A5101400967- Advanced Radiotherapy Techniques
- Advanced X-ray and CT Imaging
- Medical Imaging Techniques and Applications
- Radiation Therapy and Dosimetry
- Advanced Multi-Objective Optimization Algorithms
- Optimal Experimental Design Methods
- Statistical Methods in Clinical Trials
- Carbon and Quantum Dots Applications
- Advanced Nanomaterials in Catalysis
- Spinal Cord Injury Research
- Nerve injury and regeneration
- Advanced Neural Network Applications
- Pulsars and Gravitational Waves Research
- Molecular Sensors and Ion Detection
- Wound Healing and Treatments
- Radiomics and Machine Learning in Medical Imaging
- Mesenchymal stem cell research
- Product Development and Customization
- Gaussian Processes and Bayesian Inference
- Graphene and Nanomaterials Applications
- Gamma-ray bursts and supernovae
- Prostate Cancer Diagnosis and Treatment
- Geophysics and Gravity Measurements
- CCD and CMOS Imaging Sensors
- COVID-19 Clinical Research Studies
KTH Royal Institute of Technology
2020-2023
Tsinghua University
2023
Zhejiang University
2021-2022
RaySearch Laboratories (Sweden)
2020-2022
China Pharmaceutical University
2021
Jiangxi Normal University
2012-2018
Nankai University
2012
Abstract Following severe spinal cord injury (SCI), dysregulated neuroinflammation causes neuronal and glial apoptosis, resulting in scar cystic cavity formation during wound healing ultimately the of an atrophic microenvironment that inhibits nerve regrowth. Because this complex dynamic pathophysiology, a systemic solution for scar‐ cavity‐free with remodeling to promote regrowth has rarely been explored. A one‐step is proposed through self‐assembling, multifunctional hydrogel depot...
Abstract Hepatic ischemia‐reperfusion injury (HIRI) is a critical complication after liver surgery that negatively affects surgical outcomes of patients with the end‐stage liver‐related disease. Reactive oxygen species (ROS) are responsible for development and eventually lead to hepatic dysfunction. Selenium‐doped carbon quantum dots (Se‐CQDs) an excellent redox‐responsive property can effectively scavenge ROS protect cells from oxidation. However, accumulation Se‐CQDs in extremely low. To...
We present a framework for robust automated treatment planning using machine learning, comprising scenario-specific dose prediction and mimicking.The scenario pipeline is divided into the of nominal from input image dose, each deep learning model with U-net architecture. By specially developed dose-volume histogram-based loss function, predicted doses are ensured sufficient target coverage despite possibility training data being non-robust. Deliverable plans may then be created by solving...
Vision Transformers (ViTs) mark a revolutionary advance in neural networks with their token mixer's powerful global context capability. However, the pairwise affinity and complex matrix operations limit its deployment on resource-constrained scenarios real-time applications, such as mobile devices, although considerable efforts have been made previous works. In this paper, we introduce CAS-ViT: Convolutional Additive Self-attention Transformers, to achieve balance between efficiency...
Mitochondria, key organelles which keep in tune with energy demands for eukaryotic cells, are firmly associated neurological conditions and post-traumatic rehabilitation. In vivo fluorescence imaging of mitochondria, especially deep tissue penetration, would open a window to investigate the actual context brain. However, depth traditional two-photon mitochondrial is still limited due poor biological compatibility or low absorption cross-sections. A biocompatible mitochondria-targeted...
Abstract In this paper, we extend the general minimum lower‐order confounding (GMC) criterion to case of three‐level designs. First, review relationship between GMC and other criteria. Then introduce an aliased component‐number pattern (ACNP) a via consideration component effects, obtain some results on new criterion. All 27‐run designs, 81‐run designs with factor numbers $n=5,\ldots,20$ 243‐run resolution $IV$ or higher are tabulated. The Canadian Journal Statistics 41: 192–210; 2013 © 2012...
Objective.We propose a semiautomatic pipeline for radiation therapy treatment planning, combining ideas from machine learning-automated planning and multicriteria optimization (MCO).Approach.Using knowledge extracted historically delivered plans, prediction models spatial dose statistics are trained furthermore systematically modified to simulate changes in tradeoff priorities, creating set of differently biased predictions. Based on the predictions, an MCO problem is subsequently...
We demonstrate the application of mixture density networks (MDNs) in context automated radiation therapy treatment planning. It is shown that an MDN can produce good predictions dose distributions as well reflect uncertain decision making associated with inherently conflicting clinical tradeoffs, contrast to deterministic methods previously investigated literature. A two-component Gaussian trained on a set plans for postoperative prostate patients varying extents which rectum sparing was...
We present a method of directly optimizing on deviations in clinical goal values radiation therapy treatment planning. Using new mathematical framework which metrics derived from the dose-volume histogram are regarded as functionals an auxiliary random variable, we able to obtain volume-at-dose and dose-at-volume infinitely differentiable functions dose distribution with easily evaluable function gradients. Motivated by connection risk measures finance, is formalized this framework, also...
Abstract We present a new nonparametric mixture-of-experts model for multivariate regression problems, inspired by the probabilistic k -nearest neighbors algorithm. Using conditionally specified model, predictions out-of-sample inputs are based on similarities to each observed data point, yielding predictive distributions represented Gaussian mixtures. Posterior inference is performed parameters of mixture components as well distance metric using mean-field variational Bayes algorithm...
Purpose: We propose a general framework for quantifying predictive uncertainties of dose-related quantities and leveraging this information in dose mimicking problem the context automated radiation therapy treatment planning. Methods: A three-step pipeline, comprising feature extraction, statistic prediction mimicking, is employed. In particular, features are produced by convolutional variational autoencoder used as inputs previously developed nonparametric Bayesian statistical method,...
Abstract Mesenchymal stem cell (MSC) transplantation was suggested as a promising approach to treat spinal cord injury (SCI). However, the heterogeneity of MSC and lack appropriate delivery methods impede its clinical application. To tackle these challenges, we first generated human MSCs derived from single with great homogeneity batch quality then developed biocompatible injectable hydrogel embed cells severe SCI. In clinically relevant rat SCI model, showed that injection into lesion site...
Hepatic Ischemia-Reperfusion Injury Management Selenium-doped carbon quantum dots (Se-CQDs) with excellent anti-oxidation capability protect hepatocytes from apoptosis by limiting damage oxidative stress. Se-CQDs-lecithin nanoparticles are developed through a supramolecular self-assembly mainly driven noncovalent interactions, which effectively scavenge reactive oxygen species and inhibit the release of inflammatory cytokines, thus exerting beneficial therapeutic efficacy on hepatic...
Abstract Background: Because the COVID-19 pandemic has made comprehensive and profound impacts on world, an accurate prediction of its development trend is significant. In particular, second wave rampant to cause a dramatic increase in cases deaths globally. Methods: Using Eureqa algorithm, we predicted five European countries, including France, Germany, Italy, Spain, UK. We first built models predict daily numbers based data these countries. Based models, new COVID-19. Results: that would...
Purpose: We present a framework for robust automated treatment planning using machine learning, comprising scenario-specific dose prediction and mimicking. Methods: The scenario pipeline is divided into the of nominal from input image dose, each deep learning model with U-net architecture. By specially developed dose-volume histogram-based loss function, predicted doses are ensured sufficient target coverage despite possibility training data being non-robust. Deliverable plans may then be...