Search

Purpose: To shorten CEST acquisition time by leveraging Z‐spectrum undersampling combined with deep learning for CEST map construction from undersampled Z‐spectra. Methods: Fisher information gain analysis identified optimal frequency offsets (termed "Fisher offsets") for the multi‐pool fitting model, maximizing information gain for the amplitude and the FWHM parameters. These offsets guided initial subsampling levels. A U‐NET, trained on undersampled brain CEST images from 18 volunteers, produced CEST maps at 3 T with varied undersampling levels. Feasibility was first tested using retrospective undersampling at three levels, followed by prospective in vivo undersampling (15 of 53 offsets), reducing scan time significantly. Additionally, glioblastoma grade IV pathology was simulated to evaluate network performance in patient‐like cases. Results: Traditional multi‐pool models failed to quantify CEST maps from undersampled images (structural similarity index [SSIM] <0.2, peak SNR <20, Pearson r <0.1). Conversely, U‐NET fitting successfully addressed undersampled data challenges. The study suggests CEST scan time reduction is feasible by undersampling 15, 25, or 35 of 53 Z‐spectrum offsets. Prospective undersampling cut scan time by 3.5 times, with a maximum mean squared error of 4.4e–4, r = 0.82, and SSIM = 0.84, compared to the ground truth. The network also reliably predicted CEST values for simulated glioblastoma pathology. Conclusion: The U‐NET architecture effectively quantifies CEST maps from undersampled Z‐spectra at various undersampling levels. [ABSTRACT FROM AUTHOR]

The article from the Journal of Medical Radiation Sciences explores the application of deep learning in cancer imaging, specifically focusing on image segmentation. It discusses two studies that demonstrate the effectiveness of U‐Net and convolutional neural networks in improving body composition analysis in CT scans and rectal tumor segmentation in MRI images. The article emphasizes the need for further research to address challenges such as image quality variability across different imaging systems. Deep learning, particularly U‐Net and CNNs, shows promise in enhancing quantitative image analysis in cancer management, but the impact of image quality variations must be carefully considered for accurate and reproducible results. [Extracted from the article]

Handwriting is a natural way to communicate and exchange ideas, but converting handwritten diagrams to application‐specific digital formats requires skill and time. Automatic handwritten document conversion can save time, but diagrams and text require different recognition engines. Since accurate segmentation of handwritten diagrams can improve the accuracy of later diagram recognition steps, the authors propose to solve the problem of segmentation of text and non‐text elements of handwritten diagrams using deep semantic segmentation. The model, DeepDP is a flexible U‐net style architecture that can be tuned in complexity to a level appropriate for a particular dataset and diagram type. Experiments on a public hand‐drawn flowchart dataset and a business process diagram dataset show excellent performance, with a per pixel accuracy of 98.6% on the public flowchart datasets and improvement over the 99.3% text stroke accuracy and 96.6% non‐text stroke accuracy obtained by state of the art methods that use online stroke information. On the smaller offline business process diagram dataset, the method obtains a per‐pixel accuracy of 96.9%. [ABSTRACT FROM AUTHOR]

This study focuses on aerial images in power line inspection, using a small sample size and concentrating on accurately segmenting insulators in images and identifying potential "self‐explode" defects through deep learning methods. The research process consists of four key steps: image segmentation of insulators, identification of small connected regions, data augmentation of original samples, and detection of insulator defects using the YOLO v7 model. In this paper, due to the small sample size, sample expansion is considered first. A sliding window approach is adopted to crop images, increasing the number of training samples. Subsequently, the U‐Net neural network model for semantic segmentation is used to train insulator images, thereby generating preliminary mask images of insulators. Then, through connected region area filtering techniques, smaller connected regions are removed to eliminate small speckles in the predicted mask images, obtaining more accurate insulator mask images. The evaluation metric for image recognition, the dice coefficient, is 93.67%. To target the identification of insulator defects, 35 images with insulator defects from the original samples are augmented. These images are input into the YOLO v7 network for further training, ultimately achieving effective detection of insulator "self‐explode" defects. [ABSTRACT FROM AUTHOR]

In recent years, U‐Net and its variants have gained widespread use in medical image segmentation. One key aspect of U‐Net's design is the skip connection, facilitating the retention of detailed information and leading to finer segmentation results. However, existing research often concentrates on enhancing either the encoder or decoder, neglecting the semantic gap between them, and resulting in suboptimal model performance. In response, we introduce Multi‐Scale Fusion module aimed at enhancing the original skip connections and addressing the semantic gap. Our approach fully incorporates the correlation between outputs from adjacent encoder layers and facilitates bidirectional information exchange across multiple layers. Additionally, we introduce Channel Relation Perception module to guide the fused multi‐scale information for efficient connection with decoder features. These two modules collectively bridge the semantic gap by capturing spatial and channel dependencies in the features, contributing to accurate medical image segmentation. Building upon these innovations, we propose a novel network called MFH‐Net. On three publicly available datasets, ISIC2016, ISIC2017, and Kvasir‐SEG, we perform a comprehensive evaluation of the network. The experimental results show that MFH‐Net exhibits higher segmentation accuracy in comparison with other competing methods. Importantly, the modules we have devised can be seamlessly incorporated into various networks, such as U‐Net and its variants, offering a potential avenue for further improving model performance. [ABSTRACT FROM AUTHOR]

Across various global regions abundant in oil and natural gas reserves, the presence of substantial sub‐surface salt deposits holds significant relevance. Accurate identification of salt domes becomes crucial for enterprises engaged in oil and gas exploration. Our research introduces a precise method for the automatic detection of salt domes, leveraging advanced deep learning architectures such as U‐net, transformers, artificial intelligence generative models and liquid state machines. In comparison with state‐of‐the‐art techniques, our model demonstrates superior performance, achieving a stable and validated 96%$96\%$ intersection over the union metric, indicating high accuracy and robustness. Furthermore, the Dice similarity coefficient attaining 90%$90\%$ underscores the model's proficiency in closely aligning with ground truth across diverse scenarios. This evaluation, conducted on 1000 seismic images, reveals that our proposed architecture is not only comparable but often surpasses existing segmentation models in effectiveness and reliability. [ABSTRACT FROM AUTHOR]

Background: The quality of treatment plans for breast cancer can vary greatly. This variation could be reduced by using dose prediction to automate treatment planning. Our work investigates novel methods for training deep‐learning models that are capable of producing high‐quality dose predictions for breast cancer treatment planning. Purpose: The goal of this work was to compare the performance impact of two novel techniques for deep learning dose prediction models for tangent field treatments for breast cancer. The first technique, a "glowing" mask algorithm, encodes the distance from a contour into each voxel in a mask. The second, a gradient‐weighted mean squared error (MSE) loss function, emphasizes the error in high‐dose gradient regions in the predicted image. Methods: Four 3D U‐Net deep learning models were trained using the planning CT and contours of the heart, lung, and tumor bed as inputs. The dataset consisted of 305 treatment plans split into 213/46/46 training/validation/test sets using a 70/15/15% split. We compared the impact of novel "glowing" anatomical mask inputs and a novel gradient‐weighted MSE loss function to their standard counterparts, binary anatomical masks, and MSE loss, using an ablation study methodology. To assess performance, we examined the mean error and mean absolute error (ME/MAE) in dose across all within‐body voxels, the error in mean dose to heart, ipsilateral lung, and tumor bed, dice similarity coefficient (DSC) across isodose volumes defined by 0%–100% prescribed dose thresholds, and gamma analysis (3%/3 mm). Results: The combination of novel glowing masks and gradient weighted loss function yielded the best‐performing model in this study. This model resulted in a mean ME of 0.40%, MAE of 2.70%, an error in mean dose to heart and lung of −0.10 and 0.01 Gy, and an error in mean dose to the tumor bed of −0.01%. The median DSC at 50/95/100% isodose levels were 0.91/0.87/0.82. The mean 3D gamma pass rate (3%/3 mm) was 93%. Conclusions: This study found the combination of novel anatomical mask inputs and loss function for dose prediction resulted in superior performance to their standard counterparts. These results have important implications for the field of radiotherapy dose prediction, as the methods used here can be easily incorporated into many other dose prediction models for other treatment sites. Additionally, this dose prediction model for breast radiotherapy has sufficient performance to be used in an automated planning pipeline for tangent field radiotherapy and has the major benefit of not requiring a PTV for accurate dose prediction. [ABSTRACT FROM AUTHOR]

