-
Computerized Medical Imaging and... Jun 2024Bio-medical image segmentation models typically attempt to predict one segmentation that resembles a ground-truth structure as closely as possible. However, as medical...
BACKGROUND AND OBJECTIVES
Bio-medical image segmentation models typically attempt to predict one segmentation that resembles a ground-truth structure as closely as possible. However, as medical images are not perfect representations of anatomy, obtaining this ground truth is not possible. A surrogate commonly used is to have multiple expert observers define the same structure for a dataset. When multiple observers define the same structure on the same image there can be significant differences depending on the structure, image quality/modality and the region being defined. It is often desirable to estimate this type of aleatoric uncertainty in a segmentation model to help understand the region in which the true structure is likely to be positioned. Furthermore, obtaining these datasets is resource intensive so training such models using limited data may be required. With a small dataset size, differing patient anatomy is likely not well represented causing epistemic uncertainty which should also be estimated so it can be determined for which cases the model is effective or not.
METHODS
We use a 3D probabilistic U-Net to train a model from which several segmentations can be sampled to estimate the range of uncertainty seen between multiple observers. To ensure that regions where observers disagree most are emphasised in model training, we expand the Generalised Evidence Lower Bound (ELBO) with a Constrained Optimisation (GECO) loss function with an additional contour loss term to give attention to this region. Ensemble and Monte-Carlo dropout (MCDO) uncertainty quantification methods are used during inference to estimate model confidence on an unseen case. We apply our methodology to two radiotherapy clinical trial datasets, a gastric cancer trial (TOPGEAR, TROG 08.08) and a post-prostatectomy prostate cancer trial (RAVES, TROG 08.03). Each dataset contains only 10 cases each for model development to segment the clinical target volume (CTV) which was defined by multiple observers on each case. An additional 50 cases are available as a hold-out dataset for each trial which had only one observer define the CTV structure on each case. Up to 50 samples were generated using the probabilistic model for each case in the hold-out dataset. To assess performance, each manually defined structure was matched to the closest matching sampled segmentation based on commonly used metrics.
RESULTS
The TOPGEAR CTV model achieved a Dice Similarity Coefficient (DSC) and Surface DSC (sDSC) of 0.7 and 0.43 respectively with the RAVES model achieving 0.75 and 0.71 respectively. Segmentation quality across cases in the hold-out datasets was variable however both the ensemble and MCDO uncertainty estimation approaches were able to accurately estimate model confidence with a p-value < 0.001 for both TOPGEAR and RAVES when comparing the DSC using the Pearson correlation coefficient.
CONCLUSIONS
We demonstrated that training auto-segmentation models which can estimate aleatoric and epistemic uncertainty using limited datasets is possible. Having the model estimate prediction confidence is important to understand for which unseen cases a model is likely to be useful.
PubMed: 38878632
DOI: 10.1016/j.compmedimag.2024.102403 -
Journal of Environmental Management Jun 2024The war in Ukraine has had a devastating impact on the environment. Military actions have caused the release of hazardous substances into the environment, such as...
The war in Ukraine has had a devastating impact on the environment. Military actions have caused the release of hazardous substances into the environment, such as pollutants and toxic chemicals, that have contaminated the water, soil, and air, posing a threat to both human health and the environment. This has resulted in widespread destruction and contamination of natural habitats and resources and has disrupted wildlife populations and ecosystems. The impacts of military activity on the soils of protected areas are particularly critical, as they are the basis of biotic and landscape diversity and require special management and scientifically based monitoring measures even in peaceful conditions. In this context, this communication paper aims to provide an overview of the impacts of the war on the soils in four Ukrainian protected areas, namely Chornobyl Radiation and Ecological Biosphere Reserve; Desniansko-Starohutskyi National Nature Park; Holosiivskyi National Nature Park, and Hetmanskyi National Nature Park. To address these aspects, this paper combined GIS analysis and secondary data including soil samples obtained during field expeditions, to provide evidence of how ground battles, occupation, terrestrial land mines, and explosions can severely impact the soils. Practical and theoretical implications of the military actions are also discussed.
PubMed: 38878570
DOI: 10.1016/j.jenvman.2024.121399 -
Environment International Jun 2024
PubMed: 38870580
DOI: 10.1016/j.envint.2024.108808 -
PloS One 2024Among many types of cancers, to date, lung cancer remains one of the deadliest cancers around the world. Many researchers, scientists, doctors, and people from other...
Among many types of cancers, to date, lung cancer remains one of the deadliest cancers around the world. Many researchers, scientists, doctors, and people from other fields continuously contribute to this subject regarding early prediction and diagnosis. One of the significant problems in prediction is the black-box nature of machine learning models. Though the detection rate is comparatively satisfactory, people have yet to learn how a model came to that decision, causing trust issues among patients and healthcare workers. This work uses multiple machine learning models on a numerical dataset of lung cancer-relevant parameters and compares performance and accuracy. After comparison, each model has been explained using different methods. The main contribution of this research is to give logical explanations of why the model reached a particular decision to achieve trust. This research has also been compared with a previous study that worked with a similar dataset and took expert opinions regarding their proposed model. We also showed that our research achieved better results than their proposed model and specialist opinion using hyperparameter tuning, having an improved accuracy of almost 100% in all four models.
