Cycle-consistent Generative Adversarial Network Research Articles

Multicenter PET image harmonization using generative adversarial networks.

To improve reproducibility and predictive performance of PET radiomic features in multicentric studies by cycle-consistent generative adversarial network (GAN) harmonization approaches. GAN-harmonization was developed to harmonize whole-body PET scans to perform image style and texture translation between different centers and scanners. GAN-harmonization was evaluated by application to two retrospectively collected open datasets and different tasks. First, GAN-harmonization was performed on a dual-center lung cancer cohort (127 female, 138 male) where the reproducibility of radiomic features in healthy liver tissue was evaluated. Second, GAN-harmonization was applied to a head and neck cancer cohort (43 female, 154 male) acquired from three centers. Here, the clinical impact of GAN-harmonization was analyzed by predicting the development of distant metastases using a logistic regression model incorporating first-order statistics and texture features from baseline 18F-FDG PET before and after harmonization. Image quality remained high (structural similarity: left kidney 0.800, right kidney 0.806, liver 0.780, lung 0.838, spleen 0.793, whole-body 0.832) after image harmonization across all utilized datasets. Using GAN-harmonization, inter-site reproducibility of radiomic features in healthy liver tissue increased at least by 5 ± 14% (first-order), 16 ± 7% (GLCM), 19 ± 5% (GLRLM), 16 ± 8% (GLSZM), 17 ± 6% (GLDM), and 23 ± 14% (NGTDM). In the head and neck cancer cohort, the outcome prediction improved from AUC 0.68 (95% CI 0.66-0.71) to AUC 0.73 (0.71-0.75) by application of GAN-harmonization. GANs are capable of performing image harmonization and increase reproducibility and predictive performance of radiomic features derived from different centers and scanners.

Deep Learning Synthesis of White-Blood From Dark-Blood Late Gadolinium Enhancement Cardiac Magnetic Resonance.

Dark-blood late gadolinium enhancement (DB-LGE) cardiac magnetic resonance has been proposed as an alternative to standard white-blood LGE (WB-LGE) imaging protocols to enhance scar-to-blood contrast without compromising scar-to-myocardium contrast. In practice, both DB and WB contrasts may have clinical utility, but acquiring both has the drawback of additional acquisition time. The aim of this study was to develop and evaluate a deep learning method to generate synthetic WB-LGE images from DB-LGE, allowing the assessment of both contrasts without additional scan time. DB-LGE and WB-LGE data from 215 patients were used to train 2 types of unpaired image-to-image translation deep learning models, cycle-consistent generative adversarial network (CycleGAN) and contrastive unpaired translation, with 5 different loss function hyperparameter settings each. Initially, the best hyperparameter setting was determined for each model type based on the Fréchet inception distance and the visual assessment of expert readers. Then, the CycleGAN and contrastive unpaired translation models with the optimal hyperparameters were directly compared. Finally, with the best model chosen, the quantification of scar based on the synthetic WB-LGE images was compared with the truly acquired WB-LGE. The CycleGAN architecture for unpaired image-to-image translation was found to provide the most realistic synthetic WB-LGE images from DB-LGE images. The results showed that it was difficult for visual readers to distinguish if an image was true or synthetic (55% correctly classified). In addition, scar burden quantification with the synthetic data was highly correlated with the analysis of the truly acquired images. Bland-Altman analysis found a mean bias in percentage scar burden between the quantification of the real WB and synthetic white-blood images of 0.44% with limits of agreement from -10.85% to 11.74%. The mean image quality of the real WB images (3.53/5) was scored higher than the synthetic white-blood images (3.03), P = 0.009. This study proposed a CycleGAN model to generate synthetic WB-LGE from DB-LGE images to allow assessment of both image contrasts without additional scan time. This work represents a clinically focused assessment of synthetic medical images generated by artificial intelligence, a topic with significant potential for a multitude of applications. However, further evaluation is warranted before clinical adoption.