Autonomous navigation in agricultural environments is challenged by varying field conditions that arise in arable fields. State‐of‐the‐art solutions for autonomous navigation in such environments require expensive hardware, such as Real‐Time Kinematic Global Navigation Satellite System. This paper presents a robust crop row detection algorithm that withstands such field variations using inexpensive cameras. Existing data sets for crop row detection do not represent all the possible field variations. A data set of sugar beet images was created representing 11 field variations comprised of multiple grow stages, light levels, varying weed densities, curved crop rows, and discontinuous crop rows. The proposed pipeline segments the crop rows using a deep learning‐based method and employs the predicted segmentation mask for extraction of the central crop using a novel central crop row selection algorithm. The novel crop row detection algorithm was tested for crop row detection performance and the capability of visual servoing along a crop row. The visual servoing‐based navigation was tested on a realistic simulation scenario with the real ground and plant textures. Our algorithm demonstrated robust vision‐based crop row detection in challenging field conditions outperforming the baseline. [ABSTRACT FROM AUTHOR]

Low‐light images are captured in environments with minimal lighting, such as nighttime or underwater conditions. These images often suffer from issues like low brightness, poor contrast, lack of detail, and overall darkness, significantly impairing human visual perception and subsequent high‐level visual tasks. Enhancing low‐light images holds great practical significance. Among the various existing methods for Low‐Light Image Enhancement (LLIE), those based on the Retinex theory have gained significant attention. However, despite considerable efforts in prior research, the challenge of Retinex decomposition remains unresolved. In this study, an LLIE network based on the Retinex theory is proposed, which addresses these challenges by integrating attention mechanisms and a U‐Net‐like architecture. The proposed model comprises three modules: the Decomposition module (DECM), the Reflectance Recovery module (REFM), and the Illumination Enhancement module (ILEM). Its objective is to decompose low‐light images based on the Retinex theory and enhance the decomposed reflectance and illumination maps using attention mechanisms and a U‐Net‐like architecture. We conducted extensive experiments on several widely used public datasets. The qualitative results demonstrate that the approach produces enhanced images with superior visual quality compared to the existing methods on all test datasets, especially for some extremely dark images. Furthermore, the quantitative evaluation results based on metrics PSNR, SSIM, LPIPS, BRISQUE, and MUSIQ show the proposed model achieves superior performance, with PSNR and BRISQUE significantly outperforming the baseline approaches, where (PSNR, mean BRISQUE) values of the proposed method and the second best results are (17.14, 17.72) and (16.44, 19.65). Additionally, further experimental results such as ablation studies indicate the effectiveness of the proposed model. [ABSTRACT FROM AUTHOR]

Diagnosing diffuse infiltrative lung diseases (DILD) using high‐resolution computed tomography (HRCT) is challenging, even for expert radiologists, due to the complex and variable image patterns. Moreover, the imbalances among the six key DILD‐related patterns—normal, ground‐glass opacity, reticular opacity, honeycombing, emphysema, and consolidation—further complicate accurate segmentation and diagnosis. This study presents an enhanced U‐Net‐based segmentation technique aimed at addressing these challenges. The primary contribution of our work is the fine‐tuning of the U‐Net model using image‐level labels from 92 HRCT images that include various types of DILDs, such as cryptogenic organizing pneumonia, usual interstitial pneumonia, and nonspecific interstitial pneumonia. This approach helps to correct the imbalance among image patterns, improving the model's ability to accurately differentiate between them. By employing semantic lung segmentation and patch‐level machine learning, the fine‐tuned model demonstrated improved agreement with radiologists' evaluations compared to conventional methods. This suggests a significant enhancement in both segmentation accuracy and inter‐observer consistency. In conclusion, the fine‐tuned U‐Net model offers a more reliable tool for HRCT image segmentation, making it a valuable imaging biomarker for guiding treatment decisions in patients with DILD. By addressing the issue of pattern imbalances, our model significantly improves the accuracy of DILD diagnosis, which is crucial for effective patient care. [ABSTRACT FROM AUTHOR]

Thyroid nodules are a common endocrine system disorder for which accurate ultrasound image segmentation is important for evaluation and diagnosis, as well as a critical step in computer‐aided diagnostic systems. However, the accuracy and consistency of segmentation remains a challenging task due to the presence of scattering noise, low contrast and resolution in ultrasound images. Therefore, we propose a deep learning‐based CAD (computer‐aided diagnosis) method, STU3Net in this paper, aiming at automatic segmentation of thyroid nodules. The method employs a modified Swin Transformer combined with a CNN encoder, which is capable of extracting morphological features and edge details of thyroid nodules in ultrasound images. In decoding through the features for image reconstruction, we introduce a modified three‐layer U‐Net network with cross‐layer connectivity to further enhance image reduction. This cross‐layer connectivity enhances the network's capture and representation of the contained image feature information by creating skip connections between different layers and merging the detailed information of the shallow network with the abstract information of the deeper network. Through comparison experiments with current mainstream deep learning methods on the TN3K and BUSI datasets, we validate the superiority of the STU3Net method in thyroid nodule segmentation performance. The experimental results show that STU3Net outperforms most of the mainstream models on the TN3K dataset, with Dice and IoU reaching 0.8368 and 0.7416, respectively, which are significantly better than other methods. The method demonstrates excellent performance on these datasets and provides radiologists with an effective auxiliary tool to accurately detect thyroid nodules in ultrasound images. [ABSTRACT FROM AUTHOR]

The precise delineation of glioma tumors is of paramount importance for surgical and radiotherapy planning. Presently, the primary drawbacks associated with the manual segmentation approach are its laboriousness and inefficiency. In order to tackle these challenges, a deep learning‐based automatic segmentation technique was introduced to enhance the efficiency of the segmentation process. We proposed ADT‐UNet, an innovative algorithm for segmenting glioma tumors in MR images. ADT‐UNet leveraged attention‐dense blocks and Transformer as its foundational elements. It extended the U‐Net framework by incorporating the dense connection structure and attention mechanism. Additionally, a Transformer structure was introduced at the end of the encoder. Furthermore, a novel attention‐guided multi‐scale feature fusion module was integrated into the decoder. To enhance network stability during training, a loss function was devised that combines Dice loss and binary cross‐entropy loss, effectively guiding the network optimization process. On the test set, the DSC was 0.933, the IOU was 0.878, the PPV was 0.942, and the Sen was 0.938. Ablation experiments conclusively demonstrated that the inclusion of all the three proposed modules led to enhanced segmentation accuracy within the model. The most favorable outcomes were observed when all the three modules were employed simultaneously. The proposed methodology exhibited substantial competitiveness across various evaluation indices, with the three additional modules synergistically complementing each other to collectively enhance the segmentation accuracy of the model. Consequently, it is anticipated that this method will serve as a robust tool for assisting clinicians in auxiliary diagnosis and contribute to the advancement of medical intelligence technology. [ABSTRACT FROM AUTHOR]

Face recognition is an essential feature required for a range of computer vision applications such as security, attendance systems, emotion detection, airport check‐in, and many others. The super‐resolution of subject images is an important and challenging element in numerous scenarios. At times the images are low resolution and need to be processed through super‐resolution techniques to gain more accurate results. For the problem of image super‐resolution, deep learning‐based face recognition systems have been explored in recent years; however, low‐resolution face recognition remains an arduous task. Generative adversarial network (GAN) based models are a promising approach to address this challenge. However, conventional GAN‐based models may generate images that differ significantly from an original high‐resolution image in the test set to the point that the identity of the target face may be changed. To address this shortcoming, we propose a novel U‐Net style generator architecture, where skip‐connections between the encoder and decoder layer can help in preserving the facial characteristics of the input image in the generated image, thus curbing the generator's ability to generate an entirely new image and training it to generate an image more similar in characteristics to the original image. In addition to statistical metrics like structural similarity index measure and Fréchet inception distance, we compute the pixel‐wise distance between the original and model‐generated images to ascertain that our model generates as close to the original images as possible. While we train the model for 4× super‐resolution (64 × 64 images to 256 × 256), our architecture can also be trained for an arbitrary resizing scale. Finally, the number of faces detected over high‐resolution images generated by our model is shown to be higher than state‐of‐the‐art high‐resolution image creation models for face recognition tasks. [ABSTRACT FROM AUTHOR]

