Elsevier

Academic Radiology

Volume 15, Issue 8, August 2008, Pages 1004-1016
Academic Radiology

Original investigation
Classification of Parenchymal Abnormality in Scleroderma Lung Using a Novel Approach to Denoise Images Collected via a Multicenter Study1

https://doi.org/10.1016/j.acra.2008.03.011Get rights and content

Rationale and Objectives

Computerized classification techniques have been developed to offer accurate and robust pattern recognition in interstitial lung disease using texture features. However, these techniques still present challenges when analyzing computed tomographic (CT) image data from multiprotocols because of disparate acquisition protocols or from standardized, multicenter clinical trials because of noise variability. Our objective is to investigate the utility of denoising thin section CT image data to improve the classification of scleroderma disease patterns. The patterns are lung fibrosis (LF), groundglass (GG), honeycomb (HC), or normal lung (NL) within small regions of interest (ROIs).

Methods

High-resolution CT images were scanned in a multicenter clinical trial for the Scleroderma Lung Study. A thoracic radiologist contoured a training set (38 patients) consisting of 148 ROIs with 46 LF, 85 GG, 4 HC, and 13 NL patterns and contoured a test set (33 new patients) consisting of 132 ROIs with 44 LF, 72 GG, 4 HC, and 12 NL patterns. The corresponding CT slices of a contoured ROI were denoised using Aujol's mathematic partial differential equation algorithm. The algorithm's noise parameter was estimated as the standard deviation of grey-level signal (in Hounsfield units) in a homogeneous, non-lung region: the aorta. Within each contoured ROI, every pixel within a 4 × 4 neighborhood was sampled (4 × 4 grid sampling). All sampled pixels from a contoured ROI were assumed to be the same disease pattern as labeled by the radiologist. 5,690 pixels (3,009 LF, 1,994 GG, 348 HC, and 339 NL) and 5,045 pixels (2,665 LF, 1,753 GG, 291 HC, and 336 NL) were sampled in training and test sets, respectively. Next, 58 texture features from the original and denoised image were calculated for each pixel. Using a multinomial logistic model, subsets of features (one from original and another from denoised images) were selected to classify disease patterns. Finally, pixels were classified into disease patterns using a support vector machine procedure.

Results

From the training set, multinomial logistic model selected 45 features from the original images and 38 features from denoised images to classify disease patterns. Using the test set, the overall pixel classification rate by SVM increased from 87.8% to 89.5% with denoising. The specific classification rates (original/denoised) were 96.3/96.4% for LF, 88.8/89.4% for GG, 21.3/28.9% for HC, and 73.5/88.4% for NL. Denoising significantly improved the NL and overall classification rates (P = .037 and P = .047 respectively) at ROI level.

Conclusions

Analyzing multicenter data using a denoising approach led to more parsimonious classification models with increasing accuracy. This approach offers a novel alternate classification strategy for heterogeneous technical and disease components. Furthermore, the model offers the potential to discriminate the multiple patterns of scleroderma disease correctly.

Section snippets

Materials and methods

The use of the anonymous data set from Scleroderma Lung Study (16) was approved by our local institutional review board.

Results

Representative results of implementing Aujol's algorithm to denoise CT image data are shown in Figure 4. The image pairs correspond to the original and denoised examples of abnormalities (LF, GG, and HC) and NL shown in Figure 3. In original images, the noise parameter equaled 50, which was the upper bound of SD in the aorta across all patients.

