LF-SegNet: A Fully Convolutional Encoder–Decoder Network for Segmenting Lung Fields from Chest Radiographs

Segmentation of lung fields is an important pre-requisite step in chest radiographic computer-aided diagnosis systems as it precisely defines the region-of-interest on which different operations are applied. However, it is immensely challenging due to extreme variations in shape and size of lungs. Manual segmentation is also prone to large inter-observer and intra-observer variations. Thus, an automated method for lung field segmentation with sufficiently high accuracy is unsparingly required. This paper presents a deep learning-based fully convolutional encoder-decoder network for segmenting lung fields from chest radiographs. The major contribution of this work is in the unique design of the encoder-decoder network that makes it especially suitable for lung field segmentation. The proposed network is trained, tested and evaluated on publicly available standard datasets. The result of evaluation indicates that the performance of the proposed method, i.e. accuracy of 98.73% and overlap of 95.10%, is better than state-of-the-art methods.

[1]  S. Armato,et al.  Automated lung segmentation in digitized posteroanterior chest radiographs. , 1998, Academic radiology.

[2]  Isabelle Borget,et al.  Evaluation of the accuracy of a computer-aided diagnosis (CAD) system in breast ultrasound according to the radiologist's experience. , 2012, Academic radiology.

[3]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[4]  Edmund A Franken,et al.  Satisfaction of Search in Chest Radiography 2015. , 2015, Academic radiology.

[5]  Roberto Cipolla,et al.  SegNet: A Deep Convolutional Encoder-Decoder Architecture for Robust Semantic Pixel-Wise Labelling , 2015, CVPR 2015.

[6]  John W. Robinson,et al.  Reporting instructions significantly impact false positive rates when reading chest radiographs , 2016, European Radiology.

[7]  Bram van Ginneken,et al.  Segmentation of anatomical structures in chest radiographs using supervised methods: a comparative study on a public database , 2006, Medical Image Anal..

[8]  H. K. Huang,et al.  Feature selection in the pattern classification problem of digital chest radiograph segmentation , 1995, IEEE Trans. Medical Imaging.

[9]  Ellen M. Kok,et al.  Does the Use of a Checklist Help Medical Students in the Detection of Abnormalities on a Chest Radiograph? , 2017, Journal of Digital Imaging.

[10]  Clement J. McDonald,et al.  Lung Segmentation in Chest Radiographs Using Anatomical Atlases With Nonrigid Registration , 2014, IEEE Transactions on Medical Imaging.

[11]  Luis Perez,et al.  The Effectiveness of Data Augmentation in Image Classification using Deep Learning , 2017, ArXiv.

[12]  Dinggang Shen,et al.  Segmenting Lung Fields in Serial Chest Radiographs Using Both Population-Based and Patient-Specific Shape Statistics , 2008, IEEE Transactions on Medical Imaging.

[13]  Clarimar José Coelho,et al.  Computer-aided diagnosis in chest radiography for detection of childhood pneumonia , 2008, Int. J. Medical Informatics.

[14]  Ajay Mittal,et al.  Lung field segmentation in chest radiographs: a historical review, current status, and expectations from deep learning , 2017, IET Image Process..

[15]  Z. Qin,et al.  An evaluation of automated chest radiography reading software for tuberculosis screening among public- and private-sector patients , 2017, European Respiratory Journal.

[16]  M. Myles-Worsley,et al.  The influence of expertise on X-ray image processing. , 1988, Journal of experimental psychology. Learning, memory, and cognition.

[17]  Stefan Jaeger,et al.  Two public chest X-ray datasets for computer-aided screening of pulmonary diseases. , 2014, Quantitative imaging in medicine and surgery.

[18]  Sergey Ioffe,et al.  Batch Normalization: Accelerating Deep Network Training by Reducing Internal Covariate Shift , 2015, ICML.

[19]  M. Kallergi,et al.  Improved method for automatic identification of lung regions on chest radiographs. , 2001 .

