Analysis of Chest X-Ray images

build a neural network model using to identify X-Ray images where an "effusion" is present

Problem statement

Neural networks have revolutionised image processing in several different domains. Among these is the field of medical imaging. In the following notebook, we will get some hands-on experience in working with Chest X-Ray (CXR) images.

The objective of this exercise is to identify images where an "effusion" is present. This is a classification problem, where we will be dealing with two classes - 'effusion' and 'nofinding'. Here, the latter represents a "normal" X-ray image.

This same methodology can be used to spot various other illnesses that can be detected via a chest x-ray. For the scope of this demonstration, we will specifically deal with "effusion".

Table of Contents

• General Info
• Technologies Used
• Conclusions
• Acknowledgements

General Information

• To build a CNN based model which can accurately detect abnormalities in the X-ray scans.

• Neural networks have revolutionised image processing in several different domains. Among these is the field of medical imaging. In the following notebook, we will get some hands-on experience in working with Chest X-Ray (CXR) images.

• The objective of this exercise is to identify images where an "effusion" is present. This is a classification problem, where we will be dealing with two classes - 'effusion' and 'nofinding'. Here, the latter represents a "normal" X-ray image.

• The dataset consists of X-ray scans, which were formed from the CXR datasets.

Steps Followed in the Projects

• Data Preparation: Made sure all our images were of the same resolution
• Data Pre-Processing: Morphological Operations
• Data Pre-Processing: Normalisation
• Data Pre-Processing: Augmentation
• Model Building: Choosing AUC as evaluation metrics & introducing weighted cross entropy
• Running ablation experiments
• Overfitting on a smaller version of the training set
• Hyperparameter tuning
• Mode training and evaluation

Conclusions

• Accuracy has been improved significantly after treating the class imbalance using augmentation.
• ROC has been improved by using weighted cross entropy for loss validation.
• Accuracy 80% and AUC 84%, with very less loss in train and validation.

Technologies Used

• numpy
• pandas
• pathlib
• keras
• tensorflow
• matplotlib

Acknowledgements

Give credit here.

• I would like to thank my upGrad Buddy and tutor for assisting me to complete this assignment
• Reference: Stack Overflow, Wikipedia & Google Collab

Contact

Created by [@subham0206] - feel free to contact me!