We present a parsimonious deep learning weather prediction model to forecast seven atmospheric variables with 3‐hr time resolution for up to 1‐year lead times on a 110‐km global mesh using the Hierarchical Equal Area isoLatitude Pixelization (HEALPix). In comparison to state‐of‐the‐art (SOTA) machine learning (ML) weather forecast models, such as Pangu‐Weather and GraphCast, our DLWP‐HPX model uses coarser resolution and far fewer prognostic variables. Yet, at 1‐week lead times, its skill is only about 1 day behind both SOTA ML forecast models and the SOTA numerical weather prediction model from the European Center for Medium‐Range Weather Forecasts. We report several improvements in model design, including switching from the cubed sphere to the HEALPix mesh, inverting the channel depth of the U‐Net, and introducing gated recurrent units (GRU) on each level of the U‐Net hierarchy. The consistent east‐west orientation of all cells on the HEALPix mesh facilitates the development of location‐invariant convolution kernels that successfully propagate weather patterns across the globe without requiring separate kernels for the polar and equatorial faces of the cube sphere. Without any loss of spectral power after the first 2 days, the model can be unrolled autoregressively for hundreds of steps into the future to generate realistic states of the atmosphere that respect seasonal trends, as showcased in 1‐year simulations. Plain Language Summary: Weather forecasting traditionally relies on numerical weather prediction models that solve physical equations to simulate the evolution of the atmosphere. Such numerical models are compute intensive, and their performance is increasingly challenged by less compute demanding but still highly sophisticated machine learning (ML) approaches. Yet, a downside for many of these new ML models is they tend to drift away from climatology while producing excessively smoothed fields if they are iteratively stepped forward for several months. Here, a parsimonious machine learning model is developed to forecast just seven atmospheric variables that can be stepped forward to give realistic weather patterns over a full year. Despite using at least a factor of 10 less variables than the 67–227 in the best ML models, our model generates 8‐day forecasts with errors that are only a day behind those from state‐of‐the‐art ML forecasts. Our model provides a path toward sub‐seasonal and seasonal forecasting that could potentially improve planning for agriculture, water resources, disaster preparedness, and energy production. Key Points: The model forecasts seven atmospheric variables, an order of magnitude less than that used in state‐of‐the‐art ML weather forecast modelsForecasts are generated on the HEALPix mesh, facilitating the development of location invariant convolution kernelsWithout converging to climatology, the model produces realistic atmospheric states in 365‐day iterative rollouts [ABSTRACT FROM AUTHOR]

Background: In recent years, the increasing prevalence of skin cancers, particularly malignant melanoma, has become a major concern for public health. The development of accurate automated segmentation techniques for skin lesions holds immense potential in alleviating the burden on medical professionals. It is of substantial clinical importance for the early identification and intervention of skin cancer. Nevertheless, the irregular shape, uneven color, and noise interference of the skin lesions have presented significant challenges to the precise segmentation. Therefore, it is crucial to develop a high‐precision and intelligent skin lesion segmentation framework for clinical treatment. Methods: A precision‐driven segmentation model for skin cancer images is proposed based on the Transformer U‐Net, called BiADATU‐Net, which integrates the deformable attention Transformer and bidirectional attention blocks into the U‐Net. The encoder part utilizes deformable attention Transformer with dual attention block, allowing adaptive learning of global and local features. The decoder part incorporates specifically tailored scSE attention modules within skip connection layers to capture image‐specific context information for strong feature fusion. Additionally, deformable convolution is aggregated into two different attention blocks to learn irregular lesion features for high‐precision prediction. Results: A series of experiments are conducted on four skin cancer image datasets (i.e., ISIC2016, ISIC2017, ISIC2018, and PH2). The findings show that our model exhibits satisfactory segmentation performance, all achieving an accuracy rate of over 96%. Conclusion: Our experiment results validate the proposed BiADATU‐Net achieves competitive performance supremacy compared to some state‐of‐the‐art methods. It is potential and valuable in the field of skin lesion segmentation. [ABSTRACT FROM AUTHOR]

Many agricultural areas of the western United States and other parts of the world practice irrigation using a variety of irrigation methods. Maps of irrigation methods are needed but existing technologies are often unable to distinguish between different irrigation methods when they co‐exist on the same landscape. In this study, we develop a deep learning irrigation methods mapping tool for broad scale application. The technique uses a U‐Net model trained on Landsat 5‐ and 8‐derived input images. Training data consisted in irrigation method classified as Flood, Sprinkler or Other on agricultural fields from the Utah Water Related Land Use data set and additional labeling in selected areas of southern Idaho. An ensemble of 10 trained models had an overall accuracy of 0.78. Precision for Flood, Sprinkler and Other were 0.73, 0.82, and 0.80 while recall values were 0.75, 0.74, and 0.84 respectively. Model performance was generally stable throughout the training years but varied by areas. The best performance was obtained in regions with uniform irrigation method across large patches while small fields of contrasting irrigation method with their surroundings were inadequately predicted. Model prediction in an irrigated watershed of southern Idaho for 2006, 2011, 2013, and 2016 were consistent with previously published survey data. This methodology provides a tool for water resource managers to estimate irrigation methods in agricultural watersheds where natural precipitation is low during the growing season and irrigation methods include center pivots, wheel lines and flood irrigation. Plain Language Summary: Many agricultural areas of the western United States practice irrigation using a variety of irrigation methods. Irrigation methods can be classified into 3 main groups: surface (or flood), sprinkler systems and micro‐irrigation systems. Flood and sprinkler irrigation account for 90% of irrigated areas in the United States but impact water resources differently. Flood irrigation has been associated with many adverse effects on water quality whereas sprinkler systems are promoted as improved irrigation alternatives to preserve water quantity and quality. Maps of irrigation methods are needed to improve assessment of irrigation methods on water quantity and quality. In this study, we develop an irrigation methods mapping tool by training a deep learning model on publicly available satellite imagery. The model was trained on the Utah Water Related Land Use data set and additional data from southern Idaho. The trained model correctly predicted irrigation method over 78% of the test area. This methodology provides a tool for water resource managers to estimate irrigation methods in agricultural watersheds where natural precipitation is low during the growing season and irrigation methods include center pivots, wheel lines and flood irrigation. Key Points: A deep learning model was developed to predict the type of irrigation (Flood, Sprinkler or Other) used in areas of the northwestern USAOverall the model predicted the right type of irrigation with an accuracy of 78%This tool has application in other irrigated agricultural areas of the semi‐arid northwest where irrigation methods are varied [ABSTRACT FROM AUTHOR]

In this study, we employed a method integrating optical coherence tomography (OCT) with the U‐Net and Visual Geometry Group (VGG)‐Net frameworks within a convolutional neural network for quantitative characterization of the three dimensional whole blood during the dynamic coagulation process. VGG‐Net architecture for the identification of blood droplets across three distinct coagulation stages including drop, gelation, and coagulation achieves an accuracy of up to 99%. In addition, the U‐Net architecture demonstrated proficiency in effectively segmenting uncoagulated and coagulated portions of whole blood, as well as the background. Notably, parameters such as volume of uncoagulated and coagulated segments of the whole blood were successfully employed for the precise quantification of the coagulation process, which indicates well for the potential of future clinical diagnostics and analyses. [ABSTRACT FROM AUTHOR]

Accurate segmentation of retinal vessels is crucial for the early diagnosis and treatment of eye diseases, for example, diabetic retinopathy, glaucoma, and macular degeneration. Due to the intricate structure of retinal vessels, it is essential to extract their features with precision for the semantic segmentation of medical images. In this study, an improved deep learning neural network was developed with a focus on feature extraction based on the U‐Net structure. The enhanced U‐Net combines the architecture of convolutional neural networks (CNNs) with SE blocks (squeeze‐and‐excitation blocks) to adaptively extract image features after each U‐Net encoder's convolution. This approach aids in suppressing nonvascular regions and highlighting features for specific segmentation tasks. The proposed method was trained and tested on the DRIVECHASE_DB1 and STARE datasets. As a result, the proposed model had an algorithmic accuracy, sensitivity, specificity, Dice coefficient (Dc), and Matthews correlation coefficient (MCC) of 95.62/0.9853/0.9652, 0.7751/0.7976/0.7773, 0.9832/0.8567/0.9865, 82.53/87.23/83.42, and 0.7823/0.7987/0.8345, respectively, outperforming previous methods, including UNet++, attention U‐Net, and ResUNet. The experimental results demonstrated that the proposed method improved the retinal vessel segmentation performance. [ABSTRACT FROM AUTHOR]

