Automatic lung segmentation in routine imaging is primarily a data diversity problem, not a methodology problem

Abstract

Automated segmentation of anatomical structures is a crucial step in image analysis. For lung segmentation in computed tomography, a variety of approaches exist, involving sophisticated pipelines trained and validated on different datasets. However, the clinical applicability of these approaches across diseases remains limited. We compared four generic deep learning approaches trained on various datasets and two readily available lung segmentation algorithms. We performed evaluation on routine imaging data with more than six different disease patterns and three published data sets. Using different deep learning approaches, mean Dice similarity coefficients (DSCs) on test datasets varied not over 0.02. When trained on a diverse routine dataset (n = 36) a standard approach (U-net) yields a higher DSC (0.97 $\pm$ 0.05) compared to training on public datasets such as Lung Tissue Research Consortium (0.94 $\pm$ 0.13, p = 0.024) or Anatomy 3 (0.92 $\pm$ 0.15, p = 0.001). Trained on routine data (n = 231) covering multiple diseases, U-net compared to reference methods yields a DSC of 0.98 $\pm$ 0.03 versus 0.94 $\pm$ 0.12 (p = 0.024).

10 pages, 5 figures, 5 tables

References in corpus (7)

• Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift
• Confounding variables can degrade generalization performance of radiological deep learning models
• Multi-Scale Context Aggregation by Dilated Convolutions
• Training Very Deep Networks
• The Cityscapes Dataset for Semantic Urban Scene Understanding
• nnU-Net: Self-adapting Framework for U-Net-Based Medical Image Segmentation
• Progressive and Multi-Path Holistically Nested Neural Networks for Pathological Lung Segmentation from CT Images

Cited by in corpus (34)

• TotalSegmentator: robust segmentation of 104 anatomical structures in CT images
• A Survey of Deep Learning Techniques for the Analysis of COVID-19 and their usability for Detecting Omicron
• Unsupervised Contrastive Learning based Transformer for Lung Nodule Detection
• GaNDLF: A Generally Nuanced Deep Learning Framework for Scalable End-to-End Clinical Workflows in Medical Imaging
• COVID-MTL: Multitask Learning with Shift3D and Random-weighted Loss for Automated Diagnosis and Severity Assessment of COVID-19
• Robust deformable image registration using cycle-consistent implicit representations
• Hunting imaging biomarkers in pulmonary fibrosis: Benchmarks of the AIIB23 challenge
• Mediastinal lymph nodes segmentation using 3D convolutional neural network ensembles and anatomical priors guiding
• GASNet: Weakly-supervised Framework for COVID-19 Lesion Segmentation
• Multi-stage intermediate fusion for multimodal learning to classify non-small cell lung cancer subtypes from CT and PET
• Multi-modal Medical Image Fusion For Non-Small Cell Lung Cancer Classification
• Context Matters: Graph-based Self-supervised Representation Learning for Medical Images
• Validated respiratory drug deposition predictions from 2D and 3D medical images with statistical shape models and convolutional neural networks
• Emphysema Subtyping on Thoracic Computed Tomography Scans using Deep Neural Networks
• Representation Matters: Assessing the Importance of Subgroup Allocations in Training Data
• Automated lung segmentation from CT images of normal and COVID-19 pneumonia patients
• Imbalance-Aware Self-Supervised Learning for 3D Radiomic Representations
• Estimating Total Lung Volume from Pixel-level Thickness Maps of Chest Radiographs Using Deep Learning
• CT respiratory motion synthesis using joint supervised and adversarial learning
• CAE-Transformer: Transformer-based Model to Predict Invasiveness of Lung Adenocarcinoma Subsolid Nodules from Non-thin Section 3D CT Scans
• COVID-Rate: An Automated Framework for Segmentation of COVID-19 Lesions from Chest CT Scans
• Improve Cross-Modality Segmentation by Treating T1-Weighted MRI Images as Inverted CT Scans
• Equivariant neural networks for inverse problems
• Few-Shot Airway-Tree Modeling using Data-Driven Sparse Priors
• Self-supervised Lesion Change Detection and Localisation in Longitudinal Multiple Sclerosis Brain Imaging
• Temporal Periodic Image Registration with Implicit Neural Representations
• Exploring constraints on CycleGAN-based CBCT enhancement for adaptive radiotherapy
• COVID-view: Diagnosis of COVID-19 using Chest CT
• Virtual Scanning for NSCLC Histology: Investigating the Discriminatory Power of Synthetic PET
• Imaging-Based Mortality Prediction in Patients with Systemic Sclerosis
• Bronchovascular Tree-Guided Weakly Supervised Learning Method for Pulmonary Segment Segmentation
• Can Foundation Models Really Segment Tumors? A Benchmarking Odyssey in Lung CT Imaging
• Deformable Image Registration with Stochastically Regularized Biomechanical Equilibrium
• Separation of target anatomical structure and occlusions in chest radiographs