UNet++: A Nested U-Net Architecture for Medical Image Segmentation
arXiv:1807.10165
Abstract
In this paper, we present UNet++, a new, more powerful architecture for medical image segmentation. Our architecture is essentially a deeply-supervised encoder-decoder network where the encoder and decoder sub-networks are connected through a series of nested, dense skip pathways. The re-designed skip pathways aim at reducing the semantic gap between the feature maps of the encoder and decoder sub-networks. We argue that the optimizer would deal with an easier learning task when the feature maps from the decoder and encoder networks are semantically similar. We have evaluated UNet++ in comparison with U-Net and wide U-Net architectures across multiple medical image segmentation tasks: nodule segmentation in the low-dose CT scans of chest, nuclei segmentation in the microscopy images, liver segmentation in abdominal CT scans, and polyp segmentation in colonoscopy videos. Our experiments demonstrate that UNet++ with deep supervision achieves an average IoU gain of 3.9 and 3.4 points over U-Net and wide U-Net, respectively.
arXiv comment8 pages, 3 figures, 3 tables, accepted by 4th Deep Learning in Medical Image Analysis (DLMIA) Workshop
References in corpus (1)
Cited by in corpus (36)
- RFN-Nest: An end-to-end residual fusion network for infrared and visible images
- Recent advances and clinical applications of deep learning in medical image analysis
- Swin-Unet: Unet-like Pure Transformer for Medical Image Segmentation
- OCNet: Object Context Network for Scene Parsing
- Embracing Imperfect Datasets: A Review of Deep Learning Solutions for Medical Image Segmentation
- Global Guidance Network for Breast Lesion Segmentation in Ultrasound Images
- COVID-19 Chest CT Image Segmentation -- A Deep Convolutional Neural Network Solution
- Domain adaptation based self-correction model for COVID-19 infection segmentation in CT images
- Multiscale Deep Equilibrium Models
- Fast meningioma segmentation in T1-weighted MRI volumes using a lightweight 3D deep learning architecture
- HR-Depth: High Resolution Self-Supervised Monocular Depth Estimation
- Macroscale fracture surface segmentation via semi-supervised learning considering the structural similarity
- BSN++: Complementary Boundary Regressor with Scale-Balanced Relation Modeling for Temporal Action Proposal Generation
- MAG-Net: Multi-task attention guided network for brain tumor segmentation and classification
- Deep Negative Volume Segmentation
- CF2-Net: Coarse-to-Fine Fusion Convolutional Network for Breast Ultrasound Image Segmentation
- UTRNet: High-Resolution Urdu Text Recognition In Printed Documents
- Crack Semantic Segmentation using the U-Net with Full Attention Strategy
- Efficient Palm-Line Segmentation with U-Net Context Fusion Module
- NightVision: Generating Nighttime Satellite Imagery from Infra-Red Observations
- Multi-organ Segmentation over Partially Labeled Datasets with Multi-scale Feature Abstraction
- A Comprehensive Study on Medical Image Segmentation using Deep Neural Networks
- Enhanced U-Net: A Feature Enhancement Network for Polyp Segmentation
- A Model-Driven Stack-Based Fully Convolutional Network for Pancreas Segmentation
- More than Encoder: Introducing Transformer Decoder to Upsample
- Object Based Attention Through Internal Gating
- Chest X-ray lung and heart segmentation based on minimal training sets
- EfficientHRNet: Efficient Scaling for Lightweight High-Resolution Multi-Person Pose Estimation
- Brain tumour segmentation using cascaded 3D densely-connected U-net
- UU-Nets Connecting Discriminator and Generator for Image to Image Translation
- ComboNet: Combined 2D & 3D Architecture for Aorta Segmentation
- PECNet: A Deep Multi-Label Segmentation Network for Eosinophilic Esophagitis Biopsy Diagnostics
- UXNet: Searching Multi-level Feature Aggregation for 3D Medical Image Segmentation
- Neural Architecture Search in operational context: a remote sensing case-study
- Analysis of Macula on Color Fundus Images Using Heightmap Reconstruction Through Deep Learning
- Segmentation of Microscopy Data for finding Nuclei in Divergent Images