Image segmentation using CUDA accelerated non-local means denoising and bias correction embedded fuzzy c-means (BCEFCM)

Feng, Chaolu; Zhao, Dazhe; Huang, Min

doi:10.1016/j.sigpro.2015.12.007

Cited by 51 publications

(22 citation statements)

References 40 publications

Supporting

Mentioning

Contrasting

Order By: Relevance

“…As reported in [9], it costs 6 hours for traditional NLMD to remove noises in its 3dimensional version from images in resolution of 181×217×181 with parameters 5  M and 1  d . To improve the time performance of the proposed method, we accelerate it using CUDA in the implementation with a similar strategy exploited in [10] and obtain 150 times speedup.…”

Section: Discussionmentioning

confidence: 99%

Temporally Regularized Non-local De-noising and Its Application to Cardiac Longitudinal MR Images

Feng¹,

Lu²,

Zhao³

2018

Proceedings of the 2018 Symposium on Health and Education (SOHE 2018)

Self Cite

View full text Add to dashboard Cite

Longitudinal image analysis has become a hot research topic due to the progress of medical image processing and analysis. However, noises generally exist in most of medical images, which will negatively affect statistical characteristics of the image intensities and therefore weaken the contrast between different organs to further result in difficulties to image processing methods. In this paper, traditional local de-noising methods are first reviewed with their disadvantages in introducing new artifacts being revealed. On the contrary, non-local de-noising eliminates image noises while intact image structure information is preserved. Therefore, an improved non-local method is then proposed by incorporating temporal information of longitudinal images into the formula of non-local de-noising. Finally, the proposed method is applied to eliminate noises from cardiac longitudinal MR images. The proposed method searches similar pairs in the whole image space rather than in local areas, which is particularly different from traditional methods where only neighborhood intensities are used. The similarities are considered as weightings which are used to estimate the intensity by weighted average of all similar pairs. Experimental results show that the proposed method can effectively eliminate noises from cardiac longitudinal images without weakening boundaries of the images. To improve time performance, the proposed method is accelerated using CUDA and 150× improvement is obtained. 88

show abstract

Section: Discussionmentioning

confidence: 99%

Temporally Regularized Non-local De-noising and Its Application to Cardiac Longitudinal MR Images

Feng¹,

Lu²,

Zhao³

2018

Proceedings of the 2018 Symposium on Health and Education (SOHE 2018)

Self Cite

View full text Add to dashboard Cite

show abstract

“…In [31], a set of adaptive local priors is extracted from a set of training patches by using the local Markov random fields to model the local variations of appearance and shape. Also, a fuzzy c-means segmentation algorithm (in a CUDA accelerated version) enables to deal with local image artifacts [32]. Finally, [20] enhances the spatial priors required for the EMbased segmentation approach by applying a patch search algorithm.…”

Section: Related Workmentioning

confidence: 99%

Multi-atlas label fusion by using supervised local weighting for brain image segmentation

Cárdenas-Peña

Fernández

Ferrández-Vicente

et al. 2017

TecnoL.

View full text Add to dashboard Cite

The automatic segmentation of interest structures is devoted to the morphological analysis of brain magnetic resonance imaging volumes. It demands significant efforts due to its complicated shapes and since it lacks contrast between tissues and intersubject anatomical variability. One aspect that reduces the accuracy of the multi-atlasbased segmentation is the label fusion assumption of one-to-one correspondences between targets and atlas voxels. To improve the performance of brain image segmentation, label fusion approaches include spatial and intensity information by using voxel-wise weighted voting strategies. Although the weights are assessed for a predefined atlas set, they are not very efficient for labeling intricate structures since most tissue shapes are not uniformly distributed in the images. This paper proposes a methodology of voxel-wise feature extraction based on the linear combination of patch intensities. As far as we are concerned, this is the first attempt to locally learn the features by maximizing the centered kernel alignment function. Our methodology aims to build discriminative representations, deal with complex structures, and reduce the image artifacts. The result is an enhanced patch-based segmentation of brain images. For validation, the proposed brain image segmentation approach is compared against Bayesian-based and patch-wise label fusion on three different brain image datasets. In terms of the determined Dice similarity index, our proposal shows the highest segmentation accuracy (90.3% on average); it presents sufficient artifact robustness, and provides suitable repeatability of the segmentation results. KeywordsBrain image segmentation, label fusion, multi-atlas segmentation. ResumenLa segmentación automática de estructuras de interés en imágenes de resonancia magnética cerebral requiere esfuerzos significantes, debido a las formas complicadas, el bajo contraste y la variabilidad anatómica. Un aspecto que reduce el desempeño de la segmentación basada en múltiples atlas es la suposición de correspondencias uno-a-uno entre los voxeles objetivo y los del atlas. Para mejorar el desempeño de la segmentación, las metodologías de fusión de etiquetas incluyen información espacial y de intensidad a través de estrategias de votación ponderada a nivel de voxel. Aunque los pesos se calculan para un conjunto de atlas predefinido, estos no son muy eficientes en etiquetar estructuras intrincadas, ya que la mayoría de las formas de los tejidos no se distribuyen uniformemente en las imágenes. Este artículo propone una metodología de extracción de características a nivel de voxel basado en la combinación lineal de las intensidades de un parche. Hasta el momento, este es el primer intento de extraer características locales maximizando la función de alineamiento de kernel centralizado, buscando construir representaciones discriminativas, superar la complejidad de las estructuras, y reducir la influencia de los artefactos. Para validar los resultados, la estrategia de segmentación propuesta ...

show abstract

“…Graphic processing units (GPUs) that were originally created for rendering graphics, has emerged in the last decade as co-processing units for Central Processing Units (CPU) and has become popular for General Purpose Graphical Processing Units (GP-GPU) used for high performance computing, to accelerate various digital signal processing applications, including medical image processing [9][10][11][12].…”

Section: Introductionmentioning

confidence: 99%

CPU and GPU behaviour modelling versus sequential and parallel bias field correction fuzzy C-means algorithm implementations

Cherradi¹,

Ali²,

Abbassi³

et al. 2017

ces

View full text Add to dashboard Cite

The correction of images corrupted by bias field artefact is still challenging task both at accuracy level as on the computational plane. The work in this paper focus on the second constraint by giving mathematical models of experimental execution time per iteration ETPI(s) on GPU and CPU implementations and speed-ups GPU/CPU(x) of the iterative Bias Field Correction Fuzzy C-means clustering Algorithm (Both sequential BCFCM and parallel PBCFCM versions) against the variable cluster number. In this study we characterize the behaviour of these devices against the proposed implementations to show the optimal situation that offers the best performance. The modelling results of cluster number variation show interesting behaviours and are in good accordance to experimental ones. 424 Bouchaib Cherradi et al.

show abstract

Image segmentation using CUDA accelerated non-local means denoising and bias correction embedded fuzzy c-means (BCEFCM)

Cited by 51 publications

References 40 publications

Temporally Regularized Non-local De-noising and Its Application to Cardiac Longitudinal MR Images

Temporally Regularized Non-local De-noising and Its Application to Cardiac Longitudinal MR Images

Multi-atlas label fusion by using supervised local weighting for brain image segmentation

CPU and GPU behaviour modelling versus sequential and parallel bias field correction fuzzy C-means algorithm implementations

Contact Info

Product

Resources

About