MENU

Click here to

×

Are you sure ?

Yes, do it No, cancel

Adaptive Multi-Energy Local Mean Filtering for Photon-Counting-Detector Based Multi-Energy CT

S Tao*, K Rajendran , A Ferrero , W Zhou , C McCollough , S Leng , Mayo Clinic, Rochester, MN

Presentations

(Thursday, 8/2/2018) 7:30 AM - 9:30 AM

Room: Room 202

Purpose: To develop a multi-energy non-local mean (MENLM) filtering technique with adaptive filtering strength to maximize noise reduction in all anatomic regions while preserving structural details.

Methods: Under an IRB approved protocol, in vivo chest CT images were acquired on a research whole-body photon-counting-detector CT scanner (140kV, ultra-high-resolution mode, energy threshold=25/65keV, 126 mAs, pitch=0.8, Q65 kernel, FOV=275mm, 1024x1024 matrix, 1.5mm slice). The images were filtered using a MENLM filtering technique to reduce noise. The denoising performance of MENLM is mainly determined by the filtering strength. In applications such as chest CT, a single filtering strength may not be suitable for different anatomies of interests. Visualization of lung structures requires a lower filtering strength, which is not sufficient for denoising soft tissue. Here, we demonstrate an adaptive filtering strength strategy for MENLM, which combines images generated from different filtering strengths into a single image to achieve tissue-dependent filtering. The proposed strategy is motivated by the observation that different anatomies of interests, e.g., lung and soft tissue, occupy different regions in image histogram. MENLM with single filtering strengths of h=1.0/1.5 (optimized for lung and soft tissue), and the proposed adaptive strength strategy were applied to the chest CT images and results were compared.

Results: The images generated using a low filtering strength (h=1.0) better preserved small lung structures but did not sufficiently denoise soft tissue. A higher filtering strength (h=1.5) more effectively denoised the soft tissue but also caused loss of small lung structure. The proposed adaptive filtering strategy preserved lung structure while sufficiently denoising soft tissue. Quantitative noise measurements and line profiles supported these observations.

Conclusion: An adaptive filtering strength strategy is developed for MENLM which merges images generated using different filtering strengths into a single image. The proposed method can preserve lung structures while sufficiently denoising soft tissue regions.

Funding Support, Disclosures, and Conflict of Interest: CHM receives grand support from Siemens Healthcare.

Keywords

Not Applicable / None Entered.

Taxonomy

Not Applicable / None Entered.

Contact Email