From the training set, our logistic model selected 45 and 38 texture features from original and denoised images, respectively, as significant predictors

Discussion

Computational approaches have been used to classify disease patterns in the lung, but none has investigated the influence of noise on classification accuracy. To date, most classification models are based on image data sets derived from a single platform and thus, may not be generalizable to images acquired on different platforms or to using different imaging protocols even on the same platform (3, 4, 5, 6, 7). CT image noise can be influenced by the technique used: the product of tube current

Conclusion

Accurate and robust computerized classifications based on texture features appear improved when using denoised images. This is particularly important to discriminate normal from abnormal parenchyma given that GG and other disease patterns can be mimicked by image noise, leading to false positives. Based on this initial work that has a limited number of disease types, our CT denoising method yielded higher rates of correct NL tissue classification in scleroderma patients. Texture features from a

Acknowledgments

The authors appreciate the invaluable guidance of Jean-Francois Aujol on denoise and its implementation. We also thank David Qing for programming of the contouring tools, Irene da Costa, Project Manager, Radiology, SLS, and Laura Guzman for editing and proofreading this manuscript.

References (31)

  • F. Abtin et al.

    Denoising of HRCT images by statistical postprocessing image decomposition improves detection by ground glass opacities

    (2006)
  • D.L. Donoho et al.

    Adapting to unknown smoothness via wavelet shrinkage

    J Am Stat Assoc

    (1995)
  • S.G. Mallat

    A wavelet tour of signal processing

    (1998)
  • M. Nikolova

    A variational approach to remove outliers and impulse noise

    J Math Imaging Vision

    (2004)
  • J.F. Aujol et al.

    Structure-texture image decomposition-modeling, algorithm, and parameter selection

    Int J Comput Vision

    (2006)
  • Cited by (61)

    • Cellular and Molecular Diversity in Scleroderma

      2021, Seminars in Immunology
      Citation Excerpt :

      Results of studies that assessed the efficacy of autologous stem cell transplant [53] and cyclophosphamide or mycophenolate mofetil in patients with SSc [54] utilized computer vision to quantify the extent and degree of ILD [53,55]. Chest high-resolution computed tomograpy exams of participants in the Scleroderma Lung Study I were analyzed following computer led denoising [56]. Through this approach, investigators assessed the difference in quantitative computer-aided diagnosis scores of fibrosis, ground glass, and honeycombing that summed to generate a total ILD score [55,57].

    • Deep learning for lung disease segmentation on CT: Which reconstruction kernel should be used?

      2021, Diagnostic and Interventional Imaging
      Citation Excerpt :

      Decreasing the sharpness of the reconstruction kernel reduces the amount of noise in the image, which is already known to have a significant impact on emphysema quantification [15,16]. For systemic sclerosis-related ILD, Kim et al. demonstrated that by using image denoising methods, they increased the performance of their ILD segmentation model based on texture analysis with classical machine learning methods [17]. Although images reconstructed with a MK are less noisy, our study does not demonstrate the superiority of this kernel over LK in the setting of lung disease segmentation using DL.

    • A study design for statistical learning technique to predict radiological progression with an application of idiopathic pulmonary fibrosis using chest CT images

      2021, Contemporary Clinical Trials
      Citation Excerpt :

      There are parameters in the algorithm and they include δ, λ, and τ, which are, respectively, for noise, residual, and the step size of the gradient descent (fixed point) algorithm. We refer technical details and further notational definitions to [32] and highlight the key steps in the algorithm as follows: Initialization: Set u0 = 0

    • Prediction of progression in idiopathic pulmonary fibrosis using CT scans at baseline: A quantum particle swarm optimization - Random forest approach

      2019, Artificial Intelligence in Medicine
      Citation Excerpt :

      According to the guidelines, a HRCT scan is required for diagnosis of IPF. Quantitative image analyses (QIA) using texture-based features from HRCT scans have been utilized intensively in pulmonary related diseases [5–7]. For example, QIA are used for robust classification of interstitial lung disease patterns.

    View all citing articles on Scopus
    1

    Support provided by Public Health Service grants from the National Heart Lung and Blood Institute, National Institute of Arthritis and Musculoskeletal and Skin Diseases, National Center for Research Resources of the National Institutes of Health, Bethesda, MD, and Grants U01 HL60587-01A1 and R01 HL072424 from the National Institutes of Health. Cyclophosphamide (Cytoxan®) was supplied by Bristol-Myers Squibb, Princeton, NJ.

    View full text