Standard U-Net Research Articles

End-to-End Fully Automated Lung Cancer Volume Estimation System

The volume of the tumor plays a very crucial role in deciding the stage of lung cancer which in turn helps in deciding the best treatment and its schedule. Currently used computer-based volume estimation techniques are semi-automatic with limited accuracy. For any automatic lung cancer segmentation system, lung CT scans of hundreds of patients are required along with their corresponding annotated segmentation masks. It is difficult to get accurately annotated data as cancer segmentation of CT scans done by the radiologists, is a time-consuming manual process. Also, it is subjective and prone to intra and inter-observer variability. Further, owing to the irregular shape of the cancerous tumor, accurate volume estimation becomes a challenge with regular convolution models. This paper proposes an end-to-end automatic tumor volume estimation model that estimates volume using the GPR (Gaussian Process Regression) interpolation method. The proposed modified cancer segmentation model uses deformable convolutions. This modification offers a higher segmentation accuracy in terms of IoU (Intersection over Union) and clearly defined nodule boundaries with correct retention of the nodule shape. The research was undertaken in collaboration with Nanavati Hospital, Mumbai, and all the models were validated on a real dataset obtained from the hospital. The proposed model gives a mean segmentation IoU (Intersection over Union) of 0.9035 and a volume estimation accuracy of 93.13% which are almost 5% and 3% higher than 0.8548 and 90.51% which are the corresponding results obtained using a standard U-net++ algorithm.

Foam detection in a stirred tank using deep learning neural networks

A deep learning method based on the Convolutional Neural Network (CNN) U-Net architecture has been explored as an image analysis tool for the detection and automated quantification of foam generated in a stirred tank. Multiple datasets of different sizes and with different types of foam were obtained experimentally and processed with data augmentation techniques in order to train the deep learning networks. The impact of adding attention and residual gates to the U-Net model to improve its performance, as well as the size of the dataset used to train the model have been assessed. The main factor influencing the accuracy of the models to detect the foam in the stirred tank is the size and quality of the dataset use to train the U-Net models; it must be sufficiently large and varied in terms of foam type/structure. The addition of attention gates and residual blocks to the U-Net model slightly improves the accuracy of foam detection, particularly for images that contain additional objects or obstructions, however the training time is 30 % longer than the standard U-Net model. In all cases, this image analysis method based on the U-Net model largely out performs conventional image analysis, which was not able to automatically differentiate between foam, additional objects in the tank and bubbles in the liquid. Two methods for the measurement of foam quantity (maximum foam height and foam volume) have also been developed based on the foam regions detected by the models. This is important for the application of the tool, which is to be used to understand the impact of operating conditions on foam formation in lab-/pilot-scale stirred tanks and then develop scale-up guidelines for foam control in industrial tanks.

Oceanic Precipitation Nowcasting Using a UNet-Based Residual and Attention Network and Real-Time Himawari-8 Images

Qualitative precipitation forecasting plays a vital role in marine operational services. However, predicting heavy precipitation over the open ocean presents a significant challenge due to the limited availability of ground-based radar observations far from coastal regions. Recent advancements in deep learning models offer potential for oceanic precipitation nowcasting using satellite images. This study implemented an enhanced UNet model with an attention mechanism and a residual architecture (RA-UNet) to predict the precipitation rate within a 90 min time frame. A comparative analysis with the standard UNet and UNet with an attention algorithm revealed that the RA-UNet method exhibited superior accuracy metrics, such as the critical ratio index and probability of detection, with fewer false alarms. Two typical cases demonstrated that RA-UNet had a better ability to forecast monsoon precipitation as well as intense precipitation in a tropical cyclone. These findings indicate the greater potential of the RA-UNet approach for nowcasting heavy precipitation over the ocean using satellite imagery.

Transformer based deep learning hybrid architecture for phase unwrapping

A deep learning Hybrid architecture for phase unwrapping has been proposed. The hybrid architecture is based on integration of Convolutional Neural Networks (CNN) with Vision Transformer. The performance of Hybrid architecture/network in phase unwrapping is compared against CNN based standard UNET network. Structural Similarity Index (SSIM) and Root Mean Square Error (RMSE) have been used as performance metrics to assess the performance of these deep learning networks for phase unwrapping. To train and test the networks, dataset with high mean Entropy has been generated using Gaussian filtering of random noise in Fourier plane. The Hybrid architecture is tested on test dataset and is found to have superior performance metrics against the UNET network. Their performance is also tested in noisy environment with various noise levels and Hybrid architecture demonstrated better anti-noise capability than UNET network. Hybrid architecture was successfully validated in real world scenario using experimental data from custom built Digital Holographic Microscope. With the advent of newer architectures and hardware, Deep learning networks can further improve the performance in solving inverse problems.

