arXiv subjects

Ke Zou

Publications and source records attributed to Ke Zou.

50 records · Page 3Linked to original sources

arXiv2023

Uncertainty-informed Mutual Learning for Joint Medical Image Classification and Segmentation

Classification and segmentation are crucial in medical image analysis as they enable accurate diagnosis and disease monitoring. However, current methods often prioritize the mutual learning features and shared model parameters, while neglecting the reliability of features and performances. In this paper, we propose a novel Uncertainty-informed Mutual Learning (UML) framework for reliable and interpretable medical image analysis. Our UML introduces reliability to joint classification and segmentation tasks, leveraging mutual learning with uncertainty to improve performance. To achieve this, we first use evidential deep learning to provide image-level and pixel-wise confidences. Then, an Uncertainty Navigator Decoder is constructed for better using mutual features and generating segmentation results. Besides, an Uncertainty Instructor is proposed to screen reliable masks for classification. Overall, UML could produce confidence estimation in features and performance for each link (classification and segmentation). The experiments on the public datasets demonstrate that our UML outperforms existing methods in terms of both accuracy and robustness. Our UML has the potential to explore the development of more reliable and explainable medical image analysis models. We will release the codes for reproduction after acceptance.

cs.CV↗

arXiv2023

Federated Uncertainty-Aware Aggregation for Fundus Diabetic Retinopathy Staging

Deep learning models have shown promising performance in the field of diabetic retinopathy (DR) staging. However, collaboratively training a DR staging model across multiple institutions remains a challenge due to non-iid data, client reliability, and confidence evaluation of the prediction. To address these issues, we propose a novel federated uncertainty-aware aggregation paradigm (FedUAA), which considers the reliability of each client and produces a confidence estimation for the DR staging. In our FedUAA, an aggregated encoder is shared by all clients for learning a global representation of fundus images, while a novel temperature-warmed uncertainty head (TWEU) is utilized for each client for local personalized staging criteria. Our TWEU employs an evidential deep layer to produce the uncertainty score with the DR staging results for client reliability evaluation. Furthermore, we developed a novel uncertainty-aware weighting module (UAW) to dynamically adjust the weights of model aggregation based on the uncertainty score distribution of each client. In our experiments, we collect five publicly available datasets from different institutions to conduct a dataset for federated DR staging to satisfy the real non-iid condition. The experimental results demonstrate that our FedUAA achieves better DR staging performance with higher reliability compared to other federated learning methods. Our proposed FedUAA paradigm effectively addresses the challenges of collaboratively training DR staging models across multiple institutions, and provides a robust and reliable solution for the deployment of DR diagnosis models in real-world clinical scenarios.

eess.IV↗

arXiv2023

SAM-U: Multi-box prompts triggered uncertainty estimation for reliable SAM in medical image

Recently, Segmenting Anything has taken an important step towards general artificial intelligence. At the same time, its reliability and fairness have also attracted great attention, especially in the field of health care. In this study, we propose multi-box prompts triggered uncertainty estimation for SAM cues to demonstrate the reliability of segmented lesions or tissues. We estimate the distribution of SAM predictions via Monte Carlo with prior distribution parameters, which employs different prompts as formulation of test-time augmentation. Our experimental results found that multi-box prompts augmentation improve the SAM performance, and endowed each pixel with uncertainty. This provides the first paradigm for a reliable SAM.

cs.CV↗

arXiv2023

A Multi-view Impartial Decision Network for Frontotemporal Dementia Diagnosis

Frontotemporal Dementia (FTD) diagnosis has been successfully progress using deep learning techniques. However, current FTD identification methods suffer from two limitations. Firstly, they do not exploit the potential of multi-view functional magnetic resonance imaging (fMRI) for classifying FTD. Secondly, they do not consider the reliability of the multi-view FTD diagnosis. To address these limitations, we propose a reliable multi-view impartial decision network (MID-Net) for FTD diagnosis in fMRI. Our MID-Net provides confidence for each view and generates a reliable prediction without any conflict. To achieve this, we employ multiple expert models to extract evidence from the abundant neural network information contained in fMRI images. We then introduce the Dirichlet Distribution to characterize the expert class probability distribution from an evidence level. Additionally, a novel Impartial Decision Maker (IDer) is proposed to combine the different opinions inductively to arrive at an unbiased prediction without additional computation cost. Overall, our MID-Net dynamically integrates the decisions of different experts on FTD disease, especially when dealing with multi-view high-conflict cases. Extensive experiments on a high-quality FTD fMRI dataset demonstrate that our model outperforms previous methods and provides high uncertainty for hard-to-classify examples. We believe that our approach represents a significant step toward the deployment of reliable FTD decision-making under multi-expert conditions. We will release the codes for reproduction after acceptance.

