Residual Semantic Segmentation of the Prostate from Magnetic Resonance Images [chapter]

Md Sazzad Hossain, Andrew P. Paplinski, John M. Betts
2018 Lecture Notes in Computer Science  
The diagnosis and treatment of prostate cancer requires the accurate segmentation of the prostate in Magnetic Resonance Images (MRI). Manual segmentation is currently the most accurate method of performing this task. However, this requires specialist knowledge, and is time consuming. To overcome these limitations, we demonstrate an automatic segmentation of the prostate region in MRI images using a VGG19-based fully convolutional neural network. This new network, VGG19RSeg, identifies a region
more » ... f interest in the image using semantic segmentation, that is, a pixel-wise classification of the content of the input image. Although several studies have applied fully convolutional neural networks to medical image segmentation tasks, our study introduces two new forms of residual connections (remote and neighbouring) which increases the accuracy of segmentation over the basic architecture. Our results, using this new architecture, show that the proposed VGG19RSeg can achieve a mean Dice Similarity Coefficient of 94.57%, making it more accurate than comparable methods reported in the literature.
doi:10.1007/978-3-030-04239-4_46 fatcat:7njh7sp5hnhizbqvlzquypblvu