Enhancing Landslide Detection with SBConv-Optimized U-Net Architecture Based on Multisource Remote Sensing Data

This study introduces a novel approach to landslide detection by incorporating the Spatial and Band Refinement Convolution (SBConv) module into the U-Net architecture, to extract features more efficiently. The original U-Net architecture employs convolutional layers for feature extraction, during which it may capture some redundant or less relevant features. Although this approach aids in building rich feature representations, it can also lead to an increased consumption of computational resources. To tackle this challenge, we propose the SBConv module, an efficient convolutional unit designed to reduce redundant computing and enhance representative feature learning. SBConv consists of two key components: the Spatial Refined Unit (SRU) and the Band Refined Unit (BRU). The SRU adopts a separate-and-reconstruct approach to mitigate spatial redundancy, while the BRU employs a split-transform-and-fuse strategy to decrease band redundancy. Empirical evaluation reveals that models equipped with SBConv not only show a reduction in redundant features but also achieve significant improvements in performance metrics. Notably, SBConv-embedded models demonstrate a marked increase in Recall and F1 Score, outperforming the standard U-Net model. For instance, the SBConvU-Net variant achieves a Recall of 75.74% and an F1 Score of 73.89%, while the SBConvResU-Net records a Recall of 70.98% and an F1 Score of 73.78%, compared to the standard U-Net’s Recall of 60.59% and F1 Score of 70.91%, and the ResU-Net’s Recall of 54.75% and F1 Score of 66.86%. These enhancements in detection accuracy underscore the efficacy of the SBConv module in refining the capabilities of U-Net architectures for landslide detection of multisource remote sensing data. This research contributes to the field of landslide detection based on remote sensing technology, providing a more effective and efficient solution. It highlights the potential of the improved U-Net architecture in environmental monitoring and also provides assistance in disaster prevention and mitigation efforts.

Equivariant neural operators for gradient-consistent topology optimization

Abstract Most traditional methods for solving partial differential equations (PDEs) require the costly solving of large linear systems. Neural operators (NOs) offer remarkable speed-ups over classical numerical PDE solvers. Here, we conduct the first exploration and comparison of NOs for three-dimensional topology optimization. Specifically, we propose replacing the PDE solver within the popular Solid Isotropic Material with Penalization (SIMP) algorithm, which is its main computational bottleneck. For this, the NO not only needs to solve the PDE with sufficient accuracy but also has the additional challenge of providing accurate gradients which are necessary for SIMP’s density updates. To realize this, we do three things: (i) We introduce a novel loss term to promote gradient-consistency. (ii) We guarantee equivariance in our NOs to increase the physical correctness of predictions. (iii) We introduce a novel NO architecture called U-Net Fourier neural operator (U-Net FNO), which combines the multi-resolution properties of U-Nets with the Fourier neural operator (FNO)’s focus on local features in frequency space. In our experiments we demonstrate that the inclusion of the novel gradient loss term is necessary to obtain good results. Furthermore, enforcing group equivariance greatly improves the quality of predictions, especially on small training datasets. Finally, we show that in our experiments the U-Net FNO outperforms both a standard U-Net, as well as other FNO methods.

Comprehensive analysis of synthetic learning applied to neonatal brain MRI segmentation.