Topics: Humans; Lung Neoplasms; Machine Learning; Risk Assessment
PubMed: 38870229
DOI: 10.1371/journal.pone.0305035 -
PloS One 2024Infrared target detection is widely used in industrial fields, such as environmental monitoring, automatic driving, etc., and the detection of weak targets is one of the...
Infrared target detection is widely used in industrial fields, such as environmental monitoring, automatic driving, etc., and the detection of weak targets is one of the most challenging research topics in this field. Due to the small size of these targets, limited information and less surrounding contextual information, it increases the difficulty of target detection and recognition. To address these issues, this paper proposes YOLO-ISTD, an improved method for infrared small target detection based on the YOLOv5-S framework. Firstly, we propose a feature extraction module called SACSP, which incorporates the Shuffle Attention mechanism and makes certain adjustments to the CSP structure, enhancing the feature extraction capability and improving the performance of the detector. Secondly, we introduce a feature fusion module called NL-SPPF. By introducing an NL-Block, the network is able to capture richer long-range features, better capturing the correlation between background information and targets, thereby enhancing the detection capability for small targets. Lastly, we propose a modified K-means clustering algorithm based on Distance-IoU (DIoU), called K-means_DIOU, to improve the accuracy of clustering and generate anchors suitable for the task. Additionally, modifications are made to the detection heads in YOLOv5-S. The original 8, 16, and 32 times downsampling detection heads are replaced with 4, 8, and 16 times downsampling detection heads, capturing more informative coarse-grained features. This enables better understanding of the overall characteristics and structure of the targets, resulting in improved representation and localization of small targets. Experimental results demonstrate significant achievements of YOLO-ISTD on the NUST-SIRST dataset, with an improvement of 8.568% in [email protected] and 8.618% in [email protected]. Compared to the comparative models, the proposed approach effectively addresses issues of missed detections and false alarms in the detection results, leading to substantial improvements in precision, recall, and model convergence speed.
Topics: Algorithms; Infrared Rays; Cluster Analysis; Pattern Recognition, Automated
PubMed: 38870195
DOI: 10.1371/journal.pone.0303451 -
Nanomaterials (Basel, Switzerland) Jun 2024This paper investigates the performance of vacuum gate dielectric doping-free carbon nanotube/nanoribbon field-effect transistors (VGD-DL CNT/GNRFETs) via computational...
This paper investigates the performance of vacuum gate dielectric doping-free carbon nanotube/nanoribbon field-effect transistors (VGD-DL CNT/GNRFETs) via computational analysis employing a quantum simulation approach. The methodology integrates the self-consistent solution of the Poisson solver with the mode space non-equilibrium Green's function (NEGF) in the ballistic limit. Adopting the vacuum gate dielectric (VGD) paradigm ensures radiation-hardened functionality while avoiding radiation-induced trapped charge mechanisms, while the doping-free paradigm facilitates fabrication flexibility by avoiding the realization of a sharp doping gradient in the nanoscale regime. Electrostatic doping of the nanodevices is achieved via source and drain doping gates. The simulations encompass MOSFET and tunnel FET (TFET) modes. The numerical investigation comprehensively examines potential distribution, transfer characteristics, subthreshold swing, leakage current, on-state current, current ratio, and scaling capability. Results demonstrate the robustness of vacuum nanodevices for high-performance, radiation-hardened switching applications. Furthermore, a proposal for extrinsic enhancement via doping gate voltage adjustment to optimize band diagrams and improve switching performance at ultra-scaled regimes is successfully presented. These findings underscore the potential of vacuum gate dielectric carbon-based nanotransistors for ultrascaled, high-performance, energy-efficient, and radiation-immune nanoelectronics.
PubMed: 38869587
DOI: 10.3390/nano14110962 -
Nanoscale Advances Jun 2024Due to the unique and excellent optical performance and promising prospect for various photonics applications, cavity-enhanced superfluorescence (CESF) in perovskite...
Due to the unique and excellent optical performance and promising prospect for various photonics applications, cavity-enhanced superfluorescence (CESF) in perovskite quantum dot assembled superstructures has garnered wide attention. However, the stringent requirements and high threshold for achieving CESF limit its further development and application. The high threshold of CESF in quantum dot superstructures is mainly attributed to the low radiation recombination rate of the quantum dot and the unsatisfactory light field limiting the ability of the assembled superstructures originating from low controllability of self-assembly. Herein, we propose a strategy to reduce the threshold of CESF in quantum dot superstructure microcavities from two aspects: facet engineering optimization of quantum dot blocks and controllability improvement of the assembly method. We introduce dodecahedral quantum dots with lower nonradiative recombination, substituting frequently used cubic quantum dots as assembly blocks. Besides, we adopt the micro-emulsion droplet assembly method to obtain spherical perovskite quantum dot superparticles with high packing factors and orderly internal arrangements, which are more controllable and efficient than the conventional solvent-drying methods. Based on the dodecahedral quantum dot superparticles, we realized low-threshold CESF (Pth = 15.6 μJ cm). Our work provides a practical and scalable avenue for realizing low threshold CESF in quantum dot assembled superstructure systems.
