ChestXNet: Pneumonia Detection from Chest X-rays with Convolutional Neural Networks

Project info

This project is part of AI in healthcare course on Udacity Platform.
https://www.udacity.com/course/ai-for-healthcare-nanodegree--nd320

The goal of the project is to classify the X-ray scans as pneumonia positive or negative. Different Convolutional Neural Network architectures based on DenseNet121 pre-trained model are built, trained, and evaluated.

Dataset info

The dataset used in this project:

NIH Chest X-ray Dataset comprised of 112,120 X-ray images with disease labels from 30,805 unique patients. The labels include 14 pathologies and were extracted using Natural Language Processing (NLP) from radiological reports.

Detailed info: https://www.kaggle.com/nih-chest-xrays/data

Project files

1. EDA (exploratory data analysis)
   
   
2. Build and train model:
     - Processing metadata
     - Creating training, validation and test datasets
     - Comparison of demographic distributions in the training, validation and test datasets
     - Building of different models, their training and evaluation
     - Model performance summary
     - Next steps
   
   
3. Clinical workflow intergration:
     - checking relevant DICOM matadata
     - pre-processing image for the model
     - loading trained model and its weights
     - predicting the class
   
   
4. FDA submission:
     - intended use statement
     - indication for use
     - device limitations
     - clinical impact of performance
     - algorithm architecture, training and validation description
     - database used for algorithm development
     - ground truth description