Lung Nodule Classification Using Biomarkers, Volumetric Radiomics and 3D CNNs

TL;DR

This paper introduces a hybrid algorithm combining imaging biomarkers, volumetric radiomics, and 3D CNNs to classify lung nodule malignancy, analyzing various combinations and the impact of semi-supervised learning on accuracy.

Contribution

The study evaluates the effectiveness of integrating biomarkers, radiomics, and CNNs for lung nodule classification, revealing unexpected insights about biomarker-only models and semi-supervised learning effects.

Findings

Biomarker-only models outperform combined models in accuracy.

Semi-supervised learning with KNN improves model performance.

Combining imagery, biomarkers, and radiomics yields varied results depending on the approach.

Abstract

We present a hybrid algorithm to estimate lung nodule malignancy that combines imaging biomarkers from Radiologist's annotation with image classification of CT scans. Our algorithm employs a 3D Convolutional Neural Network (CNN) as well as a Random Forest in order to combine CT imagery with biomarker annotation and volumetric radiomic features. We analyze and compare the performance of the algorithm using only imagery, only biomarkers, combined imagery + biomarkers, combined imagery + volumetric radiomic features and finally the combination of imagery + biomarkers + volumetric features in order to classify the suspicion level of nodule malignancy. The National Cancer Institute (NCI) Lung Image Database Consortium (LIDC) IDRI dataset is used to train and evaluate the classification task. We show that the incorporation of semi-supervised learning by means of K-Nearest-Neighbors (KNN) can increase the available training sample size of the LIDC-IDRI thereby further improving the accuracy of malignancy estimation of most of the models tested although there is no significant improvement with the use of KNN semi-supervised learning if image classification with CNNs and volumetric features are combined with descriptive biomarkers. Unexpectedly, we also show that a model using image biomarkers alone is more accurate than one that combines biomarkers with volumetric radiomics, 3D CNNs, and semi-supervised learning. We discuss the possibility that this result may be influenced by cognitive bias in LIDC-IDRI because malignancy estimates were recorded by the same radiologist panel as biomarkers, as well as future work to incorporate pathology information over a subset of study participants.