V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation
By Fausto Milletarì, Nassir Navab, Seyed-Ahmad Ahmadi. 2016
11,050 citations recorded by the source corpus.
Convolutional Neural Networks (CNNs) have been recently employed to solve problems from both the computer vision and medical image analysis fields. Despite their popularity, most approaches are only able to process 2D images while most medical data used in clinical practice consists of 3D volumes. In this work we propose an approach to 3D image segmentation based on a volumetric, fully convolutional, neural network. Our CNN is trained end-to-end on MRI volumes depicting prostate, and learns to predict segmentation for the whole volume at once. We introduce a novel objective function, that we optimise during training, based on Dice coefficient. In this way we can deal with situations where there is a strong imbalance between the number of foreground and background voxels. To cope with the limited number of annotated volumes available for training, we augment the data applying random non-linear transformations and histogram matching. We show in our experimental evaluation that our approach achieves good performances on challenging test data while requiring only a fraction of the processing time needed by other previous methods.
Subfields: Biomedical Engineering, Computer Vision and Pattern Recognition
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
- P. Heng
- Qi-Yan Dou
- Holger R. Roth
- Yi Yang
- Arvid Lundervold
- Adrian V. Dalca
- Andriy Myronenko
- Daguang Xu
In Computer Vision and Pattern Recognition
- Luc Van Gool
- Dacheng Tao
- Ming-Hsuan Yang
- Xiaogang Wang
- Alan L. Yuille
- Trevor Darrell
- Chunhua Shen
- Jitendra Malik
Citations in the corpus
- End-to-End Object Detection with Transformers (2020, methodology): “The final resolution of the masks has stride 4 and each mask is supervised independently using the DICE/F-1 loss [28] and Focal loss [23].”
- Segment Anything (2023, methodology): “We supervise mask prediction with the linear combination of focal loss [65] and dice loss [73] used in [14].”
- Masked-attention Mask Transformer for Universal Image Segmentation (2021, methodology): “We use the binary cross-entropy loss (instead of focal loss [34] in [14]) and the dice loss [41] for our mask loss: Lmask = ceLce + diceLdice.”
- Learning Open-World Object Proposals without Learning to Classify (2021, neutral): “Dice coefficient [42] and generalized-IoU [43].”
- Per-Pixel Classification is Not All You Need for Semantic Segmentation (2021, methodology): “We use focal loss [27] and dice loss [33] for our mask loss: Lmask(m,m) = λfocalLfocal(m,m) + λdiceLdice(m,m), and set the hyper-parameters to λfocal = 20.”
- K-Net: Towards Unified Image Segmentation (2021, methodology): “Therefore, we apply Dice loss [42] to handle this issue following previous works [50, 55, 56].”
- MaX-DeepLab: End-to-End Panoptic Segmentation with Mask Transformers (2020, methodology): “This is consistent with the literature [42, 83, 10] that uses a cross-entropy loss together with a dice loss [69] to learn better segmentation masks.”
- A Survey on Explainable Artificial Intelligence (XAI): Toward Medical XAI (2019, background): “While machine learning methods, artificial neural networks, brain-machine interfaces and related subfields have recently demonstrated promising performance in performing medical tasks, they are hardly perfect [1] [2] [3] [4] [5] [6] [7] [8].”
- Searching Learning Strategy with Reinforcement Learning for 3D Medical Image Segmentation (2019, methodology): “By default, all training jobs use the Adam optimizer, and the Dice loss is used for gradient computing [11].”
- NAS-Unet: Neural Architecture Search for Medical Image Segmentation (2019, methodology): “[30] proposed a 3D deformation architecture called V-net based on U-Net.”
- An overview of deep learning in medical imaging focusing on MRI (2018, methodology): “V-net [84] A three-dimensional version of U-net with volumetric convolutions and skip connections as in ResNet.”
- OctNet: Learning Deep 3D Representations at High Resolutions (2016, background): “Increasing the resolution naı̈vely [32, 40, 52] reduces the depth of the networks and hence their expressiveness.”
- MDETR - Modulated Detection for End-to-End Multi-Modal Understanding (2021, methodology): “In the second stage, following [2], we freeze the weights of the network, and only train a segmentation head for 35 epochs, with a learning rate drop at 25 epochs, supervised using a combination of the Dice/F1 loss[32] and the Focal loss [25].”
- Globally Optimal Gradient Descent for a ConvNet with Gaussian Inputs (2017, methodology): “Then the output of the network in Figure 1 is given by: We note that such architectures have been used in several works (Lin et al., 2013; Milletari et al., 2016), but we view them as important firstly because they capture key properties of general convolutional networks.”