Detecting images generated by diffusers.

In recent years, the field of artificial intelligence has witnessed a remarkable surge in the generation of synthetic images, driven by advancements in deep learning techniques. These synthetic images, often created through complex algorithms, closely mimic real photographs, blurring the lines between reality and artificiality. This proliferation of synthetic visuals presents a pressing challenge: how to accurately and reliably distinguish between genuine and generated images. This article, in particular, explores the task of detecting images generated by text-to-image diffusion models, highlighting the challenges and peculiarities of this field. To evaluate this, we consider images generated from captions in the MSCOCO and Wikimedia datasets using two state-of-the-art models: Stable Diffusion and GLIDE. Our experiments show that it is possible to detect the generated images using simple multi-layer perceptrons (MLPs), starting from features extracted by CLIP or RoBERTa, or using traditional convolutional neural networks (CNNs). These latter models achieve remarkable performances in particular when pretrained on large datasets. We also observe that models trained on images generated by Stable Diffusion can occasionally detect images generated by GLIDE, but only on the MSCOCO dataset. However, the reverse is not true. Lastly, we find that incorporating the associated textual information with the images in some cases can lead to a better generalization capability, especially if textual features are closely related to visual ones. We also discovered that the type of subject depicted in the image can significantly impact performance. This work provides insights into the feasibility of detecting generated images and has implications for security and privacy concerns in real-world applications. The code to reproduce our results is available at: https://github.com/davide-coccomini/Detecting-Images-Generated-by-Diffusers.

Spatially Multiplexed Speckle on 1D Sensors for High-Speed 2D Sensing Applications.

Speckle pattern-based remote vibration monitoring has recently become increasingly valuable in industrial, commercial, and medical applications. The dynamic and random nature of speckle patterns offers practical applications for imaging and measurement systems. The speckle pattern is an interference pattern generated by light scattered from a rough surface onto a remote plane. It is typically sensed using area scan cameras (2D), which are limited to framerates of 2-4 kHz and can only capture a small region of interest (ROI). In this work, we propose a technique that enables the capture of synthetic 2D speckle patterns using a 1D high-acquisition-rate sensor and a diffractive optical element (DOE) to produce image replicas. The multiple replicas are scanned by the 1D sensor simultaneously at different spatial positions. This method provides an ability to sense remote vibrations in all directions, contrary to the case with a simple 1D sensing system.

Reliable Out-of-Distribution Recognition of Synthetic Images.

Generative adversarial networks (GANs) and diffusion models (DMs) have revolutionized the creation of synthetically generated but realistic-looking images. Distinguishing such generated images from real camera captures is one of the key tasks in current multimedia forensics research. One particular challenge is the generalization to unseen generators or post-processing. This can be viewed as an issue of handling out-of-distribution inputs. Forensic detectors can be hardened by the extensive augmentation of the training data or specifically tailored networks. Nevertheless, such precautions only manage but do not remove the risk of prediction failures on inputs that look reasonable to an analyst but in fact are out of the training distribution of the network. With this work, we aim to close this gap with a Bayesian Neural Network (BNN) that provides an additional uncertainty measure to warn an analyst of difficult decisions. More specifically, the BNN learns the task at hand and also detects potential confusion between post-processing and image generator artifacts. Our experiments show that the BNN achieves on-par performance with the state-of-the-art detectors while producing more reliable predictions on out-of-distribution examples.

Is deep learning-enabled real-time personalized CT dosimetry feasible using only patient images as input?

PURPOSE: To propose a novel deep-learning based dosimetry method that allows quick and accurate estimation of organ doses for individual patients, using only their computed tomography (CT) images as input. METHODS: Despite recent advances in medical dosimetry, personalized CT dosimetry remains a labour-intensive process. Current state-of-the-art methods utilize time-consuming Monte Carlo (MC) based simulations for individual organ dose estimation in CT. The proposed method uses conditional generative adversarial networks (cGANs) to substitute MC simulations with fast dose image generation, based on image-to-image translation. The pix2pix architecture in conjunction with a regression model was utilized for the generation of the synthetic dose images. The lungs, heart, breast, bone and skin were manually segmented to estimate and compare organ doses calculated using both the original and synthetic dose images, respectively. RESULTS: The average organ dose estimation error for the proposed method was 8.3% and did not exceed 20% for any of the organs considered. The performance of the method in the clinical environment was also assessed. Using segmentation tools developed in-house, an automatic organ dose calculation pipeline was set up. Calculation of organ doses for heart and lung for each CT slice took about 2 s. CONCLUSIONS: This work shows that deep learning-enabled personalized CT dosimetry is feasible in real-time, using only patient CT images as input.

