together there
are 800 chest x rays we did augmentation over these images and normalized them
followed by giving these pre-processed images as inputs to our models.
In this project we used two models baseline CNN model and pretrained VGG16 model
and gave pre-processed images as inputs to these both models and evaluated the models
to see which performed better comparing using different performance metrics like
accuracy, specificity, sensitivity, precision and f1-score and depicted them using graphs
and tables From the above results we made a classification reports for both models.
Title: Tuberculosis Detection Using Image Processing
Abstract:
This postgraduate student project aims to develop a robust system for tuberculosis detection through image processing techniques. The existing diagnostic methods for tuberculosis often involve time-consuming and invasive procedures. The proposed system leverages the power of image processing to enhance the efficiency and accuracy of tuberculosis detection.
Existing System:
Current tuberculosis diagnosis methods involve sputum microscopy and culture, which are time-intensive and may have limitations in sensitivity. The need for a faster and more reliable detection method prompts the exploration of image processing techniques.
Proposed System:
The proposed system utilizes image processing algorithms to analyze chest X-ray images for signs of tuberculosis. By automating the detection process, the system aims to provide a quicker and more accurate diagnosis, facilitating timely intervention and treatment.
System Requirements:
- Python programming environment
- Image processing libraries (e.g., OpenCV, scikit-image)
- Machine learning libraries (e.g., TensorFlow, PyTorch)
- Dataset of chest X-ray images with labeled tuberculosis cases
- Adequate computational resources for model training and inference
Algorithms:
The project employs convolutional neural networks (CNNs) for feature extraction and classification. Transfer learning techniques may be employed to leverage pre-trained models such as ResNet or VGG for improved performance with a smaller dataset.
Hardware and Software Requirements:
- Hardware: A machine with sufficient CPU/GPU capabilities for training deep learning models
- Software: Python, relevant libraries, IDE (e.g., Jupyter Notebook)
Architecture:
The system architecture includes:
- Input layer for chest X-ray images
- Convolutional layers for feature extraction
- Fully connected layers for classification
- Output layer for tuberculosis detection
Technologies Used:
- Python for programming
- TensorFlow or PyTorch for deep learning
- OpenCV and scikit-image for image processing
- Flask for web user interface
Web User Interface:
The system will feature a web-based interface using Flask. Users can upload chest X-ray images for analysis, and the system will provide a diagnosis output, indicating the likelihood of tuberculosis presence.
Architecture Diagram
Class Diagram
Activity diagram
Sequence Diagram
