Enhancing Lung CT Image Quality through Frequency Domain Artifact Suppression

Abstract

Computed Tomography (CT) imaging plays a crucial role in the diagnosis and monitoring of lung-related diseases. However, the presence of high-density materials, such as metallic implants or surgical clips, often introduces artifacts that degrade image quality and hinder accurate clinical interpretation. This study presents a frequency domain filtering approach to suppress such reconstruction artifacts in lung CT images, thereby improving diagnostic clarity. The proposed method begins with preprocessing of CT slices, followed by transformation into the frequency domain using the Fast Fourier Transform (FFT). Specific frequency components responsible for artifacts are selectively attenuated using tailored notch and band-stop filters. This process effectively minimizes the streaking and blurring effects commonly observed around metallic objects. The filtering algorithm is first developed and tested in MATLAB, where its performance is evaluated using metrics such as Peak Signal-to-Noise Ratio (PSNR) and Structural Similarity Index (SSIM). The results show a notable improvement in visual quality and structural preservation of the lung tissues. To explore the feasibility of hardware acceleration for real-time applications, the same filtering algorithm is implemented in Verilog and simulated on an FPGA platform. The hardware implementation demonstrates low-latency performance with efficient resource utilization, making it suitable for integration in embedded medical imaging systems. This combined software-hardware approach underscores the potential of frequency domain filtering as a practical solution for metal artifact reduction in CT images and opens avenues for real-time deployment in clinical imaging workflows..

Introduction

1. Background and Motivation

Computed Tomography (CT) is essential for diagnosing lung diseases such as cancer, infections, and COPD. However, metallic implants like dental fillings or surgical clips can cause artifacts in CT images—most notably streaking and shading—which degrade image quality and hinder accurate diagnosis. Traditional methods for metal artifact reduction (MAR) include interpolation, iterative reconstruction, and dual-energy CT, but these approaches often require raw data, added radiation, or are computationally intensive.

2. Proposed Solution

This study presents a frequency domain filtering approach to suppress metal artifacts in lung CT images:

• CT images are transformed into the frequency domain using the Fast Fourier Transform (FFT).

• Artifact frequencies, which appear as high-frequency noise or directional streaks, are isolated and suppressed using custom filter masks.

• The filtered image is then reconstructed using the inverse FFT.

• The entire process is implemented in MATLAB for algorithm development and Verilog for hardware acceleration via FPGA, enabling real-time processing.

3. Algorithm Overview

The algorithm consists of several key steps:

1. Image Preprocessing: Converts color images to grayscale for consistency.

2. Artifact Detection: Uses histogram-based thresholding (Otsu’s method) to identify metal regions.

3. Mask Refinement: Applies morphological operations to clean and fill the artifact mask.

4. Image Inpainting: Fills in detected artifact areas using region-based reconstruction from nearby pixels.

5. Edge-Preserving Smoothing: Applies a bilateral filter to preserve edges while reducing residual noise.

6. Blending and Reconstruction: Replaces the artifact regions in the original image with inpainted, filtered regions for visual consistency.

4. Results and Validation

• The algorithm was tested on real lung CT images using MATLAB.

• Metrics such as PSNR and SSIM showed improvements in image quality.

• Artifacts were significantly reduced while maintaining anatomical accuracy.

• The filtered frequency domain plots and visual outputs confirmed effective artifact suppression.

5. Hardware Implementation

• The algorithm was translated into Verilog and synthesized on an FPGA.

• FPGA-based acceleration offers:

  • Low latency

  • Parallel processing

  • Real-time performance

  • Energy efficiency

• The RTL schematic and hardware simulation confirmed the feasibility of real-time deployment in clinical systems.

Conclusion

This study presents a practical approach for reducing metal artefacts in lung CT images using frequency domain filtering, supported by both software and hardware implementations. By identifying and suppressing high-frequency components responsible for artefacts, the method significantly improves image clarity while preserving important anatomical features. The use of a low-pass mask in the Fourier domain effectively minimizes streaking caused by metallic implants, leading to visually enhanced outputs. The algorithm, developed in MATLAB, demonstrates consistent performance across varied input images. To enable real-time processing, the filtering logic was also implemented in Verilog and synthesized on FPGA hardware, highlighting its potential for clinical integration in embedded systems. The combined use of signal processing and hardware acceleration makes the proposed method both efficient and adaptable. Overall, the approach offers a reliable post-processing solution for improving the diagnostic quality of CT images, especially in scenarios where access to raw projection data is limited.