In the development of medical devices and surgical robot systems, animal models are often used for evaluation, necessitating accurate organ segmentation. Deep learning‐based image segmentation provides a solution for automatic and precise organ segmentation. However, a significant challenge in this approach arises from the limited availability of training data for animal models. In contrast, human medical image datasets are readily available. To address this imbalance, this study proposes a fine‐tuning approach that combines a limited set of animal model images with a comprehensive human image dataset. Various postprocessing algorithms were applied to ensure that the segmentation results met the positioning requirements for the evaluation of a medical robot under development. As one of the target applications, magnetic resonance images were used to determine the position of the dog's prostate, which was then used to determine the target location of the robot under development. The MSD TASK5 dataset was used as the human dataset for pretraining, which involved a modified U‐Net network. Ninety‐nine pretrained backbone networks were tested as encoders for U‐Net. The cross‐training validation was performed using the selected network backbone. The highest accuracy, with an IoU score of 0.949, was achieved using the independent validation set from the MSD TASK5 human dataset. Subsequently, fine‐tuning was performed using a small set of dog prostate images, resulting in the highest accuracy of an IoU score of 0.961 across different cross‐validation groups. The processed results demonstrate the feasibility of the proposed approach for accurate prostate segmentation. [ABSTRACT FROM AUTHOR]

Gullies and ravines are common landforms in raised marine fine‐grained deposits in Norway. Gullies in marine clay are significant landforms indicative of soil erosion and natural hazards and are of high conservation value. As a result of the substantial impact of human intervention over the past century, marine clay gullies are now red‐listed. To monitor the condition of these landforms, we need to improve our understanding of their spatial extent, complexity and morphology. We explore the applicability of automated approaches that use a methodology of combining deep learning (DL), fully convolutional neural networks (FCNNs) and an unmodified U‐Net model with ArcPy libraries and ground truth data to derive a high‐resolution map of gullies in raised marine fine‐grained deposits. Predictors used comprise solely terrain derivatives to broaden the usage of the pre‐trained model to other regions. Our best model achieved a precision score of 0.82 and a recall of 0.75. We find that our pre‐trained model can successfully predict gullies, also in blind‐test areas. The model performs better in regions with similar geological settings, scoring a length‐weighted overlap of >70% with reference datasets. The novelty of this study is that we demonstrate that the model's applicability for mapping routines increases when we post‐process the predictions by eliminating noise, especially by using the predictions derived from ensembled models. We, therefore, conclude that the pre‐trained models can effectively be used to supplement the geomorphological mapping of marine clay gullies in Norway. The outcome of this research contributes towards mapping the spatial extent and condition of red‐listed landforms in Norway, as well as the development of monitoring systems for future landscape change. [ABSTRACT FROM AUTHOR]

The internal variability pertains to fluctuations originating from processes inherent to the climate component and their mutual interactions. On the other hand, forced variability delineates the influence of external boundary conditions on the physical climate system. A methodology is formulated to distinguish between internal and forced variability within the surface air temperature. The noise‐to‐noise approach is employed for training a neural network, drawing an analogy between internal variability and image noise. A large training data set is compiled using surface air temperature data spanning from 1901 to 2020, obtained from an ensemble of Atmosphere‐Ocean General Circulation Model simulations. The neural network utilized for training is a U‐Net, a widely adopted convolutional network primarily designed for image segmentation. To assess performance, comparisons are made between outputs from two single‐model initial‐condition large ensembles, the ensemble mean, and the U‐Net's predictions. The U‐Net reduces internal variability by a factor of four, although notable discrepancies are observed at the regional scale. While demonstrating effective filtering of the El Niño Southern Oscillation, the U‐Net encounters challenges in capturing the changes in the North Atlantic Ocean. This methodology holds potential for extension to other physical variables, facilitating insights into the climate change triggered by external forcings over the long term. Plain Language Summary: To anticipate future climate change, it is crucial to detect and understand the impacts of human activities. However, distinguishing the effects of anthropogenic forcing from natural climate variations in observational data is challenging. Natural climate variability, known as internal variability, arises from the chaotic nature of atmospheric and oceanic circulation, and from the interactions among the ocean, atmosphere, and land. Here, a novel approach is introduced to distinguish the changes caused by human activities from internal variability. It is applied to the surface air temperature evolution from 1901 to 2020. This method uses an artificial neural network designed to separate the internal from the human‐induced variability. An unprecedented number of climate model simulations are used, enabling precise estimation of human‐forced variability in these climate models. The spatio‐temporal variations are distinguished by applying a well‐known methodology previously used to remove noise from images. The method's performance is evaluated, revealing errors regarding the internal variability that are typically one‐fourth of the actual variations. Regions with important internal variability or with low agreement among models exhibit the largest errors. Overall, the skills are comparable to other existing approaches, but improvements are anticipated. Key Points: We present a new method to separate the forced and internal variability of the surface air temperatureWe utilize a U‐Net trained with global climate models outputs and implement a noise to noise methodology to eliminate internal variabilityThe results are assessed through the utilization of very large ensemble simulations of two distinct climate models [ABSTRACT FROM AUTHOR]

Monitoring fetal growth in utero is crucial to anomaly diagnosis. However, current computer‐vision models struggle to accurately assess the key metrics (i.e., head circumference and occipitofrontal and biparietal diameters) from ultrasound images, largely owing to a lack of training data. Mitigation usually entails image augmentation (e.g., flipping, rotating, scaling, and translating). Nevertheless, the accuracy of our task remains insufficient. Hence, we offer a U‐Net fetal head measurement tool that leverages a hybrid Dice and binary cross‐entropy loss to compute the similarity between actual and predicted segmented regions. Ellipse‐fitted two‐dimensional ultrasound images acquired from the HC18 dataset are input, and their lower feature layers are reused for efficiency. During regression, a novel region of interest pooling layer extracts elliptical feature maps, and during segmentation, feature pyramids fuse field‐layer data with a new scale attention method to reduce noise. Performance is measured by Dice similarity, mean pixel accuracy, and mean intersection‐over‐union, giving 97.90%, 99.18%, and 97.81% scores, respectively, which match or outperform the best U‐Net models. [ABSTRACT FROM AUTHOR]

Background: While videostroboscopy is recognized as the most popular approach for investigating vocal fold function, evaluating the numerical values, such as the membranous glottal gap area, remains too time consuming for clinical applications. Methods: We used a total of 2507 videostroboscopy images from 137 patients and developed five U‐Net‐based deep‐learning image segmentation models for automatic masking of the membranous glottal gap area. To further validate the models, we used another 410 images from 41 different patients. Results: During development, all five models exhibited acceptable and similar metrics. While the VGG19 U‐Net had a long inference time of 1654 ms, the other four models had more practical inference times, ranging from 16 to 138 ms. During further validation, Efficient U‐Net demonstrated the highest intersection over union of 0.8455, the highest Dice coefficient of 0.9163, and the lowest Hausdorff distance of 1.5626. The normalized membranous glottal gap area index was also calculated and validated. Efficient U‐Net and VGG19 U‐Net exhibited the lowest mean squared errors (3.5476 and 3.3842) and the lowest mean absolute errors (1.8835 and 1.8396). Conclusions: Automatic segmentation of the membranous glottal gap area can be achieved through U‐net‐based architecture. Considering the segmentation quality and speed, Efficient U‐Net is a reasonable choice for this task, while the other four models remain valuable competitors. The models' masked area enables possible calculation of the normalized membranous glottal gap area and analysis of the glottal area waveform, revealing promising clinical applications for this model. Level of Evidence: NA Laryngoscope, 134:2835–2843, 2024 [ABSTRACT FROM AUTHOR]

Photoacoustic (PA) imaging is hybrid imaging modality with good optical contrast and spatial resolution. Portable, cost‐effective, smaller footprint light emitting diodes (LEDs) are rapidly becoming important PA optical sources. However, the key challenge faced by the LED‐based systems is the low light fluence that is generally compensated by high frame averaging, consequently reducing acquisition frame‐rate. In this study, we present a simple deep learning U‐Net framework that enhances the signal‐to‐noise ratio (SNR) and contrast of PA image obtained by averaging low number of frames. The SNR increased by approximately four‐fold for both in‐class in vitro phantoms (4.39 ± 2.55) and out‐of‐class in vivo models (4.27 ± 0.87). We also demonstrate the noise invariancy of the network and discuss the downsides (blurry outcome and failure to reduce the salt & pepper noise). Overall, the developed U‐Net framework can provide a real‐time image enhancement platform for clinically translatable low‐cost and low‐energy light source‐based PA imaging systems. [ABSTRACT FROM AUTHOR]