cs.CV↗

arXiv2023

A Review of Uncertainty Estimation and its Application in Medical Imaging

The use of AI systems in healthcare for the early screening of diseases is of great clinical importance. Deep learning has shown great promise in medical imaging, but the reliability and trustworthiness of AI systems limit their deployment in real clinical scenes, where patient safety is at stake. Uncertainty estimation plays a pivotal role in producing a confidence evaluation along with the prediction of the deep model. This is particularly important in medical imaging, where the uncertainty in the model's predictions can be used to identify areas of concern or to provide additional information to the clinician. In this paper, we review the various types of uncertainty in deep learning, including aleatoric uncertainty and epistemic uncertainty. We further discuss how they can be estimated in medical imaging. More importantly, we review recent advances in deep learning models that incorporate uncertainty estimation in medical imaging. Finally, we discuss the challenges and future directions in uncertainty estimation in deep learning for medical imaging. We hope this review will ignite further interest in the community and provide researchers with an up-to-date reference regarding applications of uncertainty estimation models in medical imaging.

eess.IV↗

arXiv2022

Realistic simulation of reflection high-energy electron diffraction patterns for two-dimensional lattices using Ewald construction

Reflection high-energy electron diffraction (RHEED) is a powerful tool for characterizing crystal surface structures. However, the setup geometry leads to distorted and complicated patterns, which are not straightforward to link to the real-space structures. A program with a graphical user interface is provided here to simulate the RHEED patterns. Following the Ewald construction in the kinematic theory, we find out the exact geometric transformation in this model that determines the positions of diffraction spots. The program can deal with many forms of surface structures, including surface reconstructions or domains. The simulations exhibit great agreement with the experimental results in various cases. This program will benefit the structure analysis in thin film growth and surface science studies.

cond-mat.mtrl-sci↗

arXiv2022

TBraTS: Trusted Brain Tumor Segmentation

Despite recent improvements in the accuracy of brain tumor segmentation, the results still exhibit low levels of confidence and robustness. Uncertainty estimation is one effective way to change this situation, as it provides a measure of confidence in the segmentation results. In this paper, we propose a trusted brain tumor segmentation network which can generate robust segmentation results and reliable uncertainty estimations without excessive computational burden and modification of the backbone network. In our method, uncertainty is modeled explicitly using subjective logic theory, which treats the predictions of backbone neural network as subjective opinions by parameterizing the class probabilities of the segmentation as a Dirichlet distribution. Meanwhile, the trusted segmentation framework learns the function that gathers reliable evidence from the feature leading to the final segmentation results. Overall, our unified trusted segmentation framework endows the model with reliability and robustness to out-of-distribution samples. To evaluate the effectiveness of our model in robustness and reliability, qualitative and quantitative experiments are conducted on the BraTS 2019 dataset.

eess.IV↗

arXiv2020

Controlling the electrical and magnetic ground states by doping in the complete phase diagram of titanate Eu1-xLaxTiO3 thin films

EuTiO3, a band insulator, and LaTiO3, a Mott insulator, are both antiferromagnetic with transition temperatures ~ 5.5 K and ~ 160 K, respectively. Here, we report the synthesis of Eu1-xLaxTiO3 thin films with x = 0 to 1 by oxide molecular beam epitaxy. The films in the full range have high crystalline quality and show no phase segregation, allowing us carry out transport measurements to study their electrical and magnetic properties. From x = 0.03 to 0.95, Eu1-xLaxTiO3 films show conduction by electrons as charge carriers, with differences in carrier densities and mobilities, contrary to the insulating nature of pure EuTiO3 and LaTiO3. Following a rich phase diagram, the magnetic ground states of the films vary with increasing La-doping level, changing Eu1-xLaxTiO3 from an antiferromagnetic insulator to an antiferromagnetic metal, a ferromagnetic metal, a paramagnetic metal, and back to an antiferromagnetic insulator. These emergent properties reflect the evolutions of the band structure, mainly at the Ti t2g bands near the Fermi level, when Eu2+ are gradually replaced by La3+. This work sheds light on this method for designing the electrical and magnetic properties in strongly-correlated oxides and completes the phase diagram of the titanate Eu1-xLaxTiO3.

cond-mat.str-el↗

arXiv2020

Tuning stoichiometry and its impact on superconductivity of monolayer and multilayer FeSe on SrTiO3

Synthesis of monolayer FeSe on SrTiO3, with greatly enhanced superconductivity compared to bulk FeSe, remains difficult. Lengthy annealing within a certain temperature window is always required to achieve superconducting samples as reported by different groups around the world, but the mechanism of annealing in inducing superconductivity has not been elucidated. We grow FeSe films on SrTiO3 by molecular beam epitaxy and adjust the stoichiometry by depositing additional small amounts of Fe atoms. The monolayer films become superconducting after the Fe deposition without any annealing, and show similar superconducting transition temperatures as those of the annealed films in transport measurements. We also demonstrate on the 5-unit-cell films that the FeSe multilayer can be reversibly tuned between the non-superconducting $\sqrt{5} \times \sqrt{5}$ phase with Fe-vacancies and superconducting $1 \times 1$ phase. Our results reveal that the traditional anneal process in essence removes Fe vacancies and the additional Fe deposition serves as a more efficient way to achieve superconductivity. This work highlights the significance of stoichiometry in the superconductivity of FeSe thin films and provides an easy path for superconducting samples.

