Medical image segmentation on GPUs – A comprehensive review

Abstract

Segmentation of anatomical structures, from modalities like computed tomography (CT), magnetic resonance imaging (MRI) and ultrasound, is a key enabling technology for medical applications such as diagnostics, planning and guidance. More efficient implementations are necessary, as most segmentation methods are computationally expensive, and the amount of medical imaging data is growing. The increased programmability of graphic processing units (GPUs) in recent years have enabled their use in several areas. GPUs can solve large data parallel problems at a higher speed than the traditional CPU, while being more affordable and energy efficient than distributed systems. Furthermore, using a GPU enables concurrent visualization and interactive segmentation, where the user can help the algorithm to achieve a satisfactory result. This review investigates the use of GPUs to accelerate medical image segmentation methods. A set of criteria for efficient use of GPUs are defined and each segmentation method is rated accordingly. In addition, references to relevant GPU implementations and insight into GPU optimization are provided and discussed. The review concludes that most segmentation methods may benefit from GPU processing due to the methods' data parallel structure and high thread count. However, factors such as synchronization, branch divergence and memory usage can limit the speedup.

Performance of Medical Image Processing Algorithms Implemented in CUDA running on GPU based Machine

“…This article mentioned few GPU based image processing registration and segmentation. Smistad et al, presented a review about medical image segmentation on GPU [13]. They gave some basics of CUDA model and explain various segmentation methods like thresholding, region growing, watershed and active contours on GPU.…”

Rupture risk in abdominal aortic aneurysms: A realistic assessment of the explicit GPU approach

“…Graphics processing unit (GPU) technology is proposed for medical image segmentation (Smistad et al, 2015), or deformable models using spring-mass-damper or tensor-mass methods (Rasmusson et al, 2008) as well as implicit FE (Cecka et al, 2011;Wong et al, 2015). The Total Lagrangian Explicit Dynamic FE algorithm (TLED) (Miller et al, 2007) fits the GPU programming model well given: (1) most element and nodal operations are pleasingly parallel (independent of one another), and (2) memory requirements are very low.…”

Real-time envelope detection of ultrasound radiofrequency signals using OpenCV GPU framework

“…However, envelope detection on CPU can be a time-consuming process. To accelerate the computation, parallel GPUs, especially the GPGPU for general purpose computing on GPU, have been used in many fields including machine learning and medical image processing [21,22].…”