The Application and Exploration of Data Visualization Technology in Artistic Expression and Exhibition

Embarking on the transformation of virtual art museums, the confluence of data visualization technology and deep learning emerges as a catalyst for redefining artistic expression. Driven by the desire to change traditional approaches, this work utilizes a sophisticated deep learning model, the self-attention-based cycle-consistent generative adversarial network (SA-CCGAN), for transformative purposes. The objective of proposed Modernizing Virtual Art: The Collaboration of Data Visualization Technology and Self-Attention-Based Cycle-Consistent Generative Adversarial Network for Artistic Expression (MVA: DVT & SA-CCGAN) is to elevate artistic expression and exhibition practices within the virtual space, marking a paradigm shift in traditional approaches. SA-CCGAN, renowned for its ability to generate realistic and coherent artistic representations, serves as the cornerstone of this transformative endeavor. The workflow intricately incorporates SA-CCGAN into the virtual art museum context, enhancing stylized outcomes and capturing global geometric features. This yields a computationally efficient representation of high-dimensional visual data, revolutionizing digital art presentation. The fusion of data visualization technology, creativity, and meticulous design not only signals innovation but also has the potential to redefine the virtual art landscape, fostering accessibility, engagement, and a deeper cultural exchange in digital art exploration.

Reinventing Web Security: An Enhanced Cycle-Consistent Generative Adversarial Network Approach to Intrusion Detection

Web3.0, as the link between the physical and digital domains, faces increasing security threats due to its inherent complexity and openness. Traditional intrusion detection systems (IDSs) encounter formidable challenges in grappling with the multidimensional and nonlinear traffic data characteristic of the Web3.0 environment. Such challenges include insufficient samples of attack data, inadequate feature extraction, and resultant inaccuracies in model classification. Moreover, the scarcity of certain traffic data available for analysis by IDSs impedes the system’s capacity to document instances of malicious behavior. In response to these exigencies, this paper presents a novel approach to Web3.0 intrusion detection, predicated on the utilization of cycle-consistent generative adversarial networks (CycleGANs). Leveraging the data transformation capabilities of its generator, this method facilitates bidirectional conversion between normal Web3.0 behavioral data and potentially intrusive behavioral data. This transformative process not only augments the diversity and volume of recorded intrusive behaviors but also clandestinely simulates various attack scenarios. Furthermore, through fostering mutual competition and learning between the discriminator and generator, the approach enhances the ability to discern the defining characteristics of potential intrusive behaviors, thereby bolstering the accuracy of intrusion detection. To substantiate the efficacy of the CycleGAN-based intrusion detection method, simulation experiments were conducted utilizing public datasets, including KDD CUP 1999 (KDD), CIC-DDOS2019, CIC-IDS2018, and SR-BH 2020. The experimental findings evince the method’s remarkable accuracies across the four datasets, attaining rates of 99.81%, 97.79%, 89.25%, and 95.15%, respectively, while concurrently maintaining low false-positive rates. This research contributes novel insights and methodologies toward the advancement of Web3.0 intrusion detection through the application of CycleGAN technology, which is poised to play a pivotal role in fortifying the security landscape of Web3.0.

Unsupervised speckle denoising in digital holographic interferometry based on 4-f optical simulation integrated cycle-consistent generative adversarial network

The speckle noise generated during digital holographic interferometry (DHI) is unavoidable and difficult to eliminate, thus reducing its accuracy. We propose a self-supervised deep-learning speckle denoising method using a cycle-consistent generative adversarial network to mitigate the effect of speckle noise. The proposed method integrates a 4-f optical speckle noise simulation module with a parameter generator. In addition, it uses an unpaired dataset for training to overcome the difficulty in obtaining noise-free images and paired data from experiments. The proposed method was tested on both simulated and experimental data, with results showing a 6.9% performance improvement compared with a conventional method and a 2.6% performance improvement compared with unsupervised deep learning in terms of the peak signal-to-noise ratio. Thus, the proposed method exhibits superior denoising performance and potential for DHI, being particularly suitable for processing large datasets.