Brain segmentation from neonatal MRI images is a very challenging task due to large changes in the shape of cerebral structures and variations in signal intensities reflecting the gestational process. In this context, there is a clear need for segmentation techniques that are robust to variations in image contrast and to the spatial configuration of anatomical structures. In this work, we evaluate the potential of synthetic learning, a contrast-independent model trained using synthetic images generated from the ground truth labels of very few subjects. We base our experiments on the dataset released by the developmental Human Connectome Project, for which high-quality images are available for more than 700 babies aged between 26 and 45 weeks postconception. First, we confirm the impressive performance of a standard UNet trained on a few volumes, but also confirm that such models learn intensity-related features specific to the training domain. We then confirm the robustness of the synthetic learning approach to variations in image contrast. However, we observe a clear influence of the age of the baby on the predictions. We improve the performance of this model by enriching the synthetic training set with realistic motion artifacts and over-segmentation of the white matter. Based on extensive visual assessment, we argue that the better performance of the model trained on real T2w data may be due to systematic errors in the ground truth. We propose an original experiment allowing us to show that learning from real data will reproduce any systematic bias affecting the training set, while synthetic models can avoid this limitation. Overall, our experiments confirm that synthetic learning is an effective solution for segmenting neonatal brain MRI. Our adapted synthetic learning approach combines key features that will be instrumental for large multisite studies and clinical applications.

An encoder-decoder ConvLSTM surrogate model for simulating geological CO2 sequestration with dynamic well controls

In Geological Carbon Sequestration (GCS), effectively managing the project requires predicting state variables such as pressure and saturation. However, numerical simulation of multiphase flow in subsurface porous media involves solving large linear algebra systems, resulting in a substantial computational burden. This can make it impractical for real-time history matching or optimization. Meanwhile, deep learning-based surrogate models are emerging as fast and accurate approximators. This study delves into the application of the Encoder-Decoder Convolutional Long Short-Term Memory (ED-ConvLSTM) neural network for predicting the complex evolution of state variables under dynamic CO2 injection schemes. Inception blocks enhanced with light-weighted attention modules are introduced in the encoder to extract high-dimensional input features. ConvLSTM is employed to propagate spatial temporal information in the low-dimensional latent space. Further, progressive upsampling blocks are used to reconstruct the latent features for the desired output. Instead of taking discrete time steps as an input feature, the proposed network captures the dynamic dependencies with the inherent ConvLSTM cell. The network has access to data at only portion of the initial time steps during training stage, while it is used to predict the state variables at unseen time steps during testing stage. Results show that the network can produce excellent predictions for both pressure and saturation, even at unseen future time steps. The remarkable generalizability to different geological permeability fields is also evaluated. ED-ConvLSTM outperforms the standard U-Net by far, especially when predicting beyond the training time period. These numerical experiments demonstrate the advantages of ED-ConvLSTM in terms of prediction accuracy, extrapolability and generalizability. This study highlights the importance of incorporating recurrent connections into the deep neural networks for simulating time-dependent multiphase flow problems. The proposed methodology has great potential in GCS surrogate modeling and offers a possible approach for real-time optimization of CO2 injection.

Evaluating the Deep Learning Models Performance for Segmentation of Oral Epithelial Dysplasia: A Histological Data-Driven Approach

Oral epithelial dysplasia (OED) poses a significant precancerous risk, potentially progressing to oral squamous cell carcinoma (OSCC). Precise segmentation of OED within histopathological images is pivotal for early diagnosis and treatment planning. This study evaluates Deep Learning (DL) models for precise Oral Epithelial Dysplasia (OED) segmentation in biopsy slide images. The Vanilla UNET model is explored with the standard UNET and other transfer learning models (VGG16, VGG19, MobileNet, and DeepLabV3+) as the backbone of the model. For our application, U-Net demonstrated superior performance (IoU: 93.73%, precision: 97.96%, recall: 97.78%, F1-score: 96.76%). Visual examples highlight model strengths and limitations, providing insights beyond traditional metrics. This research advances computer-aided histopathological analysis, emphasizing DL models’ crucial role in enhancing diagnostic accuracy and patient care.

A precise model for skin cancer diagnosis using hybrid U-Net and improved MobileNet-V3 with hyperparameters optimization

