Med3D: Transfer Learning for 3D Medical Image Analysis (2019-04-01T00:00:00.000000Z)

TL;DR

A heterogeneous 3D network called Med3D is designed to co-train multi-domain 3DSeg-8 so as to make a series of pre-trained models which can accelerate the training convergence speed of target 3D medical tasks and improve accuracy ranging from 3% to 20%.

Abstract

The performance on deep learning is significantly affected by volume of training data. Models pre-trained from massive dataset such as ImageNet become a powerful weapon for speeding up training convergence and improving accuracy. Similarly, models based on large dataset are important for the development of deep learning in 3D medical images. However, it is extremely challenging to build a sufficiently large dataset due to difficulty of data acquisition and annotation in 3D medical imaging. We aggregate the dataset from several medical challenges to build 3DSeg-8 dataset with diverse modalities, target organs, and pathologies. To extract general medical three-dimension (3D) features, we design a heterogeneous 3D network called Med3D to co-train multi-domain 3DSeg-8 so as to make a series of pre-trained models. We transfer Med3D pre-trained models to lung segmentation in LIDC dataset, pulmonary nodule classification in LIDC dataset and liver segmentation on LiTS challenge. Experiments show that the Med3D can accelerate the training convergence speed of target 3D medical tasks 2 times compared with model pre-trained on Kinetics dataset, and 10 times compared with training from scratch as well as improve accuracy ranging from 3% to 20%. Transferring our Med3D model on state-the-of-art DenseASPP segmentation network, in case of single model, we achieve 94.6\% Dice coefficient which approaches the result of top-ranged algorithms on the LiTS challenge.

Authors

Sihong Chen

1 papers

Kai Ma

2 papers

Yefeng Zheng

1 papers

TL;DR

Abstract

Authors

References43 items

Evaluation of algorithms for Multi-Modality Whole Heart Segmentation: An open-access grand challenge

Fully Automatic Left Atrium Segmentation From Late Gadolinium Enhanced Magnetic Resonance Imaging Using a Dual Fully Convolutional Neural Network

Rethinking ImageNet Pre-Training

3D Deep Learning from CT Scans Predicts Tumor Invasiveness of Subcentimeter Pulmonary Adenocarcinomas.

H-DenseUNet: Hybrid Densely Connected UNet for Liver and Tumor Segmentation From CT Volumes

DenseASPP for Semantic Segmentation in Street Scenes

Can Spatiotemporal 3D CNNs Retrace the History of 2D CNNs and ImageNet?

3D Anisotropic Hybrid Network: Transferring Convolutional Features from 2D Images to 3D Anisotropic Volumes

Recurrent Saliency Transformation Network: Incorporating Multi-stage Visual Cues for Small Organ Segmentation

Revisiting Unreasonable Effectiveness of Data in Deep Learning Era

The Kinetics Human Action Video Dataset

Evaluation of segmentation methods on head and neck CT: Auto‐segmentation challenge 2015

Automatic Liver Lesion Segmentation Using A Deep Convolutional Neural Network Method

MR‐based synthetic CT generation using a deep convolutional neural network method

Understanding the Mechanisms of Deep Transfer Learning for Medical Images

Automatic segmentation of left ventricular myocardium by deep convolutional and de-convolutional neural networks

3D U-Net: Learning Dense Volumetric Segmentation from Sparse Annotation

V-Net: Fully Convolutional Neural Networks for Volumetric Medical Image Segmentation

Cloud-Based Evaluation of Anatomical Structure Segmentation and Landmark Detection Algorithms: VISCERAL Anatomy Benchmarks

DeepLab: Semantic Image Segmentation with Deep Convolutional Nets, Atrous Convolution, and Fully Connected CRFs

Identity Mappings in Deep Residual Networks

Convolutional Neural Networks for Medical Image Analysis: Full Training or Fine Tuning?

Deep Residual Learning for Image Recognition

Learning Multi-domain Convolutional Neural Networks for Visual Tracking

The Multimodal Brain Tumor Image Segmentation Benchmark (BRATS)

Food image recognition using deep convolutional network with pre-training and fine-tuning

Benchmark for Algorithms Segmenting the Left Atrium From 3D CT and MRI Datasets

Texture Feature Analysis for Computer-Aided Diagnosis on Pulmonary Nodules

Adam: A Method for Stochastic Optimization

ImageNet Large Scale Visual Recognition Challenge

Microsoft COCO: Common Objects in Context

The Cancer Imaging Archive (TCIA): Maintaining and Operating a Public Information Repository

ImageNet classification with deep convolutional neural networks

Discovering Latent Domains for Multisource Domain Adaptation

The Lung Image Database Consortium (LIDC) and Image Database Resource Initiative (IDRI): a completed reference database of lung nodules on CT scans.

A Survey on Transfer Learning

Domain adaptation from multiple sources via auxiliary classifiers

The Difficulty of Training Deep Architectures and the Effect of Unsupervised Pre-Training

Lung image database consortium: developing a resource for the medical imaging research community.

Beyond thick versus thin: Mapping cranial vault thickness patterns in recent Homo sapiens

Liver tumor segmentation challenge

Edinburgh Research Explorer The PASCAL Visual Object Classes (VOC) Challenge

Grand challenges in biomedical image analysis

Field of Study

Journal Information

Name

Volume

Venue Information

Name

Type

URL

Alternate Names