Deep learning-based glomerulus detection and classification with generative morphology augmentation in renal pathology images

Glomerulus morphology on renal pathology images provides valuable diagnosis and outcome prediction information. To provide better care, an efficient, standardized, and scalable method is urgently needed to optimize the time-consuming and labor-intensive interpretation process by renal pathologists. This paper proposes a deep convolutional neural network (CNN)-based approach to automatically detect and classify glomeruli with different stains in renal pathology images. In the glomerulus detection stage, this paper proposes a flattened Xception with a feature pyramid network (FX-FPN). The FX-FPN is employed as a backbone in the framework of faster region-based CNN to improve glomerulus detection performance. In the classification stage, this paper considers classifications of five glomerulus morphologies using a flattened Xception classifier. To endow the classifier with higher discriminability, this paper proposes a generative data augmentation approach for patch-based glomerulus morphology augmentation. New glomerulus patches of different morphologies are generated for data augmentation through the cycle-consistent generative adversarial network (CycleGAN). The single detection model shows the F1 score up to 0.9524 in H&E and PAS stains. The classification result shows that the average sensitivity and specificity are 0.7077 and 0.9316, respectively, by using the flattened Xception with the original training data. The sensitivity and specificity increase to 0.7623 and 0.9443, respectively, by using the generative data augmentation. Comparisons with different deep CNN models show the effectiveness and superiority of the proposed approach.

Investigating the causes of permeability anisotropy in heterogeneous conglomeratic sandstone using multiscale digital rock

Heterogeneous conglomeratic sandstone exhibits anisotropic physical properties, rendering a comprehensive analysis of its physical processes challenging with experimental measurements. Digital rock technology provides a visual and intuitive analysis of the microphysical processes in rocks, thereby aiding in scientific inquiry. Nevertheless, the multiscale characteristics of conglomeratic sandstone cannot be fully captured by a single-scale digital rock, thus limiting its ability to characterize the pore structure. Our work introduces a proposed workflow that employs multiscale digital rock fusion to investigate permeability anisotropy in heterogeneous rock. We utilize a cycle-consistent generative adversarial network (CycleGAN) to fuse CT scans data of different resolutions, creating a large-scale, high-precision digital rock that comprehensively represents the conglomeratic sandstone pore structure. Subsequently, the digital rock is partitioned into multiple blocks, and the permeability of each block is simulated using a pore network. Finally, the total permeability of the sample is calculated by conducting an upscaling numerical simulation using the Darcy-Stokes equation. This process facilitates the analysis of the pore structure in conglomeratic sandstone and provides a step-by-step solution for permeability. From a multiscale perspective, this approach reveals that the anisotropy of permeability in conglomeratic sandstone stems from the layered distribution of grain sizes and differences in grain arrangement across different directions.

Enhancing scanning electron microscopy imaging quality of weakly conductive samples through unsupervised learning

Scanning electron microscopy (SEM) is a crucial tool for analyzing submicron-scale structures. However, the attainment of high-quality SEM images is contingent upon the high conductivity of the material due to constraints imposed by its imaging principles. For weakly conductive materials or structures induced by intrinsic properties or organic doping, the SEM imaging quality is significantly compromised, thereby impeding the accuracy of subsequent structure-related analyses. Moreover, the unavailability of paired high–low quality images in this context renders the supervised-based image processing methods ineffective in addressing this challenge. Here, an unsupervised method based on Cycle-consistent Generative Adversarial Network (CycleGAN) was proposed to enhance the quality of SEM images for weakly conductive samples. The unsupervised model can perform end-to-end learning using unpaired blurred and clear SEM images from weakly and well-conductive samples, respectively. To address the requirements of material structure analysis, an edge loss function was further introduced to recover finer details in the network-generated images. Various quantitative evaluations substantiate the efficacy of the proposed method in SEM image quality improvement with better performance than the traditional methods. Our framework broadens the application of artificial intelligence in materials analysis, holding significant implications in fields such as materials science and image restoration.

