Predicting lung cancer survival with attention-based CT slices combination

TL;DR

This paper presents a new method for predicting lung cancer survival using CT scans and attention mechanisms, outperforming existing techniques.

Contribution

A novel attention-based approach for combining CT slices to predict 2-year survival in NSCLC patients.

Findings

The method achieved a Ctd-index of 0.584 on the LUNG1 dataset, outperforming 3D networks.

Using transfer learning improved performance on a private dataset by 0.076 in Ctd-index.

The approach is adaptable with different 2D backbones, showing better results than traditional 3D methods.

Abstract

Accurate prognosis of Non-Small Cell Lung Cancer (NSCLC) is crucial for enhancing patient care and treatment outcomes. Despite the advancements in deep learning, the task of overall survival prediction in NSCLC has not fully leveraged these techniques, yet. This study introduces a novel methodology for predicting 2-year overall survival (OS) in NSCLC patients using CT scans. Our approach integrates CT scan representations produced by EfficientNetB0 with a soft attention mechanism to identify the most relevant slices for survival risk prediction, which are then analyzed by a risk-assessment network. To validate our method and ensure reproducibility, we employed the public LUNG1 dataset and a smaller private dataset. Our approach was compared to benchmark 3D networks and two variants of our methodology: on the LUNG1 it outperformed the competitors achieving a mean \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$C^{td}$$\end{document}Ctd-index of 0.584 over tenfold cross-validation. On the LUNG1 we also demonstrated the adaptability of our method with 5 other 2D backbones replacing the EfficientNetB0, confirming that our mechanism of combining 2D slice representations to construct a 3D volume representation is more effective for OS prediction compared to a traditional 3D approach. Finally, we used transfer learning on the private dataset, showing that it can significantly enhance performance in limited data scenarios, increasing the \documentclass[12pt]{minimal}
\usepackage{amsmath}
\usepackage{wasysym}
\usepackage{amsfonts}
\usepackage{amssymb}
\usepackage{amsbsy}
\usepackage{mathrsfs}
\usepackage{upgreek}
\setlength{\oddsidemargin}{-69pt}
\begin{document}$$C^{td}$$\end{document}Ctd-index by 0.076 compared to model without transfer learning.

Figures