Yinghua Tao BEng, Presenter: Nothing to Disclose

Kari A. Pulfer, Abstract Co-Author: Nothing to Disclose

Howard A. Rowley MD, Abstract Co-Author: Research support, General Electric Company Speaker, Bracco Group Consultant, Eli Lilly and Company Consultant, W.L. Gore & Associates, Inc Consultant, H. Lundbeck A/S

Guang-Hong Chen PhD, Abstract Co-Author: Nothing to Disclose

A single bolus of intravenous contrast used to produce CT perfusion data also has the potential to simultaneously produce time-resolved CT angiograms, provided that sufficient quality can be achieved. Here we employ a novel wavelet denoising technique to obtain high signal to noise ratio, high spatial resolution, four-dimensional CT angiography images from standard clinical CT perfusion data sets.

CT DICOM image data sets from 10 human subjects were retrospectively studied under an approved IRB protocol. All subjects underwent cerebral CT perfusion scans on a 64-slice VCT scanner (GE Healthcare) using the shuttle acquisition mode. 200mAs and 80kVp were used. The DICOM images were first lowpass filtered to reduce noise, and difference images were calculated between the original and lowpass filtered images. A wavelet denoising algorithm was applied to the difference image to reduce the noise level while retaining the sharp edges in the difference image. Then, the wavelet denoised difference images were added back to the lowpass filtered image to obtain the final low noise and high spatial resolution CT images. This wavelet based denoising technique was used for all axial slices (2D method) and/or axial plus sagittal slices (2D-2D method). ROI’s were applied to one slice at all time frames per patient. Mean values and standard deviation in ROIs were compared for both the original clinical image and SNR enhanced images.

For all patients, noise (standard deviation) reduction was 49% ±3% for 2D denoised data sets, and 64% ±5% for 2D-2D denoised data sets, which corresponds to dose saving of 75% for 2D denoised data sets, and 87% for 2D-2D denoised data sets. Mean values in ROIs were within 2% difference between clinical data sets and denoised data sets.

Factor of 2 and factor of 2.6 noise reduction has been achieved using the novel wavelet denoising technique. This noise reduction method enables dose saving of 87% when it is implemented in the 2D-2D method.

After SNR enhancement, both 4D-CTA and CT perfusion maps were constructed from the same data, providing clinically useful data at very low radiation dose level.

Tao, Y, Pulfer, K, Rowley, H, Chen, G, SNR Enhancement in Time-resolved CT Angiography Using a Novel Wavelet Denoising Technique.  Radiological Society of North America 2010 Scientific Assembly and Annual Meeting, November 28 - December 3, 2010 ,Chicago IL. http://archive.rsna.org/2010/9011954.html