Detection, growth quantification and malignancy prediction of pulmonary nodules using deep convolutional networks in follow-up CT scans

TL;DR

This paper presents an automated deep learning pipeline for detecting, quantifying growth, and predicting malignancy of lung nodules in follow-up CT scans, aiding radiologists in lung cancer management.

Contribution

It introduces a novel hierarchical probabilistic U-Net for growth detection with uncertainty estimates and a two-stream 3D-CNN for malignancy classification integrating pretrained probabilities.

Findings

Pipeline achieved performance comparable to state-of-the-art methods.

Novel growth detection approach with uncertainty estimation.

Effective integration of malignancy probabilities improves classification.

Abstract

We address the problem of supporting radiologists in the longitudinal management of lung cancer. Therefore, we proposed a deep learning pipeline, composed of four stages that completely automatized from the detection of nodules to the classification of cancer, through the detection of growth in the nodules. In addition, the pipeline integrated a novel approach for nodule growth detection, which relied on a recent hierarchical probabilistic U-Net adapted to report uncertainty estimates. Also, a second novel method was introduced for lung cancer nodule classification, integrating into a two stream 3D-CNN network the estimated nodule malignancy probabilities derived from a pretrained nodule malignancy network. The pipeline was evaluated in a longitudinal cohort and reported comparable performances to the state of art.