Hough-CNN: Deep learning for segmentation of deep brain regions in MRI and ultrasound

By Fausto Milletarì, Seyed-Ahmad Ahmadi, Christine Kroll, A. Plate, Verena E. Rozanski, Juliana Maiostre, J. Levin, O. Dietrich, B. Ertl-Wagner, K. Boetzel and 1 more. 2016

378 citations recorded by the source corpus.

Abstract In this work we propose a novel approach to perform segmentation by leveraging the abstraction capabilities of convolutional neural networks (CNNs). Our method is based on Hough voting, a strategy that allows for fully automatic localisation and segmentation of the anatomies of interest. This approach does not only use the CNN classification outcomes, but it also implements voting by exploiting the features produced by the deepest portion of the network. We show that this learning-based segmentation method is robust, multi-region, flexible and can be easily adapted to different modalities. In the attempt to show the capabilities and the behaviour of CNNs when they are applied to medical image analysis, we perform a systematic study of the performances of six different network architectures, conceived according to state-of-the-art criteria, in various situations. We evaluate the impact of both different amount of training data and different data dimensionality (2D, 2.5D and 3D) on the final results. We show results on both MRI and transcranial US volumes depicting respectively 26 regions of the basal ganglia and the midbrain.

Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition, Model Architectures

How researchers see this paper

The researchers whose trusted networks rank it highest, strongest advocate first. Every researcher gets a different order.

In Biomedical Engineering

In Computer Vision and Pattern Recognition

In Model Architectures

Citations in the corpus

V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation (2016, background): “Post-processing approaches such as connected components analysis normally yield no improvement and therefore, more recent works, propose to use the network predictions in combination with Markov random fields [9], voting strategies [12] or more traditional approaches such as level-sets [2].”
A survey on deep learning in medical image analysis (2017, methodology): “Milletari et al. (2016b) proposed an extension to the U-Net layout that incorporates ResNet-like residual blocks and a Dice loss layer, rather than the conventional cross-entropy, that directly min- imizes this commonly used segmentation error measure.”
Deep Learning Techniques for Medical Image Segmentation: Achievements and Challenges (2019, methodology): “The performance relatively depends on the overlapping of the patches and the size of mini-patches [52].”
Deep Learning in Medical Ultrasound Analysis: A Review (2019, methodology): “5 [21,38,44,46,50,57,124–126].”
FastSurfer - A fast and accurate deep learning based neuroimaging pipeline (2019, neutral): “…Hossei-ni-Asl et al., 2016; Lee et al. Kim) or segmentation of a smaller number of brain (sub-)structures (Zhang et al., 2015; Akkus et al., 2017; Milletari et al., 2017; Fedorov et al., 2017; Dolz et al., 2018; Thyreau et al., 2018; Chen et al., 2018; Nogovitsyn et al., 2019; Li et al., 2019;…”
Deep Hough Voting for 3D Object Detection in Point Clouds (2019, background): “Similarly [31] learned deep features to build code-books for segmentation in MRI and ultrasiounds images.”
PSNet: prostate segmentation on MRI based on a convolutional neural network (2018, methodology): “6,8 To obtain a pixel-wise segmentation, some researchers 9 – 15 proposed a patch-wise segmentation method, which extracts small patches (e.g., 32 × 32 ) from images and then trains a patch-wise convolutional neural network (CNN) model.”
Boundary loss for highly unbalanced segmentation (2018, methodology): “The well-known Dice overlap coeﬃcient was also adopted as a regional loss function, typically outperforming CE in unbalanced medical image segmentation problems ( Milletari et al., 2016; 2017; Wong et al., 2018 ).”
3D fully convolutional networks for subcortical segmentation in MRI: A large‐scale study (2016, background): “As discussed recently in [26], considering 3D MRI data directly, instead of slice-by-slice, can improve the performance of a segmentation method.”