cond-mat.supr-con↗

arXiv2019

Epitaxial Growth of Perovskite SrBiO$_3$ Film on SrTiO$_3$ by Oxide Molecular Beam Epitaxy

Hole-doped perovskite bismuthates such as Ba$_{1-x}$K$_x$BiO$_3$ and Sr$_{1-x}$K$_x$BiO$_3$ are well-known bismuth-based oxide high-transition-temperature superconductors. Reported thin bismuthate films show relatively low quality, likely due to their large lattice mismatch with the substrate and a low sticking coefficient of Bi at high temperatures. Here, we report the successful epitaxial thin film growth of the parent compound strontium bismuthate SrBiO$_3$ on SrO-terminated SrTiO$_3$ (001) substrates by molecular beam epitaxy. Two different growth methods, high-temperature co-deposition or recrystallization cycles of low-temperature deposition plus high-temperature annealing, are developed to improve the epitaxial growth. SrBiO$_3$ has a pseudocubic lattice constant $\sim$4.25 Å, an $\sim$8.8\% lattice mismatch on SrTiO$_3$ substrate, leading to a large strain in the first few unit cells. Films thicker than 6 unit cells prepared by both methods are fully relaxed to bulk lattice constant and have similar quality. Compared to high-temperature co-deposition, the recrystallization method can produce higher quality 1-6 unit cell films that are coherently or partially strained. Photoemission experiments reveal the bonding and antibonding states close to the Fermi level due to Bi and O hybridization, in good agreement with density functional theory calculations. This work provides general guidance to the synthesis of high-quality perovskite bismuthate films.

cond-mat.supr-con↗

arXiv2018

Suppression of the Spectral Weight of Topological Surface States on the Nanoscale via Local Symmetry Breaking

In topological crystalline insulators the topological conducting surface states are protected by crystal symmetry, in principle making it possible to pattern nanoscale insulating and conductive motifs solely by breaking local symmetries on an otherwise homogenous, single-phase material. We show using scanning tunneling microscopy/spectroscopy that defects that break local symmetry of SnTe suppress electron tunneling over an energy range as large as the bulk band gap, an order of magnitude larger than that produced globally via magnetic fields or uniform structural perturbations. Complementary ab initio calculations show how local symmetry breaking obstructs topological surface states as shown by a threefold reduction of the spectral weight of the topological surface states. The finding highlights the potential benefits of manipulating the surface morphology to create devices that take advantage of the unique properties of topological surface states and can operate at practical temperatures.

cond-mat.mtrl-sci↗

arXiv2016

The role of double TiO2 layers at the interface of FeSe/SrTiO3 superconductors

We determine the surface reconstruction of SrTiO3 used to achieve superconducting FeSe films in experiments, which is different from the 1x1 TiO2 terminated SrTiO3 assumed by most previous theoretical studies. In particular, we identify the existence of a double TiO2 layer at the SrTiO3-FeSe interface that plays two important roles. First, it facilitates the epitaxial growth of FeSe. Second, ab initio calculations reveal a strong tendency for electrons to transfer from an oxygen deficient SrTiO3 surface to FeSe when the double TiO2 layer is present. As a better electron donor than previously proposed interfacial structures, the double layer helps to remove the hole pocket in the FeSe at the Γ point of the Brillouin zone and leads to a band structure characteristic of superconducting samples. The characterization of the interface structure presented here is a key step towards the resolution of many open questions about this novel superconductor.

cond-mat.str-el↗

arXiv2010

Mechanism for current saturation and energy dissipation in graphene transistors

From a combination of careful and detailed theoretical and experimental studies, we demonstrate that the Boltzmann theory including all scattering mechanisms gives an excellent account, with no adjustable parameters, of high electric field transport in single as well as double-oxide graphene transistors. We further show unambiguously that scattering from the substrate and superstrate surface optical (SO) phonons governs the high field transport and heat dissipation over a wide range of experimentally relevant parameters. Models that neglect SO phonons altogether or treat them in a simple phenomenological manner are inadequate. We outline possible strategies for achieving higher current and complete saturation in graphene devices.

cond-mat.mes-hall↗

arXiv2009

The Deposition of High-Quality HfO2 on Graphene and the Effect of Remote Oxide Phonon Scattering

We demonstrate the atomic layer deposition of high-quality HfO2 film on graphene and report the magnitude of remote oxide phonon (ROP) scattering in dual-oxide graphene transistors. Top gates with 30 nm HfO2 oxide layer exhibit excellent doping capacity of greater than 1.5x10^(13)/cm^(2). The carrier mobility in HfO2-covered graphene reaches 20,000 cm^(2)/Vs at low temperature, which is the highest among oxide-covered graphene and compares to that of pristine samples. The temperature-dependent resistivity exhibits the effect of ROP scattering from both the SiO2 substrate and the HfO2 over-layer. At room temperature, surface phonon modes of the HfO2 film centered at 54 meV dominate and limit the carrier mobility to ~20,000 cm^(2)/Vs. Our results highlight the important choice of oxide in graphene devices.

cond-mat.mtrl-sci↗