[20]  Kenji Suzuki,et al.  Computer-Aided Detection of Lung Cancer , 2017 .

[21]  O. Eden,et al.  Inter-Observer Variation in Interpretation of Chest X-Rays , 1990, Scottish medical journal.

[22]  Roberto Cipolla,et al.  SegNet: A Deep Convolutional Encoder-Decoder Architecture for Image Segmentation , 2015, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[23]  D. Sivaganesan,et al.  Wireless Distributive Personal Communication for Early Detection of Collateral Cancer Using Optimized Machine Learning Methodology , 2016, Wireless Personal Communications.

[24]  B. van Ginneken,et al.  An automated tuberculosis screening strategy combining X-ray-based computer-aided detection and clinical information , 2016, Scientific Reports.

[25]  Algimantas Juozapavičius,et al.  Computer-aided detection of interstitial lung diseases: A texture approach , 2017 .

[26]  Alejandro F. Frangi,et al.  Active shape model segmentation with optimal features , 2002, IEEE Transactions on Medical Imaging.

[27]  S K Mun,et al.  Automated segmentation of anatomic regions in chest radiographs using an adaptive-sized hybrid neural network. , 1998, Medical physics.

[28]  Dumitru Erhan,et al.  Going deeper with convolutions , 2014, 2015 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[29]  Geoffrey E. Hinton,et al.  ImageNet classification with deep convolutional neural networks , 2012, Commun. ACM.

[30]  Hidenori Itoh,et al.  Lung Segmentation in Chest Radiographs by Means of Gaussian Kernel-Based FCM with Spatial Constraints , 2009, 2009 Sixth International Conference on Fuzzy Systems and Knowledge Discovery.

[31]  Mohammad Faizal Ahmad Fauzi,et al.  Lung segmentation on standard and mobile chest radiographs using oriented Gaussian derivatives filter , 2015 .

[32]  D. S. Morillo,et al.  Computer-aided diagnosis of pneumonia in patients with chronic obstructive pulmonary disease. , 2013, Journal of the American Medical Informatics Association : JAMIA.

[33]  Chirag Agarwal,et al.  Accurate segmentation of lung fields on chest radiographs using deep convolutional networks , 2017, Medical Imaging.

[34]  Jimmy Ba,et al.  Adam: A Method for Stochastic Optimization , 2014, ICLR.

[35]  Jun-ichiro Toriwaki,et al.  Pattern recognition of chest X-ray images , 1973, Comput. Graph. Image Process..

[36]  Anup Basu,et al.  Gradient vector flow based active shape model for lung field segmentation in chest radiographs , 2009, 2009 Annual International Conference of the IEEE Engineering in Medicine and Biology Society.

[37]  Jihoon Kim,et al.  Using a Method Based on a Modified K-Means Clustering and Mean Shift Segmentation to Reduce File Sizes and Detect Brain Tumors from Magnetic Resonance (MRI) Images , 2016, Wirel. Pers. Commun..

[38]  Jian Sun,et al.  Deep Residual Learning for Image Recognition , 2015, 2016 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[39]  B. van Ginneken,et al.  Diagnostic Accuracy of Computer-Aided Detection of Pulmonary Tuberculosis in Chest Radiographs: A Validation Study from Sub-Saharan Africa , 2014, PloS one.

[40]  Andrew Zisserman,et al.  Very Deep Convolutional Networks for Large-Scale Image Recognition , 2014, ICLR.

[41]  K. Doi,et al.  Development of a digital image database for chest radiographs with and without a lung nodule: receiver operating characteristic analysis of radiologists' detection of pulmonary nodules. , 2000, AJR. American journal of roentgenology.

[42]  Ling Mao,et al.  A region based active contour method for x-ray lung segmentation using prior shape and low level features , 2010, 2010 IEEE International Symposium on Biomedical Imaging: From Nano to Macro.

[43]  David A. Spencer,et al.  Accuracy of the Interpretation of Chest Radiographs for the Diagnosis of Paediatric Pneumonia , 2014, PloS one.