&lt;title&gt;Convolution neural-network-based detection of lung structures&lt;/title&gt;

Hasegawa, Akira; Lo, Shih-Chung Benedict; Freedman, Matthew T.; Mun, Seong Ki

doi:10.1117/12.175101

Cited by 27 publications

(12 citation statements)

References 7 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…One is the pixelwise segmentation based on its image features such as density, histogram, entropy, gradients and co-occurrence matrix, etc. [6][7][8] Those image feature vectors are used to train and test the classifiers, e.g., a feed-forward backpropagation neural network. Although this method is stable, its output gray scale images from neural networks have coarse lung contours.…”

Section: Introductionmentioning

confidence: 99%

Lung contour detection in chest radiographs using <inline-formula>1-<roman>D</roman></inline-formula> convolution neural networks

Tsujii

1999

J. Electron. Imaging

View full text Add to dashboard Cite

The purposes of this research are to investigate the effectiveness of our novel lung contour detection method in chest radiographs. The proposed method consists of five sections as follows. First, in order to reduce the amount of information, the images are smoothed and subsampled from 2 k by 2.5 k pixels to 256 by 310 pixels with rescaling from 12-bit to 8-bit based on the image maximum and minimum. Second, that the image is resolved into the profiles for two directions (i.e., horizontal x and vertical y axes). Then, for each direction, those profiles are tested by the neural network which has been trained using the profiles from 14 pairs of original and target images. Note that both horizontal and vertical neural networks are trained with the horizontal and vertical profiles, respectively. For each direction, the whole two-dimensional image is reconstructed from the output profiles of the neural network. Next, the binarization process followed by the labeling process is applied to each reconstructed image individually. Finally, two postprocessed images are combined through the OR operation, and the labeling process is performed for the combined image to get the final contour. A total of 85 screening chest radiographs from Johns Hopkins University Hospital were digitized to 2 k by 2.5 k pixels with 12-bit gray scale. Fourteen images were used for the training of the neural networks and the remaining 71 images for testing. The proposed method can detect the lung contour at 94% accuracy for test images following the same rules as for the training images.

show abstract

Section: Introductionmentioning

confidence: 99%

Lung contour detection in chest radiographs using <inline-formula>1-<roman>D</roman></inline-formula> convolution neural networks

Tsujii

1999

J. Electron. Imaging

View full text Add to dashboard Cite

show abstract

“…For the segmentation of lung fields, such schemes have been proposed by Xu et al [269,270], Duryea and Boone [65], and Carrascal et al [32]. Lung segmentation by pixel classification using neural networks has been investigated by McNittGray [179,178], Hasegawa et al [103], and Tsujii et al [235]. Vittitoe et al [250], [251] developed a pixel classifier for the identification of lung regions using Markov random field modeling.…”

Section: Conclusion and Recommendationsmentioning

confidence: 99%

“…Most algorithms for the segmentation of lung fields fall in this category [198,269,270,65,4,6,7,32,82,84]. Techniques employed are (local) thresholding, region growing, edge detection, ridge detection, morphologi-RB PC PA / lateral evaluation remarks Toriwaki [233,232] × PA none describes a complete analysis system Harlow [102] × PA none Chien [37,38] × PA none only right lung, result used to detect abnormalities Hasegawa [103] × PA none Pietka [198] × PA none McNitt-Gray [179,178] × PA 16 Q uses 5 anatomical classes Duryea [65] × PA 802 Q Xu [269] × PA 1000 R outer rib cage only Xu [270] × PA 300 R diaphragm edges only Armato [7] × PA 600 R costophrenic angles only Armato [4] × lateral 200 Q Armato [6] × PA 600 R Carrascal [32] × both 65 RQ Vittitoe [250] × PA 99 Q Tsujii [235] × PA 71 Q Wilson, Brown [265,28] × PA none describes a complete analysis system Vittitoe [251] × PA 115 Q uses 6 anatomical classes Van Ginneken [82,84] (Chapter 3)…”

Section: Segmentation Lung Fieldsmentioning

confidence: 99%

Computer aided diagnosis in chest radiography

Mouton¹,

Douglas

2003

Academic Radiology

View full text Add to dashboard Cite

“…Rule-based schemes have been proposed by Xu et al [2,3], Duryea and Boone [4], and Carrascal et al [5]. Lung segmentation by pixel classification has been investigated by McNitt-Gray [6,7], Hasegawa et al [8], Tsujii et al [9], and Vittitoe et al [10].…”

Section: Introductionmentioning

confidence: 99%

Automatic Segmentation of Lung Fields in Chest Radiographs

Ginneken

Romeny

1999

Medical Image Computing and Computer-Assisted Intervention – MICCAI’99

View full text Add to dashboard Cite

Abstract. We present algorithms for the automatic delineation of lung fields in chest radiographs. We first develop a rule-based scheme that detects lung contours using a general framework for the detection of oriented edges and ridges. This algorithm is compared to several pixel classifiers using different combinations of features. We propose a hybrid system that combines both approaches. The performance of each system is compared with interobserver variability and results available from the literature. Our hybrid scheme turns out to be accurate and robust; the accuracy is 0.969 ± 0.00803, and above 94% for all 115 test images.

show abstract

<title>Convolution neural-network-based detection of lung structures</title>

Cited by 27 publications

References 7 publications

Lung contour detection in chest radiographs using <inline-formula>1-<roman>D</roman></inline-formula> convolution neural networks

Lung contour detection in chest radiographs using <inline-formula>1-<roman>D</roman></inline-formula> convolution neural networks

Computer aided diagnosis in chest radiography

Automatic Segmentation of Lung Fields in Chest Radiographs

Contact Info

Product

Resources

About