Domain adaptation for structural health monitoring via physics-informed and self-attention-enhanced generative adversarial learning

Health monitoring technologies, empowered by sensor-driven information and model updating, play an important role in assessing the status of civil structures and detecting anomalies. However, significant domain discrepancies exist in the distribution of physical parameters between numerical structural models and their real-world counterparts. As a result, many health monitoring theories struggle to be effective in practice, even if they perform perfectly in the simulated data. To bridge the domain discrepancies, this paper proposes an unsupervised domain adaptation approach based on an adapted cycle-consistent generative adversarial network (CycleGAN) that incorporates physical constraints and a self-attention mechanism. The approach focuses on the mutual transformation of the multi-channel time series obtained from numerical models and actual structures. Specifically, the physical constraints are derived from the governing equation of linear dynamic systems, while the self-attention mechanism is achieved by adding transformer structures to both the generator and discriminator. Through free-vibration experiments on a steel beam and a large-scale steel bridge model, the physical constraints and transformer structures have proven beneficial for improving the learning capability and training stability of the GAN model. Furthermore, the proposed approach is not only verified as effective in transforming acceleration responses between the test structures and their corresponding finite element models in both time and frequency domains but has also been shown to reproduce mode shapes accurately.

Learning CT-free attenuation-corrected total-body PET images through deep learning.

Total-body PET/CT scanners with long axial fields of view have enabled unprecedented image quality and quantitative accuracy. However, the ionizing radiation from CT is a major issue in PET imaging, which is more evident with reduced radiopharmaceutical doses in total-body PET/CT. Therefore, we attempted to generate CT-free attenuation-corrected (CTF-AC) total-body PET images through deep learning. Based on total-body PET data from 122 subjects (29 females and 93 males), a well-established cycle-consistent generative adversarial network (Cycle-GAN) was employed to generate CTF-AC total-body PET images directly while introducing site structures as prior information. Statistical analyses, including Pearson correlation coefficient (PCC) and t-tests, were utilized for the correlation measurements. The generated CTF-AC total-body PET images closely resembled real AC PET images, showing reduced noise and good contrast in different tissue structures. The obtained peak signal-to-noise ratio and structural similarity index measure values were 36.92 ± 5.49 dB (p < 0.01) and 0.980 ± 0.041 (p < 0.01), respectively. Furthermore, the standardized uptake value (SUV) distribution was consistent with that of real AC PET images. Our approach could directly generate CTF-AC total-body PET images, greatly reducing the radiation risk to patients from redundant anatomical examinations. Moreover, the model was validated based on a multidose-level NAC-AC PET dataset, demonstrating the potential of our method for low-dose PET attenuation correction. In future work, we will attempt to validate the proposed method with total-body PET/CT systems in more clinical practices. The ionizing radiation from CT is a major issue in PET imaging, which is more evident with reduced radiopharmaceutical doses in total-body PET/CT. Our CT-free PET attenuation correction method would be beneficial for a wide range of patient populations, especially for pediatric examinations and patients who need multiple scans or who require long-term follow-up. • CT is the main source of radiation in PET/CT imaging, especially for total-body PET/CT devices, and reduced radiopharmaceutical doses make the radiation burden from CT more obvious. • The CT-free PET attenuation correction method would be beneficial for patients who need multiple scans or long-term follow-up by reducing additional radiation from redundant anatomical examinations. • The proposed method could directly generate CT-free attenuation-corrected (CTF-AC) total-body PET images, which is beneficial for PET/MRI or PET-only devices lacking CT image poses.

Optimized self-attention based cycle-consistent generative adversarial network adopted melanoma classification from dermoscopic images.

