Iterative Label Denoising Network: Segmenting Male Pelvic Organs in CT From 3D Bounding Box Annotations

Shuai Wang,Qian Wang,Yeqin Shao,Liangqiong Qu,Chunfeng Lian,Jun Lian,Dinggang Shen

DOI: https://doi.org/10.1109/tbme.2020.2969608

IF: 4.6

2020-10-01

IEEE Transactions on Biomedical Engineering

Abstract:Obtaining accurate segmentation of the prostate and nearby organs at risk (e.g., bladder and rectum) in CT images is critical for radiotherapy of prostate cancer. Currently, the leading automatic segmentation algorithms are based on Fully Convolutional Networks (FCNs), which achieve remarkable performance but usually need large-scale datasets with high-quality voxel-wise annotations for full supervision of the training. Unfortunately, such annotations are difficult to acquire, which becomes a bottleneck to build accurate segmentation models in real clinical applications. In this paper, we propose a novel weakly supervised segmentation approach that only needs 3D bounding box annotations covering the organs of interest to start the training. Obviously, the bounding box includes many non-organ voxels that carry noisy labels to mislead the segmentation model. To this end, we propose the label denoising module and embed it into the iterative training scheme of the label denoising network (LDnet) for segmentation. The labels of the training voxels are predicted by the tentative LDnet, while the label denoising module identifies the voxels with unreliable labels. As only the good training voxels are preserved, the iteratively re-trained LDnet can refine its segmentation capability gradually. Our results are remarkable, i.e., reaching $sim$94% (prostate), $sim$91% (bladder), and $sim$86% (rectum) of the Dice Similarity Coefficients (DSCs), compared to the case of fully supervised learning upon high-quality voxel-wise annotations and also superior to several state-of-the-art approaches. To our best knowledge, this is the first work to achieve voxel-wise segmentation in CT images from simple 3D bounding bo- annotations, which can greatly reduce many labeling efforts and meet the demands of the practical clinical applications.

engineering, biomedical

What problem does this paper attempt to address?

The problem that this paper attempts to solve is to design an energy - efficient hybrid precoder and combiner in millimeter - wave (mmWave) communication systems in order to maximize the energy efficiency in point - to - point large - scale multiple - input multiple - output (MIMO) wireless communications. Specifically, the paper focuses on the design of hybrid precoders and combiners under the sub - connected architecture, where each radio - frequency (RF) link is only connected to a subset of base - station antennas, thereby achieving energy efficiency from the perspective of transmission. Since this is a non - convex and NP - hard problem and is very difficult to solve directly, the paper proposes a two - layer optimization method to solve this problem. It optimizes the analog precoder and combiner by jointly using interference alignment and fractional programming techniques, as well as the digital precoder and combiner based on effective MIMO channel coefficients. In addition, the paper also proves the convergence of the proposed algorithm and verifies the effectiveness of the proposed method through extensive simulation results, and evaluates the energy - efficiency performance under different system configurations.

Iterative Label Denoising Network: Segmenting Male Pelvic Organs in CT From 3D Bounding Box Annotations

Deep Reinforcement Learning for Weakly-Supervised Lymph Node Segmentation in CT Images

CT Male Pelvic Organ Segmentation Using Fully Convolutional Networks with Boundary Sensitive Representation

Automatic CT Segmentation from Bounding Box Annotations using Convolutional Neural Networks

Pelvic Organ Segmentation Using Distinctive Curve Guided Fully Convolutional Networks

Automated Prostate Segmentation of Volumetric CT Images Using 3D Deeply Supervised Dilated FCN.

Semi-supervised semantic segmentation of prostate and organs-at-risk on 3D pelvic CT images

2D and 3D Bladder Segmentation Using U-Net-based Deep-Learning

Domain adversarial networks and intensity-based data augmentation for male pelvic organ segmentation in cone beam CT

CHDNet: A lightweight weakly supervised segmentation network for lung CT image

Male Pelvic Multi-Organ Segmentation Using V-transformer Network.

A Combined Learning Algorithm for Prostate Segmentation on 3D CT Images

Boundary Coding Representation for Organ Segmentation in Prostate Cancer Radiotherapy

Point-Sampling Method Based on 3D U-Net Architecture to Reduce the Influence of False Positive and Solve Boundary Blur Problem in 3D CT Image Segmentation

A Learning-Based CT Prostate Segmentation Method Via Joint Transductive Feature Selection and Regression

Scribble-based 3D Multiple Abdominal Organ Segmentation via Triple-branch Multi-dilated Network with Pixel- and Class-wise Consistency

Label-efficient Multi-organ Segmentation Method with Diffusion Model

Precise segmentation of multiple organs in CT volumes using learning-based approach and information theory.

Segmentation of the prostate and organs at risk in male pelvic CT images using deep learning

Efficient Multi-Organ Segmentation from 3D Abdominal CT Images with Lightweight Network and Knowledge Distillation.

Automatic bladder segmentation from CT images using deep CNN and 3D fully connected CRF-RNN