Denoised and texture enhanced MVCT to improve soft tissue conspicuity

Area of Science:

• Medical Imaging
• Radiation Oncology
• Image Processing

Background:

• Megavoltage CT (MVCT) is used in TomoTherapy for patient alignment but has limited soft tissue visualization due to noise and low contrast.
• Existing methods like algebraic reconstruction and local denoising have not sufficiently improved soft tissue conspicuity.

Purpose of the Study:

• To develop and evaluate a novel method (DeTECT) for enhancing soft tissue information in MVCT images.
• DeTECT utilizes nonlocal means denoising and texture enhancement to improve soft tissue delineation for adaptive radiation therapy.

Main Methods:

• A Block Matching 3D (BM3D) algorithm was adapted for noise reduction while preserving texture in MVCT images.
• A saliency map was generated to further enhance low-contrast structures.
• The DeTECT method (BM3D + saliency map) was applied to phantoms, a head and neck patient, and prostate cancer patients.

Main Results:

• Postprocessed MVCT images using DeTECT showed significant improvements in contrast without loss of resolution.
• Previously inconspicuous structures like lymph nodes and veins became visible in head and neck images.
• Boundaries between organs (e.g., bladder and prostate) were clarified, and contrast-to-noise ratios (CNRs) were substantially increased across all tested cases.
• DeTECT outperformed local denoising methods like anisotropic diffusion.

Conclusions:

• The DeTECT method effectively extracts enhanced soft tissue contrast information from noisy MVCT images.
• Combining nonlocal means 3D block matching with saliency maps makes previously imperceptible soft tissue details visible to observers.
• This technique holds promise for improving patient alignment, delineation, and adaptive radiation therapy planning.