-
Segmenting Medical Images with Limited Data
Authors:
Zhaoshan Liua,
Qiujie Lv,
Chau Hung Lee,
Lei Shen
Abstract:
While computer vision has proven valuable for medical image segmentation, its application faces challenges such as limited dataset sizes and the complexity of effectively leveraging unlabeled images. To address these challenges, we present a novel semi-supervised, consistency-based approach termed the data-efficient medical segmenter (DEMS). The DEMS features an encoder-decoder architecture and in…
▽ More
While computer vision has proven valuable for medical image segmentation, its application faces challenges such as limited dataset sizes and the complexity of effectively leveraging unlabeled images. To address these challenges, we present a novel semi-supervised, consistency-based approach termed the data-efficient medical segmenter (DEMS). The DEMS features an encoder-decoder architecture and incorporates the developed online automatic augmenter (OAA) and residual robustness enhancement (RRE) blocks. The OAA augments input data with various image transformations, thereby diversifying the dataset to improve the generalization ability. The RRE enriches feature diversity and introduces perturbations to create varied inputs for different decoders, thereby providing enhanced variability. Moreover, we introduce a sensitive loss to further enhance consistency across different decoders and stabilize the training process. Extensive experimental results on both our own and three public datasets affirm the effectiveness of DEMS. Under extreme data shortage scenarios, our DEMS achieves 16.85\% and 10.37\% improvement in dice score compared with the U-Net and top-performed state-of-the-art method, respectively. Given its superior data efficiency, DEMS could present significant advancements in medical segmentation under small data regimes. The project homepage can be accessed at https://github.com/NUS-Tim/DEMS.
△ Less
Submitted 12 July, 2024;
originally announced July 2024.
-
Neutron Tomography of Spent Fuel Casks
Authors:
Zhihua Liua,
Ming Fang,
Jon George,
Ling-Jian Meng,
Angela Di Fulvio
Abstract:
Dry casks for spent nuclear fuel (SNF) ensure the safe storage of SNF and provide radiation shielding. However, the presence of the thick casks encompassing several layers of steel and concrete makes inspection of the SNF a challenging task. Fast neutron interrogation is a viable method for the nondestructive assay of dry storage casks. In this study, we performed a Monte Carlo simulation-based st…
▽ More
Dry casks for spent nuclear fuel (SNF) ensure the safe storage of SNF and provide radiation shielding. However, the presence of the thick casks encompassing several layers of steel and concrete makes inspection of the SNF a challenging task. Fast neutron interrogation is a viable method for the nondestructive assay of dry storage casks. In this study, we performed a Monte Carlo simulation-based study associated with a machine-learning-based image reconstruction method to verify the content of SNF dry storage casks. We studied the use of neutron transmission and back-scattered measurements to assess the potential damage to fuel assemblies or fuel pin diversion during transportation of dry casks. We used Geant4 to model a realistic HI-STAR 100 cask, MPC-68 canister and basket, and GE-14 fuel assembly irradiated by a D-T neutron generator. Several bundle diversion scenarios were simulated. The angular distribution of the neutrons scattered by the cask was used to identify the diversions inside the fuel cask. A fuel bundle with at least 75% of its pins removed can be identified with a drop in the back-scattered signature larger than 2σ compared with a fully loaded scenario. We combined an iterative reconstruction algorithm with a convolutional neural network (CNN) to obtain a cross-sectional image of the fuel inside the cask. The proposed imaging approach allows locating the position of a missing fuel bundle with at least 75% of the pins removed when performing tomographic imaging of a canister with an overall scan time of less than two hours, when using a commercial neutron generator with a source strength of 10^10 n/s in the 4π solid angle.
△ Less
Submitted 17 October, 2021;
originally announced October 2021.
-
Enhancement of light absorption and oxygen vacancy formation in CeO2 by transition metal do**: A DFT study
Authors:
Zhao Liua,
Hongyang Ma,
Charles C. Sorrell,
Pramod Koshy,
Judy N. Hart
Abstract:
It has been demonstrated in previous experimental and computational work that do** CeO2 with transition metals is an effective way of tuning its properties. However, each previous study on CeO2 do** has been limited to a single or a few dopants. In this paper, we systematically study the formation energies, structural stability and electronic properties of CeO2 doped with the entire range of t…
▽ More
It has been demonstrated in previous experimental and computational work that do** CeO2 with transition metals is an effective way of tuning its properties. However, each previous study on CeO2 do** has been limited to a single or a few dopants. In this paper, we systematically study the formation energies, structural stability and electronic properties of CeO2 doped with the entire range of the ten 3d transition metals using density functional theory (DFT) calculations at the hybrid level. The formation energies of oxygen vacancies, and their effects on electronic properties, were also considered. It is found that most of the 3d transition metal dopants can lower the band gap of CeO2, with V and Co do** significantly reducing the band gap to less than 2.0 eV. Furthermore, all of the dopants can lower the formation energy of oxygen vacancies, and those with higher atomic numbers, particularly Cu and Zn, are most effective for this purpose. The electronic structures of doped CeO2 compensated by oxygen vacancies show that the presence of oxygen vacancies can further lower the band gap for most of the dopants, with V-, Cr-, Fe-, Co-, Ni-, and Cu-doped CeO2 all having band gaps of less than 2.0 eV. These results suggest that do** CeO2 with 3d transition metals could enhance the photocatalytic performance under visible light and increase the oxygen vacancy concentration, and they could provide a meaningful guide for the design of CeO2-based materials with improved photocatalytic and catalytic performance as well as enhanced ionic conductivity.