The goal of image enhancement methods is to improve image's quality. The efficacy of U‐net is evident through its extensive utilization across various significant image modalities, involving computed tomography (CT) scans, magnetic resonance imaging, X‐rays, and microscopy. In this study, we provided a novel and efficient strategy to improve lung CT images based on segmentation using U‐Net architecture. Subsequently, contrast enhancement was performed using adaptive histogram equalization and dark channel prior methods. Finally, the lightness of the lung CT image was enhanced using nonlinear mapping. The contrast enhancement performance of the suggested method is quantified by various measures like the average gradient, mean of the local standard deviation, contrast enhancement measure, and structural similarity index. The performance of the suggested method is compared against other methods, and the results indicate that the suggested method achieves better quality measures of 23.4907, 55.20341, 0.961674, and 0.4143 for the four performance metrics. [ABSTRACT FROM AUTHOR]

Objective: Application of an optical method for the identification of antiresorptive drug‐related osteonecrosis of the jaw (ARONJ). Methods: We introduce shifted‐excitation Raman difference spectroscopy followed by U‐Net deep neural network refinement to determine bone tissue viability. The obtained results are validated through established histological methods. Results: Discrimination of osteonecrosis from physiological tissues was evaluated at 119 distinct measurement loci in 40 surgical specimens from 28 patients. Mean Raman spectra were refined from 11,900 raw spectra, and characteristic peaks were assigned to their respective molecular origin. Then, following principal component and linear discriminant analyses, osteonecrotic lesions were distinguished from physiological tissue entities, such as viable bone, with a sensitivity, specificity, and overall accuracy of 100%. Moreover, bone mineral content, quality, maturity, and crystallinity were quantified, revealing an increased mineral‐to‐matrix ratio and decreased carbonate‐to‐phosphate ratio in ARONJ lesions compared to physiological bone. Conclusion: The results demonstrate feasibility with high classification accuracy in this collective. The differentiation was determined by the spectral features of the organic and mineral composition of bone. This merely optical, noninvasive technique is a promising candidate to ameliorate both the diagnosis and treatment of ARONJ in the future. [ABSTRACT FROM AUTHOR]

Purpose: Subject movement during the MR examination is inevitable and causes not only image artifacts but also deteriorates the homogeneity of the main magnetic field (B0), which is a prerequisite for high quality data. Thus, characterization of changes to B0, for example induced by patient movement, is important for MR applications that are prone to B0 inhomogeneities. Methods: We propose a deep learning based method to predict such changes within the brain from the change of the head position to facilitate retrospective or even real‐time correction. A 3D U‐net was trained on in vivo gradient‐echo brain 7T MRI data. The input consisted of B0 maps and anatomical images at an initial position, and anatomical images at a different head position (obtained by applying a rigid‐body transformation on the initial anatomical image). The output consisted of B0 maps at the new head positions. We further fine‐trained the network weights to each subject by measuring a limited number of head positions of the given subject, and trained the U‐net with these data. Results: Our approach was compared to established dynamic B0 field mapping via interleaved navigators, which suffer from limited spatial resolution and the need for undesirable sequence modifications. Qualitative and quantitative comparison showed similar performance between an interleaved navigator‐equivalent method and proposed method. Conclusion: It is feasible to predict B0 maps from rigid subject movement and, when combined with external tracking hardware, this information could be used to improve the quality of MR acquisitions without the use of navigators. [ABSTRACT FROM AUTHOR]

In recent years, deep learning algorithms have achieved remarkable results in medical image segmentation. These networks with an enormous number of parameters often encounter challenges in handling image boundaries and details, which may result in suboptimal segmentation results. To solve the problem, we develop atrous spatial pyramid pooling (ASPP) and combine it with the squeeze‐and‐excitation block (SE block), as well as present the PS module, which employs a broader and multiscale receptive field at the network's bottom to obtain more detailed semantic information. We also propose the local guided block (LG block) and also its combination with the SE block to form the LS block, which can obtain more abundant local features in the feature map, so that more edge information can be retained in each down sampling process, thereby improving the performance of boundary segmentation. We propose PLU‐Net and integrate our PS module and LS block into U‐Net. We put our PLU‐Net to the test on three benchmark datasets, and the results show that by fewer parameters and FLOPs, it outperforms on medical semantic segmentation tasks. [ABSTRACT FROM AUTHOR]

This study proposes an improved U‐Net model to address the issues of large semantic differences in skip connections and insufficient utilization of cross‐channel information in magnetic resonance imaging (MRI) images leading to inaccurate segmentation of brain tumor regions in the field of brain tumor segmentation. Firstly, by adding a deep residual module to alter the receptive field, the network's ability to learn tumor information is enhanced. Secondly, a dual attention mechanism was established using Bayesian weighting technology, achieving multi‐channel and multi‐scale feature fusion, and improving the model's learning and extraction of brain tumor boundary information. Finally, the tumor features extracted from different patterns are concatenated through skip connections, effectively integrating feature information from different levels and scales, and reducing semantic differences. We evaluated the performance of the proposed model on the BraTS2018 and BraTS2019 brain tumor image segmentation datasets. The experimental results showed that for the BraTS2018 dataset, the model improved the average dice score by 12.8%, the average sensitivity by 10.4%, and the average Hausdorff Distance by 5.75 compared to traditional U‐Net. On the BraTS2019 dataset, three indicators improved by 12.6%, 11.2%, and 7.46, respectively. The experimental results show that the proposed improved U‐Net model can improve the segmentation performance of brain tumor MRI images without increasing computational time. [ABSTRACT FROM AUTHOR]

Microwave imaging has been widely used in stroke diagnosis as a non‐invasive, ionizing radiation‐free imaging method. However, the electromagnetic inverse scattering problems of microwave imaging are nonlinear and ill‐posed. To further improve the accuracy of microwave imaging algorithms to identify and reconstruct stroke regions. A novel Encoder‐Decoder and Modified U‐Net (ED‐MUNET) network for microwave imaging of stroke is proposed. The newly proposed ED‐MUNET method accomplishes the initial imaging from scattered data to stroke images through an encoder–decoder in the first step. In the second step, the feature information of the reconstructed image is extracted through the modified U‐Net network, and the reconstruction from the rough stroke image to the high‐resolution image is realized. The second step avoids black‐box operations and improves image accuracy. ED‐MUNET has fewer artifacts, the relative reconstruction error is less than 0.05, and the reconstructed images are clearer, according to the comparative experiments with other networks. The experimental results showed the superiority of the proposed method for reconstructing stroke images. [ABSTRACT FROM AUTHOR]

Background: Despite extensive efforts to obtain accurate segmentation of magnetic resonance imaging (MRI) scans of a head, it remains challenging primarily due to variations in intensity distribution, which depend on the equipment and parameters used. Purpose: The goal of this study is to evaluate the effectiveness of an automatic segmentation method for head MRI scans using a multistep Dense U‐Net (MDU‐Net) architecture. Methods: The MDU‐Net‐based method comprises two steps. The first step is to segment the scalp, skull, and whole brain from head MRI scans using a convolutional neural network (CNN). In the first step, a hybrid network is used to combine 2.5D Dense U‐Net and 3D Dense U‐Net structure. This hybrid network acquires logits in three orthogonal planes (axial, coronal, and sagittal) using 2.5D Dense U‐Nets and fuses them by averaging. The resultant fused probability map with head MRI scans then serves as the input to a 3D Dense U‐Net. In this process, different ratios of active contour loss and focal loss are applied. The second step is to segment the cerebrospinal fluid (CSF), white matter, and gray matter from extracted brain MRI scans using CNNs. In the second step, the histogram of the extracted brain MRI scans is standardized and then a 2.5D Dense U‐Net is used to further segment the brain's specific tissues using the focal loss. A dataset of 100 head MRI scans from an OASIS‐3 dataset was used for training, internal validation, and testing, with ratios of 80%, 10%, and 10%, respectively. Using the proposed approach, we segmented the head MRI scans into five areas (scalp, skull, CSF, white matter, and gray matter) and evaluated the segmentation results using the Dice similarity coefficient (DSC) score, Hausdorff distance (HD), and the average symmetric surface distance (ASSD) as evaluation metrics. We compared these results with those obtained using the Res‐U‐Net, Dense U‐Net, U‐Net++, Swin‐Unet, and H‐Dense U‐Net models. Results: The MDU‐Net model showed DSC values of 0.933, 0.830, 0.833, 0.953, and 0.917 in the scalp, skull, CSF, white matter, and gray matter, respectively. The corresponding HD values were 2.37, 2.89, 2.13, 1.52, and 1.53 mm, respectively. The ASSD values were 0.50, 1.63, 1.28, 0.26, and 0.27 mm, respectively. Comparing these results with other models revealed that the MDU‐Net model demonstrated the best performance in terms of the DSC values for the scalp, CSF, white matter, and gray matter. When compared with the H‐Dense U‐Net model, which showed the highest performance among the other models, the MDU‐Net model showed substantial improvements in the HD view, particularly in the gray matter region, with a difference of approximately 9%. In addition, in terms of the ASSD, the MDU‐Net model outperformed the H‐Dense U‐Net model, showing an approximately 7% improvements in the white matter and approximately 9% improvements in the gray matter. Conclusion: Compared with existing models in terms of DSC, HD, and ASSD, the proposed MDU‐Net model demonstrated the best performance on average and showed its potential to enhance the accuracy of automatic segmentation for head MRI scans. [ABSTRACT FROM AUTHOR]

