Davood Karimi1, Golnoosh Samei2, Claudia Kesch3, Guy Nir4, Septimiu E Salcudean2. 1. Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada. karimi@ece.ubc.ca. 2. Department of Electrical and Computer Engineering, University of British Columbia, Vancouver, BC, Canada. 3. British Columbia Cancer Agency, Vancouver, BC, Canada. 4. Department of Urologic Sciences, University of British Columbia, Vancouver, BC, Canada.
Abstract
PURPOSE: Most of the existing convolutional neural network (CNN)-based medical image segmentation methods are based on methods that have originally been developed for segmentation of natural images. Therefore, they largely ignore the differences between the two domains, such as the smaller degree of variability in the shape and appearance of the target volume and the smaller amounts of training data in medical applications. We propose a CNN-based method for prostate segmentation in MRI that employs statistical shape models to address these issues. METHODS: Our CNN predicts the location of the prostate center and the parameters of the shape model, which determine the position of prostate surface keypoints. To train such a large model for segmentation of 3D images using small data (1) we adopt a stage-wise training strategy by first training the network to predict the prostate center and subsequently adding modules for predicting the parameters of the shape model and prostate rotation, (2) we propose a data augmentation method whereby the training images and their prostate surface keypoints are deformed according to the displacements computed based on the shape model, and (3) we employ various regularization techniques. RESULTS: Our proposed method achieves a Dice score of 0.88, which is obtained by using both elastic-net and spectral dropout for regularization. Compared with a standard CNN-based method, our method shows significantly better segmentation performance on the prostate base and apex. Our experiments also show that data augmentation using the shape model significantly improves the segmentation results. CONCLUSIONS: Prior knowledge about the shape of the target organ can improve the performance of CNN-based segmentation methods, especially where image features are not sufficient for a precise segmentation. Statistical shape models can also be employed to synthesize additional training data that can ease the training of large CNNs.
PURPOSE: Most of the existing convolutional neural network (CNN)-based medical image segmentation methods are based on methods that have originally been developed for segmentation of natural images. Therefore, they largely ignore the differences between the two domains, such as the smaller degree of variability in the shape and appearance of the target volume and the smaller amounts of training data in medical applications. We propose a CNN-based method for prostate segmentation in MRI that employs statistical shape models to address these issues. METHODS: Our CNN predicts the location of the prostate center and the parameters of the shape model, which determine the position of prostate surface keypoints. To train such a large model for segmentation of 3D images using small data (1) we adopt a stage-wise training strategy by first training the network to predict the prostate center and subsequently adding modules for predicting the parameters of the shape model and prostate rotation, (2) we propose a data augmentation method whereby the training images and their prostate surface keypoints are deformed according to the displacements computed based on the shape model, and (3) we employ various regularization techniques. RESULTS: Our proposed method achieves a Dice score of 0.88, which is obtained by using both elastic-net and spectral dropout for regularization. Compared with a standard CNN-based method, our method shows significantly better segmentation performance on the prostate base and apex. Our experiments also show that data augmentation using the shape model significantly improves the segmentation results. CONCLUSIONS: Prior knowledge about the shape of the target organ can improve the performance of CNN-based segmentation methods, especially where image features are not sufficient for a precise segmentation. Statistical shape models can also be employed to synthesize additional training data that can ease the training of large CNNs.
Entities:
Keywords:
Convolutional neural networks; Deep learning; Medical image segmentation; Prostate segmentation; Statistical shape models
Authors: Stefan Klein; Uulke A van der Heide; Irene M Lips; Marco van Vulpen; Marius Staring; Josien P W Pluim Journal: Med Phys Date: 2008-04 Impact factor: 4.071
Authors: Geert Litjens; Thijs Kooi; Babak Ehteshami Bejnordi; Arnaud Arindra Adiyoso Setio; Francesco Ciompi; Mohsen Ghafoorian; Jeroen A W M van der Laak; Bram van Ginneken; Clara I Sánchez Journal: Med Image Anal Date: 2017-07-26 Impact factor: 8.545
Authors: Geert Litjens; Robert Toth; Wendy van de Ven; Caroline Hoeks; Sjoerd Kerkstra; Bram van Ginneken; Graham Vincent; Gwenael Guillard; Neil Birbeck; Jindang Zhang; Robin Strand; Filip Malmberg; Yangming Ou; Christos Davatzikos; Matthias Kirschner; Florian Jung; Jing Yuan; Wu Qiu; Qinquan Gao; Philip Eddie Edwards; Bianca Maan; Ferdinand van der Heijden; Soumya Ghose; Jhimli Mitra; Jason Dowling; Dean Barratt; Henkjan Huisman; Anant Madabhushi Journal: Med Image Anal Date: 2013-12-25 Impact factor: 8.545
Authors: Mohammad Havaei; Axel Davy; David Warde-Farley; Antoine Biard; Aaron Courville; Yoshua Bengio; Chris Pal; Pierre-Marc Jodoin; Hugo Larochelle Journal: Med Image Anal Date: 2016-05-19 Impact factor: 8.545
Authors: Michelle Bardis; Roozbeh Houshyar; Chanon Chantaduly; Karen Tran-Harding; Alexander Ushinsky; Chantal Chahine; Mark Rupasinghe; Daniel Chow; Peter Chang Journal: Radiol Imaging Cancer Date: 2021-05
Authors: Michelle D Bardis; Roozbeh Houshyar; Peter D Chang; Alexander Ushinsky; Justin Glavis-Bloom; Chantal Chahine; Thanh-Lan Bui; Mark Rupasinghe; Christopher G Filippi; Daniel S Chow Journal: Cancers (Basel) Date: 2020-05-11 Impact factor: 6.639