Skin is the exposed part of the human body that constantly protected from UV rays, heat, light, dust, and other hazardous radiation. One of the most dangerous illnesses that affect people is skin cancer. A type of skin cancer called melanoma starts in the melanocytes, which regulate the colour in human skin. Reducing the fatality rate from skin cancer requires early detection and diagnosis of conditions like melanoma. In this article, a Self-attention based cycle-consistent generative adversarial network optimized with Archerfish Hunting Optimization Algorithm adopted Melanoma Classification (SACCGAN-AHOA-MC-DI) from dermoscopic images is proposed. Primarily, the input Skin dermoscopic images are gathered via the dataset of ISIC 2019. Then, the input Skin dermoscopic images is pre-processed using adjusted quick shift phase preserving dynamic range compression (AQSP-DRC) for removing noise and increase the quality of Skin dermoscopic images. These pre-processed images are fed to the piecewise fuzzy C-means clustering (PF-CMC) for ROI region segmentation. The segmented ROI region is supplied to the Hexadecimal Local Adaptive Binary Pattern (HLABP) to extract the Radiomic features, like Grayscale statistic features (standard deviation, mean, kurtosis, and skewness) together with Haralick Texture features (contrast, energy, entropy, homogeneity, and inverse different moments). The extracted features are fed to self-attention based cycle-consistent generative adversarial network (SACCGAN) which classifies the skin cancers as Melanocytic nevus, Basal cell carcinoma, Actinic Keratosis, Benign keratosis, Dermatofibroma, Vascular lesion, Squamous cell carcinoma and melanoma. In general, SACCGAN not adapt any optimization modes to define the ideal parameters to assure accurate classification of skin cancer. Hence, Archerfish Hunting Optimization Algorithm (AHOA) is considered to maximize the SACCGAN classifier, which categorizes the skin cancer accurately. The proposed method attains 23.01%, 14.96%, and 45.31% higher accuracy and 32.16%, 11.32%, and 24.56% lesser computational time evaluated to the existing methods, like melanoma prediction method for unbalanced data utilizing optimized Squeeze Net through bald eagle search optimization (CNN-BES-MC-DI), hyper-parameter optimized CNN depending on Grey wolf optimization algorithm (CNN-GWOA-MC-DI), DEANN incited skin cancer finding depending on fuzzy c-means clustering (DEANN-MC-DI). RESEARCH HIGHLIGHTS: This manuscript, self-attention based cycle-consistent. SACCGAN-AHOA-MC-DI method is implemented in Python. (SACCGAN-AHOA-MC-DI) from dermoscopic images is proposed. Adjusted quick shift phase preserving dynamic range compression (AQSP-DRC). Removing noise and increase the quality of Skin dermoscopic images.

Diagnostic performance of deep-learning-based virtual chromoendoscopy in gastric neoplasms.

Cycle-consistent generative adversarial network (CycleGAN) is a deep neural network model that performs image-to-image translations. We generated virtual indigo carmine (IC) chromoendoscopy images of gastric neoplasms using CycleGAN and compared their diagnostic performance with that of white light endoscopy (WLE). WLE and IC images of 176 patients with gastric neoplasms who underwent endoscopic resection were obtained. We used 1,633 images (911 WLE and 722 IC) of 146 cases in the training dataset to develop virtual IC images using CycleGAN. The remaining 30 WLE images were translated into 30 virtual IC images using the trained CycleGAN and used for validation. The lesion borders were evaluated by 118 endoscopists from 22 institutions using the 60 paired virtual IC and WLE images. The lesion area concordance rate and successful whole-lesion diagnosis were compared. The lesion area concordance rate based on the pathological diagnosis in virtual IC was lower than in WLE (44.1% vs. 48.5%, p < 0.01). The successful whole-lesion diagnosis was higher in the virtual IC than in WLE images; however, the difference was insignificant (28.2% vs. 26.4%, p = 0.11). Conversely, subgroup analyses revealed a significantly higher diagnosis in virtual IC than in WLE for depressed morphology (41.9% vs. 36.9%, p = 0.02), differentiated histology (27.6% vs. 24.8%, p = 0.02), smaller lesion size (42.3% vs. 38.3%, p = 0.01), and assessed by expert endoscopists (27.3% vs. 23.6%, p = 0.03). The diagnostic ability of virtual IC was higher for some lesions, but not completely superior to that of WLE. Adjustments are required to improve the imaging system's performance.