Background: The delineation of brain arteriovenous malformations (bAVMs) is crucial for subsequent treatment planning. Manual segmentation is time‐consuming and labor‐intensive. Applying deep learning to automatically detect and segment bAVM might help to improve clinical practice efficiency. Purpose: To develop an approach for detecting bAVM and segmenting its nidus on Time‐of‐flight magnetic resonance angiography using deep learning methods. Study Type: Retrospective. Subjects: 221 bAVM patients aged 7–79 underwent radiosurgery from 2003 to 2020. They were split into 177 training, 22 validation, and 22 test data. Field Strength/Sequence: 1.5 T, Time‐of‐flight magnetic resonance angiography based on 3D gradient echo. Assessment: The YOLOv5 and YOLOv8 algorithms were utilized to detect bAVM lesions and the U‐Net and U‐Net++ models to segment the nidus from the bounding boxes. The mean average precision, F1, precision, and recall were used to assess the model performance on the bAVM detection. To evaluate the model's performance on nidus segmentation, the Dice coefficient and balanced average Hausdorff distance (rbAHD) were employed. Statistical Tests: The Student's t‐test was used to test the cross‐validation results (P < 0.05). The Wilcoxon rank test was applied to compare the median for the reference values and the model inference results (P < 0.05). Results: The detection results demonstrated that the model with pretraining and augmentation performed optimally. The U‐Net++ with random dilation mechanism resulted in higher Dice and lower rbAHD, compared to that without that mechanism, across varying dilated bounding box conditions (P < 0.05). When combining detection and segmentation, the Dice and rbAHD were statistically different from the references calculated using the detected bounding boxes (P < 0.05). For the detected lesions in the test dataset, it showed the highest Dice of 0.82 and the lowest rbAHD of 5.3%. Data Conclusion: This study showed that pretraining and data augmentation improved YOLO detection performance. Properly limiting lesion ranges allows for adequate bAVM segmentation. Level of Evidence: 4 Technical Efficacy Stage: 1 [ABSTRACT FROM AUTHOR]

Convective cold pools (CPs) mediate interactions between convective rain cells and help organize thunderstorm clusters, in particular mesoscale convective systems and extreme rainfall events. Unfortunately, the observational detection of CPs on a large scale has been hampered by the lack of relevant near‐surface data. Unlike numerical studies, where fields, such as virtual temperature or wind, are available at high resolution and frequently used to detect CPs, observational studies mainly identify CPs based on surface time series, for example, from weather stations or research vessels—thus limiting studies to a regional scope. To expand to a global scope, we here develop and evaluate a methodology for CP detection that relies exclusively on data with (a) global availability and (b) high spatiotemporal resolution. We trained convolutional neural networks to segment CPs in high‐resolution cloud‐resolving simulation output by deliberately restricting ourselves to only cloud top temperature and rainfall fields. Apart from simulations, such data are readily available from geostationary satellites that fulfill both (a) and (b). The networks employ a U‐Net architecture, popular with image segmentation, where spatial correlations at various scales must be learned. Despite the restriction imposed, the trained networks systematically identify CP pixels. Looking ahead, our methodology aims to reliably detect CPs over tropical land from space‐borne sensors on a global scale. As it also provides information on the spatial extent and the relative positioning of CPs over time, our method may unveil the role of CPs in convective organization. Plain Language Summary: Cold pools come about when rain evaporates underneath thunderstorm rain cells. Such cold pools are colder and thus denser than the surrounding air, which makes them flow over the surface. The associated gust fronts can often be felt when observing thunderstorm clouds in nature, as strong, but relatively short‐lasting, winds occurring near the thunderstorm downpour. Such cold pools can bring about clumps of thunderstorms, which can deliver large quantities of rainfall within areas of approximately 100 km across. Thus, detecting cold pools on these and larger scales is important, but so far difficult due to the challenge of observing the air currents underneath clouds from satellite. We here present a method that may be able to do just that, namely detecting cold pools from above, by identifying signatures left behind in cloud patterns, for example, arc‐like shapes along the gust front. We demonstrate the algorithms capabilities using high‐resolution simulation data, where we are able to "know" the true result. We use an artificial intelligence framework to carry out this statistical task. We suggest that our method should be applicable to satellite data and thus give new insight into cloud organization at large scales. Key Points: A neural network is trained to detect cold pools from cloud top temperature and precipitationThe method works reliably and a pseudo‐3D approach, which includes dynamic cold pool evolution, performs bestThe method offers new perspectives for the cold pool detection from geostationary satellite observations over tropical land [ABSTRACT FROM AUTHOR]

With rapid developments in convolutional neural networks for image processing, deep learning methods based on pixel classification have been extensively applied in medical image segmentation. One popular strategy for such tasks is the encoder‐decoder‐based U‐Net architecture and its variants. Most segmentation methods based on fully convolutional networks will cause the loss of spatial and contextual information due to continuous pooling operations or strided convolution when decreasing image resolution, and make less use of contextual information and global information under different receptive fields. To overcome this shortcoming, this paper proposes a novel structure called RAAU‐Net. In our proposed RAAU‐Net structure, which is a modified U‐shaped architecture, we aim to capture high‐level information while preserving spatial information and focusing on the regions of interest. RAAU‐Net comprises three main components: a feature encoder module that utilizes a pre‐trained ResNet‐18 model as a fixed feature extractor, a multi‐receptive field extraction module that we developed, and a feature decoder module. We have tested our method on several 2D medical image segmentation tasks such as retinal nerve, breast tumor, skin lesion, lung, gland, and polyp segmentation. All the indexes of the model reached the best in the dataset of skin lesions, in which Accuracy, Precision, IoU, Recall, and Dice Score were 3.26%, 5.42%, 9.92%, 6.52%, and 5.95% higher than UNet. [ABSTRACT FROM AUTHOR]

Precise and unambiguous segmentation of pulmonary nodules from the CT images is imperative for a CAD framework implementation delineated for the prognosis of lung cancer. Lung nodule segmentation is an appealing research discipline for accurate dismemberment of lung cancer but the irregularity in shades, contours, and compositions, and the affinity between the tumors and the neighboring regions makes it an arduous task. This paper proffers a series atrous convolution enhanced U‐Net which uses a series of concatenated dilated convolution blocks after every stage in the encoder and decoder path. Our approach helps in obtaining the quintessential components from the feature maps, in addition to the absolute convergence of the model. It is largely assessed on the publicly accessible LIDC‐IDRI dataset. The average Dice Similarity Coefficient (DSC) obtained is 81.10% with an Intersection over Union (IoU/ Jaccard Index) of 72.24%. Exploratory outcomes prove that our architecture achieves ameliorate performance. [ABSTRACT FROM AUTHOR]

Early diagnosis of multiple sclerosis (MS) through the delineation of lesions in the brain magnetic resonance imaging is important in preventing the deteriorating condition of MS. This study aims to develop a modified U‐Net model for automating lesions segmentation in MS more accurately. The proposed modified U‐Net uses residual dense blocks to replace the standard convolutional stacks and incorporates three axes (axial, sagittal, and coronal) of 2D slice images as input. Furthermore, a custom fusion method is also introduced for merging the predicted lesions from different axes. The model was implemented on ISBI2015 and OpenMS data sets. On ISBI2015, the proposed model achieves the best overall score of 93.090% and DSC of 0.857 on the OpenMS data set. [ABSTRACT FROM AUTHOR]