Skin cancer is a frequently occurring and possibly deadly disease that necessitates prompt and precise diagnosis in order to ensure efficacious treatment. This paper introduces an innovative approach for accurately identifying skin cancer by utilizing Convolution Neural Network architecture and optimizing hyperparameters. The proposed approach aims to increase the precision and efficacy of skin cancer recognition and consequently enhance patients' experiences. This investigation aims to tackle various significant challenges in skin cancer recognition, encompassing feature extraction, model architecture design, and optimizing hyperparameters. The proposed model utilizes advanced deep-learning methodologies to extract complex features and patterns from skin cancer images. We enhance the learning procedure of deep learning by integrating Standard U-Net and Improved MobileNet-V3 with optimization techniques, allowing the model to differentiate malignant and benign skin cancers. Also substituted the crossed-entropy loss function of the Mobilenet-v3 mathematical framework with a bias loss function to enhance the accuracy. The model's squeeze and excitation component was replaced with the practical channel attention component to achieve parameter reduction. Integrating cross-layer connections among Mobile modules has been proposed to leverage synthetic features effectively. The dilated convolutions were incorporated into the model to enhance the receptive field. The optimization of hyperparameters is of utmost importance in improving the efficiency of deep learning models. To fine-tune the model's hyperparameter, we employ sophisticated optimization methods such as the Bayesian optimization method using pre-trained CNN architecture MobileNet-V3. The proposed model is compared with existing models, i.e., MobileNet, VGG-16, MobileNet-V2, Resnet-152v2 and VGG-19 on the “HAM-10000 Melanoma Skin Cancer dataset". The empirical findings illustrate that the proposed optimized hybrid MobileNet-V3 model outperforms existing skin cancer detection and segmentation techniques based on high precision of 97.84%, sensitivity of 96.35%, accuracy of 98.86% and specificity of 97.32%. The enhanced performance of this research resulted in timelier and more precise diagnoses, potentially contributing to life-saving outcomes and mitigating healthcare expenditures.

Standardizing and improving dose predictions for head and neck cancers using complete sets of OAR contours.

Radiotherapy dose predictions have been trained with data from previously treated patients of similar sites and prescriptions. However, clinical datasets are often inconsistent and do not contain the same number of organ at risk (OAR) structures. The effects of missing contour data in deep learning-based dose prediction models have not been studied. The purpose of this study was to investigate the impacts of incomplete contour sets in the context of deep learning-based radiotherapy dose prediction models trained with clinical datasets and to introduce a novel data substitution method that utilizes automated contours for undefined structures. We trained Standard U-Nets and Cascade U-Nets to predict the volumetric dose distributions of patients with head and neck cancers (HNC) using three input variations to evaluate the effects of missing contours, as well as a novel data substitution method. Each architecture was trained with the original contour (OC) inputs, which included missing information, hybrid contour (HC) inputs, where automated OAR contours generated in software were substituted for missing contour data, and automated contour (AC) inputs containing only automated OAR contours. 120 HNC treatments were used for model training, 30 were used for validation and tuning, and 44 were used for evaluation and testing. Model performance and accuracy were evaluated with global whole body dose agreement, PTV coverage accuracy, and OAR dose agreement. The differences in these values between dataset variations were used to determine the effects of missing data and automated contour substitutions. Automated contours used as substitutions for missing data were found to improve dose prediction accuracy in the Standard U-Net and Cascade U-Net, with a statistically significant difference in some global metrics and/or OAR metrics. For both models, PTV coverage between input variations was unaffected by the substitution technique. Automated contours in HC and AC datasets improved mean dose accuracy for some OAR contours, including the mandible and brainstem, with a greater improvement seen with HC datasets. Global dose metrics, including mean absolute error, mean error, and percent error were different for the Standard U-Net but not for the Cascade U-Net. Automated contours used as a substitution for contour data improved prediction accuracy for some but not all dose prediction metrics. Compared to the Standard U-Net models, the Cascade U-Net achieved greater precision.

Image-Enhanced U-Net: Optimizing Defect Detection in Window Frames for Construction Quality Inspection

Ensuring the structural integrity of window frames and detecting subtle defects, such as dents and scratches, is crucial for maintaining product quality. Traditional machine vision systems face challenges in defect identification, especially with reflective materials and varied environments. Modern machine and deep learning (DL) systems hold promise for post-installation inspections but face limitations due to data scarcity and environmental variability. Our study introduces an innovative approach to enhance DL-based defect detection, even with limited data. We present a comprehensive window frame defect detection framework incorporating optimized image enhancement, data augmentation, and a core U-Net model. We constructed five datasets using cell phones and the Spot Robot for autonomous inspection, evaluating our approach across various scenarios and lighting conditions in real-world window frame inspections. Our results demonstrate significant performance improvements over the standard U-Net model, with a notable 7.43% increase in the F1 score and 15.1% in IoU. Our approach enhances defect detection capabilities, even in challenging real-world conditions. To enhance the generalizability of this study, it would be advantageous to apply its methodology across a broader range of diverse construction sites.