Cycle-Retinex: Unpaired Low-Light Image Enhancement via Retinex-Inline CycleGAN

Low-light image enhancement aims to recover normal-light images from the images captured under dim environments. Most existing methods could just improve the light appearance globally whereas failing to handle other degradation such as dense noise, color offset and extremely low-light. Moreover, unsupervised methods proposed in recent years lack reliable physical model as the basis, thus universality is greatly limited. To address these problems, we propose a novel low-light image enhancement method via Retinex-inline cycle-consistent generative adversarial network named Cycle-Retinex, whose training is totally dependent on unpaired datasets. Specifically, we organically combine Retinex theory with CycleGAN, by which we decouple low-light image enhancement task into two sub-tasks, <italic xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">i.e.</i> illumination map enhancement and reflectance map restoration. Retinex theory helps CycleGAN simplify low-light image enhancement problem and CycleGAN provides synthetic paired images to guide the training of Retinex decomposition network. We further introduce a self-augmented method to address the color distortion and noise problem, thus making the network learn to enhance low-light images adaptively. Extensive experiments show that the proposed method can achieve promising results. The source code is available at <uri xmlns:mml="http://www.w3.org/1998/Math/MathML" xmlns:xlink="http://www.w3.org/1999/xlink">https://github.com/mummmml/Cycle-Retinex</uri> .

A Sim-to-Real Learning-Based Framework for Contact-Rich Assembly by Utilizing CycleGAN and Force Control

Deep reinforcement learning (RL) has succeeded in robotic manipulation applications. However, training robots in the real world is challenging due to sample efficiency and safety concerns. Sim-to-real transfer has been proposed to address the aforementioned concerns but introduces the reality gap. In this work, we introduce a sim-to-real learning framework for vision-based assembly tasks and perform training in a simulation environment by employing raw image inputs from a single camera to address the aforementioned issues. We build a robotic Peg-in-Hole (PiH) training environment that requires low-level simulation knowledge. We also present a domain adaptation method based on a cycle-consistent generative adversarial network (CycleGAN) and a force control transfer approach to bridge the reality gap. The proposed framework, trained in a simulation environment with different environmental scenes, can be successfully transferred to a real PiH setup with a UR5e robot. We then reproduce these results with a Diana7 robot and different peg shapes to verify the generalization ability of the framework.

Cycle‐Consistent Generative Adversarial Network and Crypto Hash Signature Token‐based Block chain Technology for Data Aggregation with Secured Routing in Wireless Sensor Networks

SummaryA fundamental component of Wireless Sensor Networks (WSN) is routing, because it is responsible for data transmission to the base stations (BS). Routing attacks are attacks that have the ability to interfere with the operation of WSN. A reliable routing system is needed for guaranteeing routing security including WSN effectiveness. Many studies have been conducted to improve trust among routing nodes, cryptographic algorithms and centralized routing decisions. Nevertheless, most of the routing techniques are not practical because it is challenging to detect suspicious actions of routing nodes. Generally, there is lack of reliable technique for preventing malicious node attacks. Therefore in this manuscript, Cycle‐Consistent Generative Adversarial Network (CCGAN) optimized with Ebola Optimization Search Algorithm (EOSA) and Crypto Hash Signature (CHS) Token‐based Block chain (BC) Technology for Data Aggregation with Secured Routing in Wireless Sensor Networks (SR‐CCGAN‐EOSA‐BDA‐WSN) is proposed for data aggregation with secured optimum routing in WSN. The proposed methodology uses a Proof of Authority (PoA) method in the block chain network to authenticate the process of node transmission. A Cycle‐Consistent Generative Adversarial Network optimized with the Ebola optimization algorithm methodology (CCGAN‐EOSA) is used to select the validation group required for proofing and select the proper next hop as a forwarding node proficient of securely and easily transmitting messages. The performance metrics, like delay, average latency with energy consumption, block chain token transactions throughput is analyzed. The performance of SR‐CCGAN‐EOSA‐BDAWSN proposed method provides 76.26%, 65.57%, and 42.9% lesser delay under 30% spiteful routing environment; 73.06% are compared with existing methods.

