Skip to main page content
U.S. flag

An official website of the United States government

Dot gov

The .gov means it’s official.
Federal government websites often end in .gov or .mil. Before sharing sensitive information, make sure you’re on a federal government site.

Https

The site is secure.
The https:// ensures that you are connecting to the official website and that any information you provide is encrypted and transmitted securely.

Access keys NCBI Homepage MyNCBI Homepage Main Content Main Navigation
. 2023 Jun;36(3):1237-1247.
doi: 10.1007/s10278-022-00772-y. Epub 2023 Jan 25.

Effect of Contrast Level and Image Format on a Deep Learning Algorithm for the Detection of Pneumothorax with Chest Radiography

Myeong Seong Yoon #  1   2   3 , Gitaek Kwon #  4   5 , Jaehoon Oh  6   7 , Jongbin Ryu  8 , Jongwoo Lim  4   2 , Bo-Kyeong Kang  2   9 , Juncheol Lee  1 , Dong-Kyoon Han  3
Affiliations

Effect of Contrast Level and Image Format on a Deep Learning Algorithm for the Detection of Pneumothorax with Chest Radiography

Myeong Seong Yoon et al. J Digit Imaging. 2023 Jun.

Abstract

Under the black-box nature in the deep learning model, it is uncertain how the change in contrast level and format affects the performance. We aimed to investigate the effect of contrast level and image format on the effectiveness of deep learning for diagnosing pneumothorax on chest radiographs. We collected 3316 images (1016 pneumothorax and 2300 normal images), and all images were set to the standard contrast level (100%) and stored in the Digital Imaging and Communication in Medicine and Joint Photographic Experts Group (JPEG) formats. Data were randomly separated into 80% of training and 20% of test sets, and the contrast of images in the test set was changed to 5 levels (50%, 75%, 100%, 125%, and 150%). We trained the model to detect pneumothorax using ResNet-50 with 100% level images and tested with 5-level images in the two formats. While comparing the overall performance between each contrast level in the two formats, the area under the receiver-operating characteristic curve (AUC) was significantly different (all p < 0.001) except between 125 and 150% in JPEG format (p = 0.382). When comparing the two formats at same contrast levels, AUC was significantly different (all p < 0.001) except 50% and 100% (p = 0.079 and p = 0.082, respectively). The contrast level and format of medical images could influence the performance of the deep learning model. It is required to train with various contrast levels and formats of image, and further image processing for improvement and maintenance of the performance.

Keywords: Artificial intelligence; Contrast level; Deep learning; Image format; Pneumothorax.

PubMed Disclaimer

Conflict of interest statement

The authors declare no competing interests.

Figures

Fig. 1
Fig. 1
Flow chart of data collection and analysis in research for pneumothorax detection with deep learning algorithms
Fig. 2
Fig. 2
Overview of detailed workflow used to extract the features of the image. In our workflow, ResNet-50 [31] was utilized as our backbone networks
Fig. 3
Fig. 3
Receiver-operator characteristics (ROC) curves of overall performance for the detection of pneumothorax using ResNet-50 network with chest radiograph over five tests according to 50%, 75%, 100%, 125%, and 150% contrast level and image formats. (a) DICOM format, (b) JPEG format
Fig. 4
Fig. 4
Confusion matrices of the test performance for the detection of pneumothorax using ResNet-50 network with five contrast level images (50%, 75%, 100%, 125%, and 150%) in (a) DICOM and (b) JPEG. P, positive; N, negative
Fig. 5
Fig. 5
Examples of right and wrong detection of pneumothorax using ResNet-50 network according to contrast level of the chest radiograph. A On chest radiography with a pneumothorax in the right lung, all images were detected to pneumothorax according to all contrast levels. B On chest radiography with a pneumothorax in the left lung, standard and dark images of 50% and 75% contrast levels were detected as pneumothorax, but light images of 125% and 150% contrast levels were predicted as normal images. C On chest radiography with normal images, images of 50%, 75%, and 150% contrast levels were predicted as pneumothorax. Blue arrows indicate the visceral pleural line of the pneumothorax and red boxes indicate images of false positive or negative
See this image and copyright information in PMC

Similar articles

See all similar articles

References

    1. Mintz Y, Brodie R. Introduction to artificial intelligence in medicine. Minim Invasive Ther Allied Technol. 2019;28(2):73–81. doi: 10.1080/13645706.2019.1575882. - DOI - PubMed
    1. Hosny A, Parmar C, Quackenbush J, Schwartz LH, Aerts HJWL. Artificial intelligence in radiology. Nat Rev Cancer. 2018;18(8):500–510. doi: 10.1038/s41568-018-0016-5. - DOI - PMC - PubMed
    1. Syed AB, Zoga AC. Artificial Intelligence in Radiology: Current Technology and Future Directions. Semin Musculoskelet Radiol. 2018;22(5):540–545. doi: 10.1055/s-0038-1673383. - DOI - PubMed
    1. Chassagnon G, Vakalopoulou M, Paragios N, Revel MP. Artificial intelligence applications for thoracic imaging. Eur J Radiol. 2020;123:108774. doi: 10.1016/j.ejrad.2019.108774. - DOI - PubMed
    1. Hwang EJ, Park S, Jin KN, Kim JI, Choi SY, Lee JH, et al: Deep Learning-Based Automatic Detection Algorithm Development and Evaluation Group. Development and Validation of a Deep Learning-based Automatic Detection Algorithm for Active Pulmonary Tuberculosis on Chest Radiographs. Clin Infect Dis 69(5):739–747, 2019 - PMC - PubMed

Publication types