Elsevier

Academic Radiology

Volume 8, Issue 7, July 2001, Pages 629-638
Academic Radiology

Improved Method for Automatic Identification of Lung Regions on Chest Radiographs

https://doi.org/10.1016/S1076-6332(03)80688-8Get rights and content

Abstract

Rationale and Objectives

The authors performed this study to evaluate an algorithm developed to help identify lungs on chest radiographs.

Materials and Methods

Forty clinical posteroanterior chest radiographs obtained in adult patients were digitized to 12-bit gray-scale resolution. In the proposed algorithm, the authors simplified the current approach of edge detection with derivatives by using only the first derivative of the horizontal and/or vertical image profiles. In addition to the derivative method, pattern classification and image feature analysis were used to determine the region of interest and lung boundaries. Instead of using the traditional curve-fitting method to delineate the lung, the authors applied an iterative contour-smoothing algorithm to each of the four detected boundary segments (costal, mediastinal, lung apex, and hemidiaphragm edges) to form a smooth lung boundary.

Results

The algorithm had an average accuracy of 96.0% for the right lung and 95.2% for the left lung and was especially useful in the delineation of hemidiaphragm edges. In addition, it took about 0.775 second per image to identify the lung boundaries, which is much faster than that of other algorithms noted in the literature.

Conclusion

The computer-generated segmentation results can be used directly in the detection and compensation of rib structures and in lung nodule detection.

Section snippets

Data Set

Forty adult posteroanterior chest radiographs were used in this study. All radiographs were obtained from the research database at the University of South Florida Medical School, Tampa, and digitized by using a scanner (Lumiscan-75; Lumisys, Tucson, Ariz) with a pixel size of 100 μm. The digital chest images had 12-bit gray-scale resolution. For better viewing and faster processing, each input image was first subsampled to one-fourth of the original size (average size, 2,000 × 2,400). Although

Results

We performed a pixel-by-pixel analysis to qualify the performance of the algorithm (6). Each pixel of an image was placed into one of four categories: true-positive, where both the algorithm and the human observer considered the pixel to be within the lung; true-negative, where both considered the pixel to be outside the lung; false-positive, where the computer found the pixel inside the lung and the observer did not; and false-negative, where the observer detected the pixel in the lung but the

Discussion

First derivatives are usually considered unreliable in edge detection because they are based on the local feature of the image and are sensitive to noise. We successfully used first derivatives, however, after making some modifications on the basis of image feature analysis and pattern classification. In this algorithm, we first reduced the irrelevant information (including the noise) contained in the image by identifying the ROI (the lung region). Then, we used first derivatives of the image

References (13)

  • MS Brown et al.

    Knowledge-based method for segmentation and analysis of lung boundaries in chest x-ray images

    Comput Med Imaging Graph

    (1998)
  • T Ishida et al.

    Computerized analysis of interstitial disease in chest radiographs: improvement of geometric-pattern feature analysis

    Med Phys

    (1997)
  • MJ Carreira et al.

    Computer-aided diagnoses: automatic detection of lung nodules

    Med Phys

    (1998)
  • N Nakamori et al.

    Image feature analysis and computer-aided diagnosis in digital radiography: automated analysis of size of heart and lung in chest images

    Med Phys

    (1990)
  • O Tsujii et al.

    Automated segmentation of anatomic regions in chest radiographs using an adaptive-sized hybrid neural network

    Med Phys

    (1998)
  • MF McNitt-Gray et al.

    Feature selection in the pattern classification problem of digital chest radiograph segmentation

    IEEE Trans Med Imaging

    (1995)
There are more references available in the full text version of this article.

Cited by (77)

View all citing articles on Scopus
View full text