PubMed: 38868834
DOI: 10.1039/d4na00188e -
Environmental Politics 2024Solar geoengineering (also known as solar radiation modification) is garnering more attention (and controversy) among media and policymakers in response to the impacts...
Solar geoengineering (also known as solar radiation modification) is garnering more attention (and controversy) among media and policymakers in response to the impacts of climate change. Such debates have become more prominent following the first-ever field trials of stratospheric aerosol injection (SAI) in 2022. How the lay public perceives solar geoengineering remains unclear, however. We use nationally representative samples ( = 3013) in Mexico, United States, and United Kingdom to examine public perceptions of risks and benefits, support, and policy preferences. We also employ an information-framing design that presented individuals with media-style reports on SAI activities differing along three dimensions: location, actor, and scale and purpose. Support for SAI is found to be generally higher in Mexico; perceptions of risks and benefits do not differ between countries. Information about SAI activities has a limited effect. There is evidence that activities conducted by universities receive more support than those by start-up companies.
PubMed: 38868558
DOI: 10.1080/09644016.2023.2301262 -
Frontiers of Optoelectronics Jun 2024In this paper, we first present an experimental demonstration of terahertz radiation pulse generation with energy up to 5 pJ under the electron emission during ultrafast...
In this paper, we first present an experimental demonstration of terahertz radiation pulse generation with energy up to 5 pJ under the electron emission during ultrafast optical discharge of a vacuum photodiode. We use a femtosecond optical excitation of metallic copper photocathode for the generation of ultrashort electron bunch and up to 45 kV/cm external electric field for the photo-emitted electron acceleration. Measurements of terahertz pulses energy as a function of emitted charge density, incidence angle of optical radiation and applied electric field have been provided. Spectral and polarization characteristics of generated terahertz pulses have also been studied. The proposed semi-analytical model and simulations in COMSOL Multiphysics prove the experimental data and allow for the optimization of experimental conditions aimed at flexible control of radiation parameters.
PubMed: 38866994
DOI: 10.1007/s12200-024-00123-5 -
Frontiers in Oncology 2024The ability to dynamically adjust target contours, derived Boolean structures, and ultimately, the optimized fluence is the end goal of online adaptive radiotherapy...
INTRODUCTION
The ability to dynamically adjust target contours, derived Boolean structures, and ultimately, the optimized fluence is the end goal of online adaptive radiotherapy (ART). The purpose of this work is to describe the necessary tests to perform after a software patch installation and/or upgrade for an established online ART program.
METHODS
A patch upgrade on a low-field MR Linac system was evaluated for post-software upgrade quality assurance (QA) with current infrastructure of ART workflow on (1) the treatment planning system (TPS) during the initial planning stage and (2) the treatment delivery system (TDS), which is a TPS integrated into the delivery console for online ART planning. Online ART QA procedures recommended for post-software upgrade include: (1) user interface (UI) configuration; (2) TPS beam model consistency; (3) segmentation consistency; (4) dose calculation consistency; (5) optimizer robustness consistency; (6) CT density table consistency; and (7) end-to-end absolute ART dose and predicted dose measured including interruption testing. Differences of calculated doses were evaluated through DVH and/or 3D gamma comparisons. The measured dose was assessed using an MR-compatible A26 ionization chamber in a motion phantom. Segmentation differences were assessed through absolute volume and visual inspection.
RESULTS
(1) No UI configuration discrepancies were observed. (2) Dose differences on TPS pre-/post-software upgrade were within 1% for DVH metrics. (3) Differences in segmentation when observed were small in general, with the largest change noted for small-volume regions of interest (ROIs) due to partial volume impact. (4) Agreement between TPS and TDS calculated doses was 99.9% using a 2%/2-mm gamma criteria. (5) Comparison between TPS and online ART plans for a given patient plan showed agreement within 2% for targets and 0.6 cc for organs at risk. (6) Relative electron densities demonstrated comparable agreement between TPS and TDS. (7) ART absolute and predicted measured end-to-end doses were within 1% of calculated TDS.
DISCUSSION
An online ART QA program for post-software upgrade has been developed and implemented on an MR Linac system. Testing mechanics and their respective baselines may vary across institutions, but all necessary components for a post-software upgrade QA have been outlined and detailed. These outlined tests were demonstrated feasible for a low-field MR Linac system; however, the scope of this work may be applied and adapted more broadly to other online ART platforms.
PubMed: 38863634
DOI: 10.3389/fonc.2024.1358487