Full-resolution Lung Nodule Segmentation from Chest X-ray Images using Residual Encoder-Decoder Networks

Michael James Horry,Subrata Chakraborty,Biswajeet Pradhan,Manoranjan Paul,Jing Zhu,Prabal Datta Barua,U. Rajendra Acharya,Fang Chen,Jianlong Zhou

2023-07-13

Abstract:Lung cancer is the leading cause of cancer death and early diagnosis is associated with a positive prognosis. Chest X-ray (CXR) provides an inexpensive imaging mode for lung cancer diagnosis. Suspicious nodules are difficult to distinguish from vascular and bone structures using CXR. Computer vision has previously been proposed to assist human radiologists in this task, however, leading studies use down-sampled images and computationally expensive methods with unproven generalization. Instead, this study localizes lung nodules using efficient encoder-decoder neural networks that process full resolution images to avoid any signal loss resulting from down-sampling. Encoder-decoder networks are trained and tested using the JSRT lung nodule dataset. The networks are used to localize lung nodules from an independent external CXR dataset. Sensitivity and false positive rates are measured using an automated framework to eliminate any observer subjectivity. These experiments allow for the determination of the optimal network depth, image resolution and pre-processing pipeline for generalized lung nodule localization. We find that nodule localization is influenced by subtlety, with more subtle nodules being detected in earlier training epochs. Therefore, we propose a novel self-ensemble model from three consecutive epochs centered on the validation optimum. This ensemble achieved a sensitivity of 85% in 10-fold internal testing with false positives of 8 per image. A sensitivity of 81% is achieved at a false positive rate of 6 following morphological false positive reduction. This result is comparable to more computationally complex systems based on linear and spatial filtering, but with a sub-second inference time that is faster than other methods. The proposed algorithm achieved excellent generalization results against an external dataset with sensitivity of 77% at a false positive rate of 7.6.

Image and Video Processing,Computer Vision and Pattern Recognition,Machine Learning

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to automatically locate lung nodules in chest X - ray images. Specifically, the research aims to develop an efficient and generalizable computer vision method or architecture that can locate lung nodules from full - resolution chest X - ray images, and at the same time, the generalization ability of the algorithm is more important than in - sample fitting. The paper pays special attention to the following points: 1. **Establish an end - to - end deep - learning encoder - decoder network**: It is proved for the first time that an end - to - end deep - learning encoder - decoder network trained on a small data set can effectively segment and locate lung nodules. 2. **Use full - resolution images**: Research shows that using full - resolution images as input images can achieve the best results, which questions the common practice of image down - sampling. Processing large images by adding convolutional layers instead of down - sampling is a better strategy. 3. **The influence of image pre - processing steps**: It systematically shows the significant influence of these intuitive but unvalidated image pre - processing steps such as lung region segmentation and histogram equalization on the generalization ability of the algorithm. 4. **Algorithm performance evaluation**: The research also explores the performance of the algorithm in the following aspects: - Rating of the subtlety of nodules - Malignancy of nodules - Location of nodules - Diagnosis of patients - Gender of patients Through these studies, the paper hopes to provide an efficient and generalizable solution that can be actually applied in a clinical environment and improve the accuracy and efficiency of lung nodule detection.

Full-resolution Lung Nodule Segmentation from Chest X-ray Images using Residual Encoder-Decoder Networks

Detection-Guided Deep Learning-Based Model with Spatial Regularization for Lung Nodule Segmentation

Lung Nodule Detection via Deep Reinforcement Learning

Debiasing pipeline improves deep learning model generalization for X-ray based lung nodule detection

Lung Nodule Segmentation Combining Sobel Operator and Mask R-CNN

Automated Pulmonary Nodule Detection in CT Images Using Deep Convolutional Neural Networks.

Deep Deconvolutional Residual Network Based Automatic Lung Nodule Segmentation

Pulmonary Nodule Classification Using a Multiview Residual Selective Kernel Network

Identification of Pulmonary Nodules Via CT Images with Hierarchical Fully Convolutional Networks.

P3.13-037 Deep Learning System for Lung Nodule Detection

Multi-Level Cross Residual Network for Lung Nodule Classification

Automatic Lung Nodule Detection Using a 3D Deep Convolutional Neural Network Combined with a Multi-Scale Prediction Strategy in Chest CTs

DeepSEED: 3D Squeeze-and-Excitation Encoder-Decoder Convolutional Neural Networks for Pulmonary Nodule Detection

Detection of Pulmonary Nodules on Chest X-ray with Fusion of Location and Scale Informa-tion

Lung Nodule Detection Based on the Combination of Morphometric and Texture Features

Lung cancer classification and identification framework with automatic nodule segmentation screening using machine learning

Highly accurate model for prediction of lung nodule malignancy with CT scans

Lung Nodule Diagnosis on 3D Computed Tomography Images Using Deep Convolutional Neural Networks

Deep convolutional neural networks for multiplanar lung nodule detection: Improvement in small nodule identification

A Novel Deep Learning Model Based on Multi-Scale and Multi-View for Detection of Pulmonary Nodules

MEDS-Net: Self-Distilled Multi-Encoders Network with Bi-Direction Maximum Intensity projections for Lung Nodule Detection