Related Experiment Videos
Breast volume denoising and noise characterization by 3D wavelet transform.
1University of Rochester, Department of Radiology, 601 Elmwood Avenue, Rochester, NY 14642, USA.
Summary
This study introduces a novel volumetric denoising method for 3D breast imaging using a separable 3D wavelet transform (WT). The technique effectively reduces noise in digital breast volumes, enhancing subsequent analysis.
Area of Science:
- Medical Imaging
- Signal Processing
- Computational Biology
Background:
- Cone-beam computed tomography (CBCT) generates digital breast volumes for 3D tissue analysis.
- Data denoising is a critical preprocessing step for accurate volumetric breast segmentation.
- Existing denoising methods may not fully address the complexities of 3D breast data.
Purpose of the Study:
- To develop and evaluate a novel volumetric denoising technique for 3D breast imaging.
- To improve the quality of digital breast volumes obtained from CBCT.
- To facilitate more accurate segmentation and analysis of breast tissues.
Main Methods:
- A separable 3D wavelet transform (WT) scheme combining 2D WT and 1D WT was employed.
- The method involves wavelet decomposition, attenuation of high-pass subbands, and wavelet synthesis.
- Multilevel WT provides a multiresolution representation, with noise primarily in high-pass subbands.
Main Results:
- The proposed technique effectively reduces noise and irregularities in 3D breast volumes.
- Wavelet transform characterizes subband information using energy, variance, and entropy.
- 3D visualization aids in perceiving the spatial structure within subbands.
Conclusions:
- The separable 3D WT denoising technique is effective for 3D breast imaging data.
- This method enhances the quality of digital breast volumes for improved analysis.
- The approach shows promise for applications in breast cancer detection and diagnosis.