References

[1] X. Huang, J. Wang, F. Tang, T. Zhong, and Y. Zhang, “Metal artifact reduction on cervical CT images by deep residual learning,” Biomedical Engineering Online, vol. 17, no. 1, pp.?1–15, Nov.?2018. [2] H. Liao, W\\.-A. Lin, S.-K. Zhou, and J. Luo, “ADN: Artifact disentanglement network for unsupervised metal artifact reduction,” IEEE Transactions on Medical Imaging, vol. 39, no. 2, pp.?634–643, Feb.?2019. [3] W\\.-A. Lin, H. Liao, C. Peng, X. Sun, J. Zhang, J. Luo, R. Chellappa, and S.-K. Zhou, “DuDoNet: Dual domain network for CT metal artifact reduction,” in Proc. IEEE/CVF Conf. Computer Vision and Pattern Recognition, Long Beach, CA, Jun.?2019, pp.?10512–10521. [4] Y. Nakao, K. Imanishi, N. Ueda, Y. Imai, T. Kirita, and T. Matsuda, “Regularized three-dimensional generative adversarial nets for unsupervised metal artifact reduction in head and neck CT images,” IEEE Access, vol.?8, pp.?109453–109465, 2020. [5] N. J. Lee, J. Gu, and J. C. Ye, “Unsupervised CT metal artifact learning using attention-guided ??CycleGAN,” IEEE Transactions on Medical Imaging, vol. 40, no.?12, pp.?3932–3944, Dec.?2021. [6] Y. Zhang and H. Yu, “Convolutional neural network based metal artifact reduction in X?ray computed tomography,” IEEE Transactions on Medical Imaging, vol. 37, no. 6, pp.?1370–1381, Jun.?2018. [7] M. U. Ghani and W. Clem Karl, “Fast enhanced CT metal artifact reduction using data domain deep learning,” arXiv, Apr.?2019. [8] Y. Su, J. Wang, Y. Li, K. Shang, and D. Liang, “RetinexFlow for CT metal artifact reduction,” arXiv, Jun.?2023. [9] H. Wang, Y. Li, H. Zhang, J. Chen, K. Ma, D. Meng, and Y. Zheng, “InDuDoNet: An interpretable dual domain network for CT metal artifact reduction,” arXiv, Sep.?2021. [10] Y. Yu, Z. Zhang, X. Li, H. Ren, W. Zhao, and L. Xing, “Metal artifact reduction in 2D CT images with self?supervised cross?domain learning,” arXiv, Sep.?2021. [11] Y. Rodríguez-Gallo, R. Orozco-Morales, and M. Pérez-Díaz, “Metal artifact reduction by morphological image filtering for computed tomography,” in World Congress on Medical Physics and Biomedical Engineering 2018, L. Lhotska et al., Eds., Springer, Singapore, 2019, pp.?219–222. [12] H. Lell, E. Meyer, M. Schmid et al., “Frequency split metal artefact reduction in pelvic computed tomography,” European Radiology, vol.?23, pp.?2137–2145, 2013. (Cited by multiple recent studies.) [13] R. Wellenberg, E. Hakvoort, C. Slump, M. Boomsma, M. Maas, and G. Streekstra, “Metal artifact reduction techniques in musculoskeletal CT-imaging: phantom and clinical study,” European Journal of Radiology, vol.?107, pp.?60–69, 2018. [14] C.-B. Schönlieb, “Inpainting-filtering for metal artifact reduction (IMIF?MAR) in computed tomography,” Physical and Engineering Sciences in Medicine, vol.?44, pp.?55–66, 2021. [15] N. Nagornov, P. Lyakhov, M. Valueva, and M. Bergerman, “RNS?based FPGA accelerators for high?quality 3D medical image wavelet processing using scaled filter coefficients,” IEEE Access, vol.?10, pp.?2022–2036, 2022..

Copyright

Copyright © 2025 Humera Kouser K, Dr. Sunil T D, Dr. M.C. Chandrashekhar . This is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.