Many algorithms in the field of image processing support high degree of inherent parallelism. Edge detection is one of the most important processes in medical image processing. Edge detection is an independent process to support parallel computation of each pixel intensity changes by their neighbourhood pixels. Magnetic resonance imaging (MRI) scanner provides stack of 2D slices with millions of pixels and thus require much time for edge detection process in central processing unit (CPU) systems. In the proposed work, graphics processing unit (GPU) based parallel edge detection methods are developed for MRI volume using compute unified device architecture (CUDA). Each pixel operation in edge detection is independent and thus GPU provides high level data parallelism using threads per voxel method. Basic edge detection operators such as Roberts, Prewitt, Sobel, Marr- Hildreth and Canny are used in this experiment. The computational time of parallel GPU-CUDA based methods were compared with the serial CPU implementation. Results showed that parallel implementation is about 11× to 98× times faster than the serial CPU implementation.

Edge detection, GPU, CUDA, Roberts, Prewitt; Sobel, Marr- Hildreth, Canny

[1] J.L. Prince, J.M. Links, “Medical Imaging Signals and Systems”, Pearson Prentice Hall, 2nd Edition, 2014.
[2] D. Kirk, W. Hwu, “Programming Massively Parallel Processors - A Hands-on Approach”, 3rd Edition, pp. 1-514, 2016.
[3] T.Kalaiselvi, P. Sriramakrishnan, “Rapid brain tissue segmentation process by modified FCM algorithm with CUDA enabled GPU machine”, International Journal of Imaging System and Technology, pp. 1–12, 2018. DOI: 10.1002/ima.22267
[4] M. Chouchene, F.E. Sayadi, Y. Said, M. Atri, R. Tourki, “Efficient implementation of Sobel edge detection algorithm on CPU, GPU and FPGA”, International Journal of Advanced Media and Communication, Vol. 5, No. 2, pp. 105-117, 2014.

[5] H.X. Gong, L Hao, “Roberts edge detection algorithm based on GPU”, Journal of Chemical and Pharmaceutical Research, Vol. 6, No. 7, pp. 1308-1314, 2014
[6] A. Jain, A. Namdev, M. Chawla, “Parallel Edge Detection By Sobel Algorithm Using CUDA C”, In the proceedings of the International Conference on Electrical, Electronics and Computer Science, India, pp. 1-6, 2016.
[7] M. Kossain, M.A. Ashique, M.A. Ibtehaz, J. Uddin, “Parallel Edge Detection using Sobel Algorithm with Contract-time Anytime Algorithm in CUDA”, Proceedings of the International Conference on Parallel and Distributed Processing Techniques and Applications (PDPTA), CSREA Press, pp. 130-135, 2016
[8] K. Ogawa, Y. Ito, K. Nakano, “Efficient Canny edge detection using a GPU”, In the proceedings of First International Conference on Networking and Computing (ICNC), Japan, pp. 279-280, 2010.
[9] T.L.B. Cheikh, G. Beltrame, G. Nicolescu, F. Cheriet, S. Tahar, “Parallelization strategies of the canny edge detector for multi-core CPUs and many-core GPUs”, In the Proceedings of the International on New Circuits and Systems Conference (NEWCAS), Canada, pp. 49-52, 2012.
[10] Z. Emrani, S. Bateni, H. Rabbani, “A New Parallel Approach for Accelerating the GPU-Based Execution of Edge Detection Algorithms”, Journal of medical signals and sensors, Vol. 7, No. 1, pp. 33-42, 2017.
[11] T. Kalaiselvi, P. Sriramakrishnan, K. Somasundaram, “Survey of using GPU CUDA programming model in medical image analysis”, Informatics in Medicine Unlocked, Vol. 9, pp. 133 – 144, 2017
[12] F. Rob, CUDA application design and development, 1st Ed. Elsevier, pp. 1–336, 2011.
[13] L. Roberts, J. Tippet, “Machine Perception of Three Dimensional Solids”, Optical and Electro-Optical Information Processing, Cambridge, MA: IT Press; 1965.
[14] J.M. Prewitt, “Object Enhancement and Extraction”, Vol. 75, New York, Academic Press 1970.
[15] D. Marr, E. Hildreth, “Theory of Edge Detection”, Proceedings of the Royal Society of London, Series B, Biological Sciences, Vol. 207, No. 1167, pp. 187–217, 1980.
[16] J.F. Canny, “A computation approach to edge detection”, IEEE Transactions on Pattern Analysis and Machine Intelligence, Vol. 8, No. 6, pp. 769-798, 1986.
[17] S. Cuomo, P.D. Michele, F. Piccialli, “3D data denoising via nonlocal means filter by using parallel GPU strategies”, Computational Mathematical Methods in Medicine, pp. 1–14, 2014.
[18] V. Podlozhnyuk, “Image convolution with CUDA”, 2007.