Automated seismic semantic segmentation using attention U-Net

Seismic facies mapping from a 3D seismic cube is of significant value to various seismic interpretation and characterization tasks. Traditional facies mapping is based on examining sedimentary environments and stratigraphic sequences that provide distinct characteristics used for facies mapping. Given the complex nature of the task, manual facies mapping is typically time and labor consuming, and the quality of the decisions varies as a function of expertise. This complexity is further increased with the ever-increasing size of 3D seismic data sets. Deep-learning methods have indicated a promising potential to perform fast, accurate, and automated segmentation tasks. We investigate the application of machine-learning techniques, particularly state-of-the-art deep convolutional neural networks (CNNs), as a framework to perform accurate automated seismic facies pixel-wise segmentation. The workflow consists of a CNN-based U-Net architecture that adopts modern computer vision techniques. We develop three major changes to the standard U-Net to boost the performance for seismic semantic segmentation tasks: (1) using residual building blocks in the encoder, (2) using transformer-like attention gates after each residual block, and (3) using frequency spectrum data, in addition to seismic amplitude, as input to the network. We indicate that this implementation achieves higher accuracy metrics outperforming recently published state-of-the-art benchmarks. The performance of our method is validated using two 3D seismic data sets, the F3 Netherlands data set and the Penobscot data set acquired offshore Nova Scotia, Canada. Experimentation involves training on a set of samples and tuning the hyperparameters, followed by quantitative evaluation of the trained network. Our workflow produces high-quality segmentation with significantly reduced artifacts, improved edge detection, and improved lateral consistency throughout the seismic survey.

Preliminary analysis of predicting the first recurrence in patients with neovascular age-related macular degeneration using deep learning

BackgroundTo predict, using deep learning, the first recurrence in patients with neovascular age-related macular degeneration (nAMD) after three monthly loading injections of intravitreal anti-vascular endothelial growth factor (anti-VEGF).MethodsOptical coherence tomography (OCT) images were obtained at baseline and after the loading phase. The first recurrence was defined as the initial appearance of a new retinal hemorrhage or intra/subretinal fluid accumulation after the initial resolution of exudative changes after three loading injections. Standard U-Net architecture was used to identify the three retinal fluid compartments, which include pigment epithelial detachment, subretinal fluid, and intraretinal fluid. To predict the first recurrence of nAMD, classification learning was conducted to determine whether the first recurrence occurred within three months after the loading phase. The recurrence classification architecture was built using ResNet50. The model with retinal regions of interest of the entire region and fluid region on OCT at baseline and after the loading phase is presented.ResultsA total of 1,444 eyes of 1,302 patients were included. The mean duration until the first recurrence after the loading phase was 8.20 ± 15.56 months. The recurrence classification system revealed that the model with the fluid region of OCT after the loading phase provided the highest classification performance, with an area under the receiver operating characteristic curve (AUC) of 0.725 ± 0.012. Heatmap analysis revealed that three pathological fluids, subsided choroidal neovascularization lesions, and hyperreflective foci were important areas for the first recurrence.ConclusionsThe deep learning algorithm allowed for the prediction of the first recurrence for three months after the loading phase with adequate feasibility. An automated prediction system may assist in establishing patient-specific treatment plans and the provision of individualized medical care for patients with nAMD.

A Complete Strategy to Achieve High Precision Automatic Segmentation of Challenging Experimental X‐Ray Computed Tomography Data Using Low‐Resemblance Synthetic Training Data

It is shown that preconditioning of experimental X‐ray computed tomography (XCT) data is critical to achieve high‐precision segmentation scores. The challenging experimental XCT datasets and deep convolutional neural networks (DCNNs) are used that are trained with low‐resemblance synthetic XCT data. The material used is a 6‐phase Al–Si metal matrix composite‐reinforced with ceramic fibers and particles. To achieve generalization, in our past studies, specific data augmentation techniques were proposed for the synthetic XCT training data. In addition, two toolsets are devised: (1) special 3D DCNN architecture (3D Triple_UNet), slicing the experimental XCT data from multiple views (MultiView Forwarding), the i.S.Sy.Da.T.A. iterative segmentation algorithm, and (2) nonlocal means (NLM) conditioning (filtering) for the experimental XCT data. This results in good segmentation Dice scores across all phases compared to more standard approaches (i.e., standard UNet architecture, single view slicing, standard single training, and NLM conditioning). Herein, the NLM filter is replaced with the deep conditioning framework BAM SynthCOND introduced in a previous publication, which can be trained with synthetic XCT data. This leads to a significant segmentation precision increase for all phases. The proposed methods are potentially applicable to other materials and imaging techniques.