Improved artificial intelligence discrimination of minor histological populations by supplementing with color-adjusted images

Despite the dedicated research of artificial intelligence (AI) for pathological images, the construction of AI applicable to histopathological tissue subtypes, is limited by insufficient dataset collection owing to disease infrequency. Here, we present a solution involving the addition of supplemental tissue array (TA) images that are adjusted to the tonality of the main data using a cycle-consistent generative adversarial network (CycleGAN) to the training data for rare tissue types. F1 scores of rare tissue types that constitute < 1.2% of the training data were significantly increased by improving recall values after adding color-adjusted TA images constituting < 0.65% of total training patches. The detector also enabled the equivalent discrimination of clinical images from two distinct hospitals and the capability was more increased following color-correction of test data before AI identification (F1 score from 45.2 ± 27.1 to 77.1 ± 10.3, p < 0.01). These methods also classified intraoperative frozen sections, while excessive supplementation paradoxically decreased F1 scores. These results identify strategies for building an AI that preserves the imbalance between training data with large differences in actual disease frequencies, which is important for constructing AI for practical histopathological classification.

Study of peripheral dose from low-dose CT to adaptive radiotherapy of postoperative prostate cancer.

The increasing use of computed tomography (CT) for adaptive radiotherapy (ART) has raised concerns about the peripheral radiation dose. This study investigates the feasibility of low-dose CT (LDCT) for postoperative prostate cancer ART to reduce the peripheral radiation dose, and evaluates the peripheral radiation dose of different imaging techniques and propose an image enhancement method based on deep learning for LDCT. A linear accelerator integrated with a 16-slice fan-beam CT from UIH (United Imaging Healthcare, China) was utilized for prostate cancer ART. To reduce the tube current of CT for ART, LDCT was acquired. Peripheral doses of normal-dose CT (NDCT), LDCT, and mega-voltage computed tomography (MV-CT) were measured using a cylindrical Virtual Water™ phantom and an ion chamber. A deep learning model of LDCT for abdominal and pelvic-based cycle-consistent generative adversarial network was employed to enhance the image quality of LDCT. Six postoperative prostate cancer patients were selected to evaluate the feasibility of low-dose CT network restoration images (RCT) by the deep learning model for ART. The three aspects among NDCT, LDCT, and RCT were compared: the Hounsfield Unit (HU) of the tissue, the Dice Similarity Coefficient (DSC) criterion of target and organ, and dose calculation differences. In terms of peripheral dose, the LDCT had a surface measurement point dose of approximately 1.85 mGy at the scanning field, while the doses of NDCT and MV-CT were higher at 22.85 mGy and 29.97 mGy, respectively. However, the image quality of LDCT was worse than NDCT. When compared to LDCT, the tissue HU value of RCT showed a significant improvement and was closer to that of NDCT. The DSC results for target CTV between RCT and NDCT were also impressive, reaching up to 94% for bladder and femoral heads, 98% for rectum, and 94% for the target organ. Additionally, the dose calculation differences for the ART plan based on LDCT and NDCT were all within 1%. Overall, these findings suggest that RCT can provide an effective alternative to NDCT and MV-CT with similar or better outcomes in HU values of tissue and organ damage. More testing is required before clinical application.