△ Less
Submitted 11 December, 2020;
originally announced December 2020.
-
Knowledge-aware Attention Network for Protein-Protein Interaction Extraction
Authors:
Huiwei Zhou,
Zhuang Liu1,
Shixian Ning,
Chengkun Lang,
Yingyu Lin,
Lei Du
Abstract:
Protein-protein interaction (PPI) extraction from published scientific literature provides additional support for precision medicine efforts. However, many of the current PPI extraction methods need extensive feature engineering and cannot make full use of the prior knowledge in knowledge bases (KB). KBs contain huge amounts of structured information about entities and relationships, therefore pla…
▽ More
Protein-protein interaction (PPI) extraction from published scientific literature provides additional support for precision medicine efforts. However, many of the current PPI extraction methods need extensive feature engineering and cannot make full use of the prior knowledge in knowledge bases (KB). KBs contain huge amounts of structured information about entities and relationships, therefore plays a pivotal role in PPI extraction. This paper proposes a knowledge-aware attention network (KAN) to fuse prior knowledge about protein-protein pairs and context information for PPI extraction. The proposed model first adopts a diagonal-disabled multi-head attention mechanism to encode context sequence along with knowledge representations learned from KB. Then a novel multi-dimensional attention mechanism is used to select the features that can best describe the encoded context. Experiment results on the BioCreative VI PPI dataset show that the proposed approach could acquire knowledge-aware dependencies between different words in a sequence and lead to a new state-of-the-art performance.
△ Less
Submitted 7 January, 2020;
originally announced January 2020.
-
Extraordinary acoustic reflection enhancement by acoustically-transparent thin plates
Authors:
Zhaojian He,
Shasha Peng,
Chunyin Qiu,
Rui Hao,
Gangqiang Liu,
Manzhu Ke,
Jun Mei,
Weijia Wen,
Zhengyou Liu?
Abstract:
We report an observation of the extraordinary high reflection of acoustic waves in water by thin epoxy plates partitioned by subwavelength cuts, whereas such plates without structure are acoustically-transparent as the acoustic properties of epoxy are close to water. It is demonstrated that this exotic phenomenon results from the resonant excitation of the local modes within the individual pieces…
▽ More
We report an observation of the extraordinary high reflection of acoustic waves in water by thin epoxy plates partitioned by subwavelength cuts, whereas such plates without structure are acoustically-transparent as the acoustic properties of epoxy are close to water. It is demonstrated that this exotic phenomenon results from the resonant excitation of the local modes within the individual pieces derived by the cuts. The experiment agrees well with the theory.
△ Less
Submitted 16 April, 2011; v1 submitted 23 December, 2009;
originally announced December 2009.
-
An efficient approach of controlling traffic congestion in scale-free networks
Authors:
Zonghua Liua,
Weichuan Ma,
Huan Zhang,
Yin Sun,
P. M. Hui
Abstract:
We propose and study a model of traffic in communication networks. The underlying network has a structure that is tunable between a scale-free growing network with preferential attachments and a random growing network. To model realistic situations where different nodes in a network may have different capabilities, the message or packet creation and delivering rates at a node are assumed to depe…
▽ More
We propose and study a model of traffic in communication networks. The underlying network has a structure that is tunable between a scale-free growing network with preferential attachments and a random growing network. To model realistic situations where different nodes in a network may have different capabilities, the message or packet creation and delivering rates at a node are assumed to depend on the degree of the node. Noting that congestions are more likely to take place at the nodes with high degrees in networks with scale-free character, an efficient approach of selectively enhancing the message-processing capability of a small fraction (e.g. 3%) of the nodes is shown to perform just as good as enhancing the capability of all nodes. The interplay between the creation rate and the delivering rate in determining non-congested or congested traffic in a network is studied more numerically and analytically.
△ Less
Submitted 11 June, 2008;
originally announced June 2008.