Total water storage anomalies reconstruction using noise-augmented u-shaped network: A case study in the Yangtze River Basin

The satellite mission of Gravity Recovery and Climate Experiment (GRACE) and its follow-on mission (GRACE-FO) have characterized global total water storage anomalies (TWSA) with unprecedented accuracy. However, the data gap between GRACE and GRACE-FO from July 2017 to May 2018 represents challenges for interpreting long-term water storage changes. In this study, we present a deep learning model, a noise-augmented u-shaped network (NA-UNet), to bridge the gap over the Yangtze River Basin (YRB). This model has been trained on precipitation, temperature, and hydrological model data. We show that the NA-UNet model achieves a state-of-the-art TWSA reconstruction, outperforming the standard UNet model and previous studies. At the basin scale, the NA-UNet model agrees favorably well with GRACE observations, with a correlation coefficient (CC) of 0.99, Nash-Sutcliff efficient (NSE) of 0.97, and normalized root-mean-square error (NRMSE) of 0.04 during the testing period. At the grid-cell scale, our model has a much more stable performance with median CC/NSE/NRMSE values of 0.99/0.96/0.05. This significant improvement in the gap-filling ability addresses the issue with discontinuous TWSA observations while laying the foundation for predicting future changes in total water storage.

Segmenting Cervical Arteries in Phase Contrast Magnetic Resonance Imaging Using Convolutional Encoder–Decoder Networks

Phase contrast (PC) magnetic resonance imaging (MRI) is a primary method used to quantify blood flow. Cerebral blood flow (CBF) is an important hemodynamic parameter to characterize cerebrovascular and neurological diseases. However, a critical step in CBF quantification using PC MRI is vessel segmentation, which is largely manual, and thus time-consuming and prone to interrater variability. Here, we present encoder–decoder deep learning models to automate segmentation of neck arteries to accurately quantify CBF. The PC-MRI data were collected from 46 Moyamoya (MM) patients and 107 healthy control (HC) participants. Three segmentation U-Net models (Standard, Nested, and Attention) were compared. The PC MRI images were taken before and 15 min after vasodilation. The models were assessed based on their ability to detect the internal carotid arteries (ICAs), external carotid arteries (ECAs), and vertebral arteries (VAs), using the Dice score coefficient (DSC) of overlap between manual and predicted segmentations and receiver operator characteristic (ROC) metric. Analysis of variance, Wilcoxon rank-sum test, and paired t-test were used for comparisons. The Standard U-NET, Attention U-Net, and Nest U-Net models achieved results of mean DSCs of 0.81 ± 0.21, and 0.85 ± 0.14, and 0.85 ± 0.13, respectively. The ROC curves revealed high area under the curve scores for all methods (≥0.95). While the Nested and Attention U-Net architectures accomplished reliable segmentation performance for HC and MM subsets, Standard U-Net did not perform as well in the subset of MM patients. Blood flow velocities calculated by the models were statistically comparable. In conclusion, optimized deep learning architectures can successfully segment neck arteries in PC MRI images and provide precise quantification of their blood flow.

Head and Neck Cancer Segmentation in FDG PET Images: Performance Comparison of Convolutional Neural Networks and Vision Transformers.

