Low dose CT image restoration using a localized patch database

doi:10.1109/nssmic.2013.6829131

Cited by 2 publications

(1 citation statement)

References 9 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…There are, however, two shortcomings with this method: (1) the registration can be errorprone, not only because of the noise artifacts, but also due to the likely presence of high variations between the possible targets and the images stored in the database and (2) the need to store a large number of full-size images can lead to excessive hardware requirements, such as disk space and management. In order to overcome these challenges we recently presented some preliminary work (Ha et al 2013) that replaced the image database with a database of small 7 × 7 patches. This has several advantages: (1) it simplifies the restoration pipeline because it eliminates the need for registration and (2) by storing only patches we can remove a good amount of redundancy and in its place store more diversity.…”

Section: Introductionmentioning

confidence: 99%

Low dose CT image restoration using a database of image patches

Mueller

2015

Phys. Med. Biol.

View full text Add to dashboard Cite

Reducing the radiation dose in CT imaging has become an active research topic and many solutions have been proposed to remove the significant noise and streak artifacts in the reconstructed images. Most of these methods operate within the domain of the image that is subject to restoration. This, however, poses limitations on the extent of filtering possible. We advocate to take into consideration the vast body of external knowledge that exists in the domain of already acquired medical CT images, since after all, this is what radiologists do when they examine these low quality images. We can incorporate this knowledge by creating a database of prior scans, either of the same patient or a diverse corpus of different patients, to assist in the restoration process. Our paper follows up on our previous work that used a database of images. Using images, however, is challenging since it requires tedious and error prone registration and alignment. Our new method eliminates these problems by storing a diverse set of small image patches in conjunction with a localized similarity matching scheme. We also empirically show that it is sufficient to store these patches without anatomical tags since their statistics are sufficiently strong to yield good similarity matches from the database and as a direct effect, produce image restorations of high quality. A final experiment demonstrates that our global database approach can recover image features that are difficult to preserve with conventional denoising approaches.

show abstract

Section: Introductionmentioning

confidence: 99%

Low dose CT image restoration using a database of image patches

Mueller

2015

Phys. Med. Biol.

View full text Add to dashboard Cite

show abstract

Non-sample fuzzy based convolutional neural network model for noise artifact in biomedical images

Byeon,

Patel,

Vidhate

et al. 2024

Discov Appl Sci

View full text Add to dashboard Cite

The use of a light-weight deep learning Convolutional Neural Network (CNN) augmented with the power of Fuzzy Non-Sample Shearlet Transformation (FNSST) has successfully solved the problem of reducing noise and artifacts in Low-Dose Computed Tomography (LDCT) pictures. Both the Normal-Dose Computed Tomography (NDCT) and the Low-Dose Computed Tomography (LDCT) images from the dataset are subjected to the FNSST decomposition procedure during the training phase, producing high-frequency sub-images that act as input for the CNN. The CNN creates a meaningful connection between the high-frequency sub-images from LDCT and their corresponding residual sub-images during the training operation. The CNN is given the capacity to distinguish between LDCT high-frequency sub-images and expected high-frequency sub-images, which frequently have varying levels of noise or artifacts, especially in a fuzzy setting. The FNSST-CNN then successfully distinguishes LDCT high-frequency sub-images from the expected high-frequency sub-images during the testing phase, thereby reducing noise and artifacts. When compared to other approaches like KSVD, BM3D, and conventional image domain CNNs, the performance of FNSST-CNN is impressive as shown by better peak signal-to-noise ratios, stronger structural similarity, and a closer likeness to NDCT pictures.

show abstract

Low dose CT image restoration using a localized patch database

Cited by 2 publications

References 9 publications

Low dose CT image restoration using a database of image patches

Low dose CT image restoration using a database of image patches

Non-sample fuzzy based convolutional neural network model for noise artifact in biomedical images

Contact Info

Product

Resources

About