Dimensionality Reduction in Deep Learning for Chest X-Ray Analysis of Lung Cancer

TL;DR

This study evaluates various dimensionality reduction techniques, including lung segmentation, bone shadow exclusion, and t-SNE, to improve deep learning analysis of chest X-rays for lung cancer detection, showing that combined preprocessing enhances model accuracy.

Contribution

The paper introduces a comprehensive evaluation of multiple preprocessing techniques, demonstrating that their combination significantly improves deep learning performance on chest X-ray analysis.

Findings

Preprocessed dataset with all techniques achieved highest accuracy.

Lung segmentation and bone shadow exclusion improved model training.

t-SNE filtering outliers further enhanced performance.

Abstract

Efficiency of some dimensionality reduction techniques, like lung segmentation, bone shadow exclusion, and t-distributed stochastic neighbor embedding (t-SNE) for exclusion of outliers, is estimated for analysis of chest X-ray (CXR) 2D images by deep learning approach to help radiologists identify marks of lung cancer in CXR. Training and validation of the simple convolutional neural network (CNN) was performed on the open JSRT dataset (dataset #01), the JSRT after bone shadow exclusion - BSE-JSRT (dataset #02), JSRT after lung segmentation (dataset #03), BSE-JSRT after lung segmentation (dataset #04), and segmented BSE-JSRT after exclusion of outliers by t-SNE method (dataset #05). The results demonstrate that the pre-processed dataset obtained after lung segmentation, bone shadow exclusion, and filtering out the outliers by t-SNE (dataset #05) demonstrates the highest training rate and best accuracy in comparison to the other pre-processed datasets.