Dimensionality Reduction in Deep Learning for Chest X-Ray Analysis of Lung Cancer
This incremental work addresses the problem of improving lung cancer detection accuracy for radiologists by optimizing pre-processing steps in medical imaging.
The study evaluated the efficiency of combining lung segmentation, bone shadow exclusion, and t-SNE outlier filtering for dimensionality reduction in deep learning-based chest X-ray analysis of lung cancer, finding that this combined pre-processing method achieved the highest training rate and best accuracy compared to other datasets.
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.