Convolutional neural networks (CNNs) have a proven track record in medical image segmentation. Recently, Vision Transformers were introduced and are gaining popularity for many computer vision applications, including object detection, classification, and segmentation. Machine learning algorithms such as CNNs or Transformers are subject to an inductive bias, which can have a significant impact on the performance of machine learning models. This is especially relevant for medical image segmentation applications where limited training data are available, and a model's inductive bias should help it to generalize well. In this work, we quantitatively assess the performance of two CNN-based networks (U-Net and U-Net-CBAM) and three popular Transformer-based segmentation network architectures (UNETR, TransBTS, and VT-UNet) in the context of HNC lesion segmentation in volumetric [F-18] fluorodeoxyglucose (FDG) PET scans. For performance assessment, 272 FDG PET-CT scans of a clinical trial (ACRIN 6685) were utilized, which includes a total of 650 lesions (primary: 272 and secondary: 378). The image data used are highly diverse and representative for clinical use. For performance analysis, several error metrics were utilized. The achieved Dice coefficient ranged from 0.833 to 0.809 with the best performance being achieved by CNN-based approaches. U-Net-CBAM, which utilizes spatial and channel attention, showed several advantages for smaller lesions compared to the standard U-Net. Furthermore, our results provide some insight regarding the image features relevant for this specific segmentation application. In addition, results highlight the need to utilize primary as well as secondary lesions to derive clinically relevant segmentation performance estimates avoiding biases.

MCSTransWnet: A new deep learning process for postoperative corneal topography prediction based on raw multimodal data from the Pentacam HR system

This work provides a new multimodal fusion generative adversarial net (GAN) model, Multiple Conditions Transform W-net (MCSTransWnet), which primarily uses femtosecond laser arcuate keratotomy surgical parameters and preoperative corneal topography to predict postoperative corneal topography in astigmatism-corrected patients. The MCSTransWnet model comprises a generator and a discriminator, and the generator is composed of two sub-generators. The first sub-generator extracts features using the U-net model, vision transform (ViT) and a multi-parameter conditional module branch. ​The second sub-generator uses a U-net network for further image denoising. The discriminator uses the pixel discriminator in Pix2Pix. Currently, most GAN models are convolutional neural networks; however, due to their feature extraction locality, it is difficult to comprehend the relationships among global features. Thus, we added a vision Transform network as the model branch to extract the global features. It is normally difficult to train the transformer, and image noise and geometric information loss are likely. Hence, we adopted the standard U-net fusion scheme and transform network as the generator, so that global features, local features, and rich image details could be obtained simultaneously. Our experimental results clearly demonstrate that MCSTransWnet successfully predicts postoperative corneal topographies (structural similarity ​= ​0.765, peak signal-to-noise ratio ​= ​16.012, and Fréchet inception distance ​= ​9.264). Using this technique to obtain the rough shape of the postoperative corneal topography in advance gives clinicians more references and guides changes to surgical planning and improves the success rate of surgery.

FCD-AttResU-Net: An improved forest change detection in Sentinel-2 satellite images using attention residual U-Net

Forest Change Detection (FCD) is a critical component of natural resource monitoring and conservation strategies, enabling informed decision-making. Various methods utilizing the power of artificial intelligence (AI) have been developed for detecting and categorizing changes in forest cover using remote sensing (RS) data. One prominent AI-powered approach is the U-Net, a deep learning (DL) architecture famous for its segmentation proficiency. However, the standard U-Net architecture fails to effectively capture intricate spatial dependencies and long-range contextual information present in remote sensing imagery. To address this research gap, we introduce an attention-residual-based novel DL model which leverages the U-Net architecture and Sentinel-2 satellite images to map alterations in forest vegetation cover in the tropical region. Our novel model enhances the U-Net architecture by seamlessly integrating the strengths of the U-Net, harnessing attention mechanisms strategically to amplify crucial features, and leveraging cutting-edge residual connections to facilitate the smooth flow of information and gradient propagation. These meticulous design choices enabled the precise feature extraction, resulting in improved computational performance of the proposed method compared to the Standard U-Net, Deeplabv3+, Deep Res-U-Net, and Attention U-Net. The classification results demonstrate the enhanced efficiency of our model, achieving a Mean Intersection over Union (MIoU) of 0.9330 on our test dataset. This performance surpasses the Attention U-Net (0.9146), Standard U-Net (0.9029), Deeplabv3+ (0.9247), and Deep Res-U-Net (0.9282). The comparative analysis of ground truth reproductions unveiled the superior detection capabilities of our model in accurately identifying forest and non-forest polygons, surpassing both the standard U-Net, and the U-Net augmented with attention mechanism, along with other state-of-the-art techniques, thereby highlighting its enhanced efficacy. The model’s broad applicability can support forest managers and ecologists in rapidly evaluating the long-term ramifications of infrastructure initiatives, such as roads, on tropical forests, including those in Brunei.