Chest Xray Pneumonia Detection Neural Network Model

This project aims to detect pneumonia from chest X-ray images using a Convolutional Neural Network (CNN). The model is trained on a dataset of chest X-ray images and evaluated for its performance. The project is ongoing, and I aim to fine-tune the model in the future.

If you are seeing this, it means I am still working on the project.

Prediction Accuracy: Currently at 84%

Data

The dataset consisted of chest X-ray images divided into two main categories: NORMAL and PNEUMONIA. The images were sourced from a publicly available medical imaging dataset.

You can get the data here: https://doi.org/10.17632/rscbjbr9sj.2

Methodology

1. Data Preprocessing: Images were resized and normalized to a uniform scale. Data augmentation techniques like rotation, width shift, height shift, and horizontal flip were applied to increase the robustness of the model.

2. Model Development: A CNN was designed with multiple convolutional, max pooling, and dense layers. Dropout layers were also incorporated to prevent overfitting.

3. Training: The model was trained on the training dataset, with a portion set aside for validation to monitor the training process for signs of overfitting.

4. Evaluation: After training, the model was evaluated on a separate test set to measure its accuracy and ability to generalize to new data.

Results

• The CNN achieved an accuracy of approximately 84% on the validation dataset.

• The model tended to predict most test images as having pneumonia, indicating potential issues with class imbalance or model sensitivity.

Discussion

• The results suggest that while the model can identify pneumonia in many cases, its tendency to over-predict pneumonia points to the need for further tuning. Adjustments to the model’s architecture, training process, or data preprocessing steps could improve its specificity and overall accuracy. I intend to work on these improvements in the future, so you might see a higher accuracy rate in the fine-tuned model.

Purpose

• I have built this model for my own study purposes, which is why you may encounter undefined and unexplained steps and codes. I am actively working to clean up the first draft of my model, so please stay tuned for more updates.