Computed tomography perfusion imaging denoising using Gaussian process regression

Abstract

Brain perfusion weighted images acquired using dynamic contrast studies have an important clinical role in acute stroke diagnosis and treatment decisions. However, computed tomography (CT) images suffer from low contrast-to-noise ratios (CNR) as a consequence of the limitation of the exposure to radiation of the patient. As a consequence, the developments of methods for improving the CNR are valuable. The majority of existing approaches for denoising CT images are optimized for 3D (spatial) information, including spatial decimation (spatially weighted mean filters) and techniques based on wavelet and curvelet transforms. However, perfusion imaging data is 4D as it also contains temporal information. Our approach using gaussian process regression (GPR), which takes advantage of the temporal information, to reduce the noise level. Over the entire image, GPR gains a 99% CNR improvement over the raw images and also improves the quality of haemodynamic maps allowing a better identification of edges and detailed information. At the level of individual voxel, GPR provides a stable baseline, helps us to identify key parameters from tissue time-concentration curves and reduces the oscillations in the curve. GPR is superior to the comparable techniques used in this study.

Wavelet analysis of cardiac optical mapping data

“…Wavelet filtering also retains the spatial features in regions with localized important changes. In this respect, wavelet denoising is generally superior to Gaussian smoothing because the wavelet domain filter adapts "automatically" (because of the thresholding algorithm) to the noise distribution of the input image, while the Gaussian smoothing kernel must match the object size to be detected [61,[63][64][65].…”

Feasibility of Single-Input Tracer Kinetic Modeling with Continuous-Time Formalism in Liver 4-Phase Dynamic Contrast-Enhanced CT

“…To enhance contrast-to-noise ratios, DCE-CT images were denoised by use of multiple observations Gaussian process regression [20]. To reduce movement-induced artifacts, we coregistered each set of dynamic images with the portal-phase image as a template by using the Insight Segmentation and Registration Toolkit [21].…”

Phase-preserving approach in denoising computed tomography medical images

“…It is used to determine many types of human health issues such as brain stroke, intracranial haemorrhage, trauma, skull fracture (Bhadauria and Dewal 2013) and breast cancer (Chen and Ning 2004). Moreover, CT scan images are the most used modality for acute stroke patients Zhu et al 2012). CT scans are widely used due to various reasons, such as lower prices (Attivissimo et al 2010), less scanning time, wide availability, simplicity of access, perfect recognition of calcification, outstanding detection of bony details (Bhadauria and Dewal 2012), shorter imaging time (Bhadauria and Dewal 2012), better clarity (Sharma and Jindal 2011), and it is used for patients who are unable to remain stationary during the scanning procedure (Bhadauria and Dewal 2012).…”

“…Therefore, lots of studies have proven that the type of noise affecting CT images is the additive white Gaussian noise (Gravel et al 2004;Attivissimo et al 2010;Sun et al 2013). Numerous issues contributed to the existence of noise with CT images such as image acquisition modes, transmission, storage and display devices (Trinh et al 2012), errors in photon counting statistics (Chen et al 2011;Rahim et al 2012), imaging hardware problems, finite exposure time (Pham 2012), missing data through the image acquisition procedure or even problems in sensors and detectors of the CT imaging system, absorption of fewer amounts of x-ray photons (Chen et al 2011) and limitations of the exposure to radiation on the patient (Zhu et al 2012). Currently, certain types of CT imaging systems possess the capability to achieve isotropic acquisition of the whole chest with submillimetre resolution within a single breath hold.…”

“…To reduce the health risk from exposure to radiation while making a CT scan, it is desirable to use a radiation dose as low as possible because the radiation increases the danger of inducing cancer (Zheng et al 2013). Therefore, CT scanning balances the trade-off between image quality and the amount of radiation exposure on patients (Zhu et al 2012). As a result, reducing the radiation dose would increase the noise in the obtained images (Kozaitis et al 2011).…”