Automation of the various steps of radiotherapy is gaining importance nowadays. Predicting the amount of radiation dose received by the tumor and nearby organs is one among them. Many deep learning architectures that predict the 3D dose distribution images from corresponding CT and contour images have been proposed in the literature. However, a detailed investigation has yet to be done on the significance of input images (CT and contour) in predicting dose distributions. This study introduces a novel two‐stage deep learning framework using a transfer learning technique for the same domain. The main objective of this approach is to accurately extract valuable information from the CT and contour images to determine the amount of radiation dose received by the organs. Training and testing are performed on the public dataset—OpenKBP, consisting of 340 oropharyngeal cancer patient data. The model performance is evaluated using the metrics dose score and DVH score. The proposed model outperforms the single‐stage deep learning models by 0.34% for the DVH score and 0.14% for the dose score. While comparing the mean dose difference between the predicted and actual dose values for each organ, the proposed model shows better performance in almost all cases. The results imply that medical professionals can utilize the predicted dose distributions to aid the optimization process in the treatment planning systems. [ABSTRACT FROM AUTHOR]

Summary: Manual investigation of blood cell count is sometimes erroneous due to interoperability error, fatigue error, requiring expert skill and time consuming too. In particular, investigation of white blood cell (WBC) gains importance in identifying diseases like leukemia, leukopenia, etc. WBC does not possess regular structure because they move throughout the blood stream and hence analyzing WBC and its types for structure and shape is quite challenging. To aid in hematology, this work provides classification of WBC classification based on modified U‐Net and support vector machines (SVM). A modified U‐Net architecture is developed to segment WBC followed by feature extraction and classification by radial basis function‐support vector machine (RBF‐SVM). Experiments indicated that the modified U Net segmentation can detect the WBC nucleus with a dice similarity coefficient of 0.972. The proposed U‐Net‐SVM can recognize WBCs in Raabin‐WBC, LISC, and BCCD datasets with an accuracy of 99.45%, 98.62%, and 98.81%, respectively. Further investigation on leukemia dataset, ALL‐IDB2, revealed an accuracy of 99.42% with 100% sensitivity and specificity. The proposed model can be used to investigate WBCs and hence provide a great support to the hematologists in analyzing the blood smear for various disease identifications. [ABSTRACT FROM AUTHOR]

This paper presents a new approach which uses the tools within artificial intelligence (AI) software libraries as an alternative way of solving partial differential equations (PDEs) that have been discretised using standard numerical methods. In particular, we describe how to represent numerical discretisations arising from the finite volume and finite element methods by pre‐determining the weights of convolutional layers within a neural network. As the weights are defined by the discretisation scheme, no training of the network is required and the solutions obtained are identical (accounting for solver tolerances) to those obtained with standard codes often written in Fortran or C++. We also explain how to implement the Jacobi method and a multigrid solver using the functions available in AI libraries. For the latter, we use a U‐Net architecture which is able to represent a sawtooth multigrid method. A benefit of using AI libraries in this way is that one can exploit their built‐in technologies to enable the same code to run on different computer architectures (such as central processing units, graphics processing units or new‐generation AI processors) without any modification. In this article, we apply the proposed approach to eigenvalue problems in reactor physics where neutron transport is described by diffusion theory. For a fuel assembly benchmark, we demonstrate that the solution obtained from our new approach is the same (accounting for solver tolerances) as that obtained from the same discretisation coded in a standard way using Fortran. We then proceed to solve a reactor core benchmark using the new approach. For both benchmarks we give timings for the neural network implementation run on a CPU and a GPU, and a serial Fortran code run on a CPU. [ABSTRACT FROM AUTHOR]

For the effective diagnosis of cardio vascular disease (CVD), anatomical characteristics of the heart must be examined, which depends on segmenting the cardiac tissues of interest and then classifying them into appropriate pathological groups. In recent years, deep learning (DL)‐based computer aided design (CAD) segmentation has been employed to automate the segmentation process. Despite the evolution of several DL methods, they still fail due to the shape variation of the heart in patients and the availability of a limited amount of data. This paper proposes an effective Saliency and Active Contour‐based Attention UNet3+ algorithm to segment the ventricles of the heart, which is a challenging task for most researchers, especially with an irregularly shaped right ventricle (RV) that varies over cardiac phases. The algorithm outperforms other state‐of‐the‐art methods in DC metrics, which proves its efficiency in automating the segmentation process. [ABSTRACT FROM AUTHOR]

The worldwide spread of the coronavirus (COVID‐19) outbreak has proven devastating to public health. The severity of pneumonia relies on a rapid and accurate diagnosis of COVID‐19 in CT images. Accordingly, a computed tomography (CT) scan is an excellent screening tool for detecting COVID‐19. This paper proposes a deep learning‐based strategy for recognizing and segmenting a COVID‐19 lesion from chest CT images, which would introduce an accurate computer aided decision criteria for the physicians about the severity rate of the patients. Two main stages have been proposed for detecting COVID‐19; first, a convolutional neural network (CNN) deep structure recognizes and classifies COVID‐19 from CT images. Second, a U‐Net deep structure segments the COVID‐19 regions in a semantic manner. The proposed system is trained and evaluated on three different CT datasets for COVID‐19, two of which are used to illustrate the system's segmentation performance and the other is to demonstrate the system's classification ability. Experiment results reveal that the proposed CNN can achieve classification accuracy greater than 0.99, and the proposed U‐Net model outperforms the state‐of‐the‐art in segmentation with an IOU greater than 0.92. [ABSTRACT FROM AUTHOR]

This paper explores a deep‐learning approach to evaluate the position of circular delimiters in cartridge case images. These delimiters define two regions of interest (ROI), corresponding to the breech face and the firing pin impressions, and are placed manually or by an image‐processing algorithm. This positioning bears a significant impact on the performance of the image‐matching algorithms for firearm identification, and an automated evaluation method would be beneficial to any computerized system. Our contribution consists in optimizing and training U‐Net segmentation models from digital images of cartridge cases, intending to locate ROIs automatically. For the experiments, we used high‐resolution 2D images from 1195 samples of cartridge cases fired by different 9MM firearms. Our results show that the segmentation models, trained on augmented data sets, exhibit a performance of 95.6% IoU (Intersection over Union) and 99.3% DC (Dice Coefficient) with a loss of 0.014 for the breech face images; and a performance of 95.9% IoU and 99.5% DC with a loss of 0.011 for the firing pin images. We observed that the natural shapes of predicted circles reduce the performance of segmentation models compared with perfect circles on ground truth masks suggesting that our method provide a more accurate segmentation of the real ROI shape. In practice, we believe that these results could be useful for firearms identification. In future work, the predictions may be used to evaluate the quality of delimiters on specimens in a database, or they could determine the region of interest on a cartridge case image. [ABSTRACT FROM AUTHOR]

Skilled seasonal forecasting will effectively reduce the economic losses caused by droughts and floods. Because of the powerful data mining capability of deep learning networks, it is increasingly applied in studies of seasonal rainfall prediction. However, there remain two prominent issues in the modeling process: the lack of enough training samples and the effect of a small number of extreme values on the model optimization. To tackle these deficiencies, we combine strategies such as principal component analysis, reduction of model hidden layers, and early‐stopping with Attention U‐Net to construct a rainfall classification forecasting model. These steps reduced the model outfitting and improved the model generalization. The results show that the prediction accuracy of this network with leads of 1–3 months is obviously better than that of the numerical model. Further analysis also supports that the spatial features of precipitation predicted by the network are very close to the observations. Plain Language Summary: Accurate summer precipitation forecasts can effectively reduce economic losses and casualties caused by heavy precipitation and flooding in China. Nowadays, seasonal forecasting still mainly relies on numerical models based on physical knowledge. However, the predictors of rainfall are so broad and complex that some physical processes are not yet fully understood. As a result, the improvement of the model skills encountered a bottleneck. Recently, deep learning (DL) network can achieve complex function approximations of data by automatically extracting and learning data features. Yet the existing DL seasonal prediction models suffer from over‐fitting and interference from noise and extreme values. Here, we use Attention U‐Net to predict rainfall levels and combine strategies such as principal component analysis, reduced model depth, and early‐stopping. This network not only solves the two drawbacks mentioned above. It also shows the superiority in the prediction results. First, the rainfall level prediction of this network is more accurate than the numerical model at 1–3 months ahead. Second, the network captures the spatial features with excellent performance. The predicted distributions are largely consistent with the true values. Key Points: The Attention U‐Net model outperforms the numerical model in predicting summer precipitation levels with a forecast lead of 1 monthConverting a regression task into a classification task can eliminate the bad effects of noises and extremesThe principal component analysis and the early‐stopping strategy facilitate the model's generalization [ABSTRACT FROM AUTHOR]

