Resolution enhancement of lung 4D-CT data using multiscale interphase iterative nonlocal means.

Area of Science:

• Medical Imaging
• Radiotherapy Physics
• Image Processing

Background:

• Four-dimensional computed tomography (4D-CT) is vital for lung cancer radiotherapy, offering tumor motion data.
• Increased scanning duration in 4D-CT elevates radiation dose, leading to dose reduction at the cost of spatial resolution.
• Reduced resolution causes artifacts and uncertainty in tumor localization, potentially damaging healthy tissues.

Purpose of the Study:

• To develop a novel postprocessing algorithm for enhancing the resolution of lung 4D-CT data.
• To address the trade-off between radiation dose and image resolution in lung cancer radiotherapy.
• To improve the accuracy of tumor localization and minimize damage to surrounding healthy tissues.

Main Methods:

• A patch-based mechanism propagates information across different 4D-CT phases to reconstruct intermediate slices.
• Utilizes structurally matching and spatially nearby patches for patch reconstruction.
• Employs a quad-tree technique for adaptive image partitioning and an iterative strategy for detailed enhancement.

Main Results:

• The algorithm significantly enhances image structures and reduces artifacts in lung 4D-CT data.
• Quantitative analysis shows a 3-4 dB increase in peak signal-to-noise ratio.
• Demonstrates a 3%-5% improvement in structural similarity index compared to standard interpolation methods.

Conclusions:

• A new algorithm effectively improves 4D-CT resolution.
• The developed algorithm surpasses conventional interpolation-based methods.
• Results show markedly improved structural clarity and significantly reduced artifacts in 4D-CT images.