AGLC-GAN: Attention-based global-local cycle-consistent generative adversarial networks for unpaired single image dehazing

Image dehazing is a critical image pre-processing task to estimate the haze-free images corresponding to the input hazy images. Despite the recent advances, the task of image dehazing remains challenging, especially in the unsupervised scenario. Several efforts can be found in the literature to dehaze images in a supervised set-up, where a huge number of paired (clear and hazy images) images are required for training. The supervised approaches often become biased towards the nature of haze present in the training hazy images, and produce less realistic images for query hazy images. We propose an Attention-based Global–Local Cycle-consistent Generative Adversarial Network (AGLC-GAN) for Unpaired Single Image Dehazing. The proposed CycleGAN-based AGLC-GAN model contains a dehazing generator encapsulating an autoencoder-like network with an attention mechanism comprising channel attention and pixel attention to deal with uneven haze intensity across the image. We use a global–local consistent discriminator to identify spatially varying haze and improve the stability of the discriminator. We adopt cyclic perceptual consistency loss to maintain consistency in the feature space. A dynamic feature enhancement module and an adaptive mix-up module are included in the proposed generator to dynamically obtain more spatially structured features and hence, adaptively preserve the flow of shallow features. Furthermore, we extensively experiment with the proposed model on multiple benchmark datasets for evaluating the efficacy of removing haze. The results of the experiments conducted in the study, demonstrate a significant quantitative and qualitative improvement over the existing methods for unpaired image dehazing. 11The code is available at the following link: https://github.com/jr925/AGLC-GAN

Vortex Beam Transmission Compensation in Atmospheric Turbulence Using CycleGAN

To improve the robustness of vortex beam transmission and detection in the face of atmospheric turbulence and to guarantee accurate recognition of orbital angular momentum (OAM), we present an end-to-end dynamic compensation technique for vortex beams using an improved cycle-consistent generative adversarial network (CycleGAN). This approach transforms the problem of vortex beam distortion compensation into one of image translation. The Pix2pix and CycleGAN models were extended with a structural similarity loss function to constrain turbulence distortion compensation in luminance, contrast and structure. Experiments were designed to evaluate the compensation performance from subjective and objective indicators. The simulation results demonstrate that the optical OAM intensity map is very similar to that of the target OAM light after compensation. The mean value of structural similarity is close to 1. The recognition accuracy of the OAM is improved by 4.4% compared to no distortion compensation, demonstrating that the improved CycleGAN-based compensation scheme can guarantee excellent detection accuracy without reconstructing the wavefront and saving optical hardware. The method can be implemented in real-time optical communications in atmospheric turbulence environments.

Comparison and evaluation of different deep learning models of synthetic CT generation from CBCT for nasopharynx cancer adaptive proton therapy.

Cone-beam computed tomography (CBCT) scanning is used for patient setup in image-guided radiotherapy. However, its inaccurate CT numbers limit its applicability in dose calculation and treatment planning. This study compares four deep learning methods for generating synthetic CT (sCT) to determine which method is more appropriate and offers potential for further clinical exploration in adaptive proton therapy for nasopharynx cancer. CBCTs and deformed planning CT (dCT) from 75 patients (60/5/10 for training, validation and testing) were used to compare cycle-consistent Generative Adversarial Network(cycleGAN), Unet, Unet+cycleGAN and conditionalGenerative Adversarial Network (cGAN) for sCT generation. The sCT images generated by each method were evaluated against dCT images using mean absolute error (MAE), structural similarity (SSIM), peak signal-to-noise ratio (PSNR), spatial non-uniformity (SNU) and radial averaging in the frequency domain. In addition, dosimetric accuracy was assessed through gamma analysis, differences in water equivalent thickness (WET), and dose-volume histogram metrics. The cGAN model has demonstrated optimal performance in the four models across various indicators. In terms of image quality under global condition, the average MAE has been reduced to 16.39HU, SSIM has increased to 95.24%, and PSNR has increased to 28.98. Regarding dosimetric accuracy, the gamma passing rate (2%/2mm) has reached 99.02%, and the WET difference is only 1.28mm. The D95 value of CTVs coverage and Dmax value of spinal cord, brainstem show no significant differences between dCT and sCT generated by cGAN model. The cGAN model has been shown to be a more suitable approach for generating sCT using CBCT, considering its characteristics and concepts. The resulting sCT has the potential for application in adaptive proton therapy.