Post-processing multiple-frame super-resolution in ultrasound imaging

Morin, Renaud; Basarab, Adrian; Ploquin, Marie; Kouamé, Denis

doi:10.1117/12.910711

Cited by 7 publications

(6 citation statements)

References 20 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…We summarize here the SR studies implemented in the postprocessing phase of USI since the pre-processing phase of USI is out of the scope of this paper. A multi-frame SR task was experimented in [39]. The authors mainly focused on the registration step to get a better ultrasound image quality.…”

Section: B Super Resolution In Ultrasound Imagingmentioning

confidence: 99%

Super Resolution of B-Mode Ultrasound Images With Deep Learning

Temiz

Bilge

2020

IEEE Access

View full text Add to dashboard Cite

Ultrasound offers a safe, non-invasive, and inexpensive way of imaging. However, due to its natural intrinsic imaging characteristics, it produces poor quality images with low resolution (LR) compared to other medical imaging modalities. Various image enhancement techniques have been extensively studied to overcome these shortcomings. Super-resolution (SR) is one of these methods, which endeavor to obtain high resolution (HR) images from LR images while enlarging them. Numerous studies have already utilized different SR techniques in various stages of ultrasound imaging (USI). Unlike other studies, which aimed at obtaining SR in the pre-processing phase or early stages of the post-processing phase of USI, we achieved SR on B-mode ultrasound images, which is the last stage of USI. We constructed a deep convolutional neural network (CNN) and trained it with a very large dataset of B-mode ultrasound images for the scale factors 2, 3, 4, and 8. We evaluated the performance of our proposed model quantitatively with eight image quality measures. The quantitative results revealed that our algorithm is much more successful than other methods at each magnification factor. Furthermore, we also verified that there is a statistically significant difference between our approach and others. Besides, qualitative analysis of the reconstructed images also confirms that it produces much better quality HR images than other methods in terms of the human visual system. INDEX TERMS Ultrasound, super-resolution, deep learning, convolutional neural network.

show abstract

Section: B Super Resolution In Ultrasound Imagingmentioning

confidence: 99%

Super Resolution of B-Mode Ultrasound Images With Deep Learning

Temiz

Bilge

2020

IEEE Access

View full text Add to dashboard Cite

show abstract

“…We have recently shown in [11] that classical motion estimation, usually involved in multiframe SR frameworks (see [10] and references therein), provide poor results when performed using US images. Indeed, these methods fail to estimate the true elastic motion and therefore break the speckle characteristics, resulting in an image degradation [11]. The main issue is that the motion estimation must be performed on the LR images with sub-pixel accuracy in order to provide improvements in the resulting HR image.…”

Section: Image Sequence Modelmentioning

confidence: 99%

“…The CNR was successively computed using one LR input image and the three HR images based on bicubic interpolation, classic SR as reported in [11] and our method (see the corresponding images in Fig. 6).…”

Section: In-vitro Phantom Imagesmentioning

confidence: 99%

“…The desired high resolution (HR) was obtained by using the relative sub-pixel motion between LR images and has been greatly improved [10] during the past decades. The straightforward implementation of such algorithms for US imaging was unsuccessful, due to the intrinsic nature of tissue elastic motions, speckle and the point spread function (PSF) [11]. Note that unlike strain compounding [12][13][14], which consists in averaging several ultrasound images acquired under different strain conditions in order to reduce the speckle, our goal is to improve the resolution of US images without altering their intrinsic characteristics.…”

Section: Introductionmentioning

confidence: 99%

See 1 more Smart Citation

Motion estimation-based image enhancement in ultrasound imaging

et al. 2015

Self Cite

View full text Add to dashboard Cite

OATAO is an open access repository that collects the work of Toulouse researchers and makes it freely available over the web where possible. This is an author-deposited version published in : http://oatao.univ-toulouse.fr/ Eprints ID : 14818 High resolution medical ultrasound (US) imaging is an ongoing challenge in many diagnosis applications and can be achieved either by instrumentation or by post-processing. Though many works have considered the issue of resolution enhancement in optical imaging, very few works have investigated this issue in US imaging. In optics, several algorithms have been proposed to achieve super-resolution (SR) image reconstruction, which consists of merging several low resolution images to create a higher resolution image. However, the straightforward implementation of such techniques for US imaging is unsuccessful, due to the interaction of ultrasound with tissue and speckle. We show how to overcome the limit of SR in this framework by refining the registration part of common multiframe techniques. For this purpose, we investigate motion estimation methods adapted to US imaging. Performance of the proposed technique is evaluated on both realistic simulated US images (providing an estimated best-case performance) and real US sequences of phantom and in-vivo thyroid images. Compared to classical SR methods, our technique brings both quantitative and qualitative improvements. Resolution gain was found to be 1.41 for the phantom sequence and 1.12 for the thyroid sequence and a quantitative study using the phantom further confirmed the spatial resolution enhancement. Furthermore, the contrast-to-noise ratio was increased by 27% and 13% for simulated and experimental US images, respectively.

show abstract

“…Lately, deconvolution techniques have been extensively studied in the general image domain to perform the estimation of a high resolution (HR) image. Early attempts to perform this task include total variation (TV) deconvolution [5] and complex deconvolution framework [6], but more recent works have focused on multiple frame super-resolution (SR) [7,8], image restoration based on tetrolets shrinkage [9], deconvolution with an inaccurate optical PSF [10], parametric inverse filtering [11,12], and blind deconvolution without SR [13]. Among the previously mentioned methods, blind deconvolution is a category of efficient but challenging approaches since both the HR image and the unknown blur must be estimated from the observed image.…”

Section: Introductionmentioning

confidence: 99%

Semi-blind deconvolution for resolution enhancement in ultrasound imaging

Morin

Bidon

Basarab

et al. 2013

2013 IEEE International Conference on Image Processing

Self Cite

View full text Add to dashboard Cite

In the field of ultrasound imaging, resolution enhancement is an up-to-date challenging task. Many device-based approaches have been proposed to overcome the low resolution nature of ultrasound images but very few works deal with post-processing methods. This paper investigates a novel approach based on semi-blind deconvolution formulation and alternating direction method framework in order to perform the ultrasound image restoration task. The algorithm performance is addressed using optical images and synthetic ultrasound data for a various range of criteria. The results demonstrate that our technique is more robust to uncertainties in the a priori ultrasonic pulse than classical non-blind deconvolution methods.

show abstract

Post-processing multiple-frame super-resolution in ultrasound imaging

Cited by 7 publications

References 20 publications

Super Resolution of B-Mode Ultrasound Images With Deep Learning

Super Resolution of B-Mode Ultrasound Images With Deep Learning

Motion estimation-based image enhancement in ultrasound imaging

Semi-blind deconvolution for resolution enhancement in ultrasound imaging

Contact Info

Product

Resources

About