Saturn Kilometric Radiation (SKR) is a non‐thermal auroral emission with peak emission occurring at 100–400 kHz. Its properties have been extensively studied since Cassini's arrival at Saturn until mission end with its Radio and Plasma Wave Science (RPWS) experiment. Low Frequency Extensions (LFEs) of SKR which consist of global intensifications of SKR accompanied by extensions of the main SKR band down to lower frequencies have been studied in particular. Low Frequency Extensions result from internally driven tail reconnection and from solar wind compressions of the magnetosphere, which also trigger tail reconnection. They have been cataloged through visual inspection with two approaches, using an intensity threshold for LFEs in 2006 (Reed et al., 2018, https://doi.org/10.1002/2017ja024499) and more recently O'Dwyer et al. (2023a, https://doi.org/10.25546/103103) produced a sample of LFEs detected by Cassini/RPWS by fitting their exact frequency‐time coordinates with polygons. In this study we use the latter catalog of LFEs as a training set for an image based machine learning algorithm to classify all LFEs detected by Cassini/RPWS. The inputs to the model are multi‐channel images consisting of spectrogram images in flux density and degree of circular polarization. The outputs of the model are binary masks showing the exact location of the LFE in frequency‐time space. The median Intersection Over Union across the testing and training set were calculated to be 0.97 and 0.98, respectively. The output of this study is a list of all 4,874 LFEs detected using this method. The list of LFE frequency‐time coordinates is available for use amongst the scientific community. Plain Language Summary: We are using radio observations from the Cassini spacecraft that was in orbit around the planet Saturn for 13 years. We want to search for characteristic features of Saturn's auroral radio emissions (called Saturn Kilometric Radiation or SKR) in the data stream from the radio instrument—specifically events called Low Frequency Extensions (LFEs). The edges of these events can be tracked in time‐frequency spectrograms of Cassini radio observations. We find several hundred examples of the LFEs that we're looking for, and feed these into a computer algorithm which learns what they look like. The algorithm can then be applied to new/unseen data and we allow it to search for similar events. The end result is an extensive catalog of all the LFEs observed throughout the 13‐year near‐Saturn mission by the radio instrument of Cassini. This catalog can be used by the scientific community as a basis for statistical studies of Saturn's radio emissions. The machine learning aspect of this work can be adapted through something known as transfer learning to other planets where we look for similar features in data. Key Points: Supervised learning applied to database of labeled polygons marked on radio spectrogramsFocus on Low Frequency Extensions of Saturn Kilometric Radiation to return a full catalog from the Cassini missionA modified U‐Net architecture achieved median Intersection over Union values of 0.98 and 0.97 across the training and testing set [ABSTRACT FROM AUTHOR]

Summary: Low segmentation accuracy and background noise exisit in rectal cancer lesion segmentation. To segment colorectal tumors in CT images accurately, we propose an improved U‐Net network for colorectal tumor image segmentation dilated‐pyramid‐attention U‐Net (DPA‐UNet). In this paper, we use the U‐Net network and combine techniques such as dilated convolution, weighted feature pyramid structure (W‐FPN), and convolutional block attention module (CBAM) mechanism. Firstly, CBAM and W‐FPN are combined to extract dense pixel‐level features for pixel labeling. Secondly, after the third network output layer, three serially dilated depth‐separable dilated convolutional layers with dilation rates of 1, 2, and 4, are added respectively to expand the feature map receptive field. Finally, the DPA‐UNet model is compared and analyzed with other new network structures. The experimental results show that DPA‐UNet achieves automatic segmentation of the colorectal cancer image region of interest (ROI). [ABSTRACT FROM AUTHOR]

Accurate segmentation of brain gliomas (BG) is a crucial and challenging task for effective treatment planning in BG therapy. This study presents the weight loss function and dropout U‐Net with ConvNeXt block (WD‐UNeXt), which precisely segments BG from few shot MRI. The ConvNeXt block, which comprises the main body of the network, is a structure that can extract more detailed features from images. The weight loss function addresses the issue of category imbalance, thereby enhancing the network's ability to achieve more precise segmentation. The training set of BraTS2019 was used to train the network and apply it to test data. Dice similarity coefficient (DSC), sensitivity (Sen), specificity (Spec) and Hausdorff distance (HD) were used to assess the performance of the method. The experimental results demonstrate that the DSC of whole tumour, tumour core and enhancing tumour reached 0.934, 0.911 and 0.851, respectively. Sen of the sub‐regions achieved 0.922, 0.911 and 0.867. Spec and HD reached 1.000, 1.000, 1.000 and 3.224, 2.990, 2.844, respectively. Compared with the performance of state‐of‐the‐art methods, the DSC and HD of WD‐UNeXt were improved to varying degrees. Therefore, this method has considerable potential for the segmentation of BG. [ABSTRACT FROM AUTHOR]

In this article, we propose a new U‐Net‐based approach for intraretinal cyst segmentation across different vendors that improve some of the challenges faced by previous deep‐based techniques. The proposed method has two main steps: (1) prior information embedding and input data adjustment, (2) the segmentation model. In the first step, we inject the information into the network in a way that overcomes some of the network limitations in receiving data and learning important contextual knowledge. And in the next step, we introduce a connection module between the encoder and decoder parts that transfers information more effectively from the encoder to the decoder. Two public datasets, namely, OPTIMA and KERMANY, are employed to evaluate the proposed method. The results show that the proposed method is an efficient vendor‐independent approach for the segmentation of intraretinal cystoid fluid with mean Dice values of 0.78 and 0.81 on the OPTIMA and KERMANY datasets, respectively. [ABSTRACT FROM AUTHOR]

Lung cancer is one of the deadliest cancers in the world and is a serious threat to human life. Lung nodules are an early manifestation of lung cancer, early detection and treatment of which can improve the survival rate of patients. In order to accurately segment the lung nodule regions in lung CT images, CA‐UNet, an encoding and decoding structure based on convolution and attention fusion, is proposed based on the U‐Net network. It has improved on two points: First, at the skip connection, the global feature information is extracted using the Swin Transformer block and then fused with the pre‐extraction features and subsequently fed into the corresponding layer of the decoder; second, each channel information is reweighted in the decoder by the channel attention module so that the network focuses on more important channels. Experimental results on the LIDC‐IDRI public database of lung nodules showed that the intersection of union, dice similarity coefficient, precision, and recall of the algorithm were 82.42%, 89.86%, 89.07%, and 92.44%, respectively. The algorithm has better segmentation performance compared to other segmentation methods. [ABSTRACT FROM AUTHOR]

Fast and precise reconstruction algorithm is desired for for multifocal structured illumination microscopy (MSIM) to obtain the super‐resolution image. This work proposes a deep convolutional neural network (CNN) to learn a direct mapping from raw MSIM images to super‐resolution image, which takes advantage of the computational advances of deep learning to accelerate the reconstruction. The method is validated on diverse biological structures and in vivo imaging of zebrafish at a depth of 100 µm. The results show that high‐quality, super‐resolution images can be reconstructed in one‐third of the runtime consumed by conventional MSIM method, without compromising spatial resolution. Last but not least, a fourfold reduction in the number of raw images required for reconstruction is achieved by using the same network architecture, yet with different training data. [ABSTRACT FROM AUTHOR]

Recent advancements in transmission electron microscopy (TEM) have enabled the study of atomic structures of materials at unprecedented scales as small as tens of picometers (pm). However, accurately detecting atomic positions from TEM images remains a challenging task. Traditional Gaussian fitting and peak‐finding algorithms are effective under ideal conditions but perform poorly on images with strong background noise or contamination areas (shown as ultrabright or ultradark contrasts). Moreover, these traditional algorithms require parameter tuning for different magnifications. To overcome these challenges, AtomID‐Net is presented, a deep neural network model for atomic detection from multiscale low‐SNR experimental images of scanning TEM (scanning transmission electron microscopy (STEM)). The model is trained on real images, which allows the robust and efficient detection of atomic positions, even in the presence of background noise and contamination. The evaluation on a test set of 50 images with a resolution of 800 × 800 yields an average F1‐Score of 0.964, which demonstrates significant improvements over existing peak‐finding algorithms. [ABSTRACT FROM AUTHOR]