Time-series Medical Image Classification

Image Classification Model for Temporal Disease Progression of Chest X-ray dataset

Introduction

Disease progression modeling (DPM) uses mathematical functions and scientific principles to describe the quantitative progression of a disease over time, providing valuable insights for the development and use of medicines. Disease Progression models have the potential to improve patient outcomes, reduce healthcare costs, and accelerate the development of new treatments. One such task is to predict the three states of disease progression (improving, stable, or worsening) given the current and past multi-image frontal chest X-ray images. This work focuses on fine-tuning and evaluating the pre-trained Torch X-ray Vision model Cohen et al. (2021) for the temporal image classification task.

Setup

1. Download the Chest X-ray dataset here: https://drive.google.com/file/d/1rKGrW57Nr6AN-jOQOPOQzQiHk8LD3q2o/view?usp=sharing
2. Download the MS-CXR-T dataset and move MS_CXR_T_temporal_image_classification_v1.0.0.csv to the root directory
3. Run python feature_extraction.py {feature_extractor} to extract features from any of the four extractors used
4. Run python logistic_regression_baseline.py {feature_extractor} to run the baseline logistic regression model
5. Run python model2.py {feature_extractor} to run the second model, five separate independent classifiers
6. Run python model3.py {feature_extractor} to run the third model, a combined model with loss masking