Generation of fluoroscopy-alike radiographs as alternative datasets for deep learning in interventional radiology.

In fluoroscopy-guided interventions (FGIs), obtaining large quantities of labelled data for deep learning (DL) can be difficult. Synthetic labelled data can serve as an alternative, generated via pseudo 2D projections of CT volumetric data. However, contrasted vessels have low visibility in simple 2D projections of contrasted CT data. To overcome this, we propose an alternative method to generate fluoroscopy-like radiographs from contrasted head CT Angiography (CTA) volumetric data. The technique involves segmentation of brain tissue, bone, and contrasted vessels from CTA volumetric data, followed by an algorithm to adjust HU values, and finally, a standard ray-based projection is applied to generate the 2D image. The resulting synthetic images were compared to clinical fluoroscopy images for perceptual similarity and subject contrast measurements. Good perceptual similarity was demonstrated on vessel-enhanced synthetic images as compared to the clinical fluoroscopic images. Statistical tests of equivalence show that enhanced synthetic and clinical images have statistically equivalent mean subject contrast within 25% bounds. Furthermore, validation experiments confirmed that the proposed method for generating synthetic images improved the performance of DL models in certain regression tasks, such as localizing anatomical landmarks in clinical fluoroscopy images. Through enhanced pseudo 2D projection of CTA volume data, synthetic images with similar features to real clinical fluoroscopic images can be generated. The use of synthetic images as an alternative source for DL datasets represents a potential solution to the application of DL in FGIs procedures.

An Image Turing Test on Realistic Gastroscopy Images Generated by Using the Progressive Growing of Generative Adversarial Networks.

Generative adversarial networks (GAN) in medicine are valuable techniques for augmenting unbalanced rare data, anomaly detection, and avoiding patient privacy issues. However, there were limits to generating high-quality endoscopic images with various characteristics, such as peristalsis, viewpoints, light sources, and mucous patterns. This study used the progressive growing of GAN (PGGAN) within the normal distribution dataset to confirm the ability to generate high-quality gastrointestinal images and investigated what barriers PGGAN has to generate endoscopic images. We trained the PGGAN with 107,060 gastroscopy images from 4165 normal patients to generate highly realistic 5122 pixel-sized images. For the evaluation, visual Turing tests were conducted on 100 real and 100 synthetic images to distinguish the authenticity of images by 19 endoscopists. The endoscopists were divided into three groups based on their years of clinical experience for subgroup analysis. The overall accuracy, sensitivity, and specificity of the 19 endoscopist groups were 61.3%, 70.3%, and 52.4%, respectively. The mean accuracy of the three endoscopist groups was 62.4 [Group I], 59.8 [Group II], and 59.1% [Group III], which was not considered a significant difference. There were no statistically significant differences in the location of the stomach. However, the real images with the anatomical landmark pylorus had higher detection sensitivity. The images generated by PGGAN showed highly realistic depictions that were difficult to distinguish, regardless of their expertise as endoscopists. However, it was necessary to establish GANs that could better represent the rugal folds and mucous membrane texture.

Incorporating the synthetic CT image for improving the performance of deformable image registration between planning CT and cone-beam CT.

Objective: To develop a contrast learning-based generative (CLG) model for the generation of high-quality synthetic computed tomography (sCT) from low-quality cone-beam CT (CBCT). The CLG model improves the performance of deformable image registration (DIR). Methods: This study included 100 post-breast-conserving patients with the pCT images, CBCT images, and the target contours, which the physicians delineated. The CT images were generated from CBCT images via the proposed CLG model. We used the Sct images as the fixed images instead of the CBCT images to achieve the multi-modality image registration accurately. The deformation vector field is applied to propagate the target contour from the pCT to CBCT to realize the automatic target segmentation on CBCT images. We calculate the Dice similarity coefficient (DSC), 95 % Hausdorff distance (HD95), and average surface distance (ASD) between the prediction and reference segmentation to evaluate the proposed method. Results: The DSC, HD95, and ASD of the target contours with the proposed method were 0.87 ± 0.04, 4.55 ± 2.18, and 1.41 ± 0.56, respectively. Compared with the traditional method without the synthetic CT assisted (0.86 ± 0.05, 5.17 ± 2.60, and 1.55 ± 0.72), the proposed method was outperformed, especially in the soft tissue target, such as the tumor bed region. Conclusion: The CLG model proposed in this study can create the high-quality sCT from low-quality CBCT and improve the performance of DIR between the CBCT and the pCT. The target segmentation accuracy is better than using the traditional DIR.

DeepSWI: Using Deep Learning to Enhance Susceptibility Contrast on T2*-Weighted MRI.

BACKGROUND: Although susceptibility-weighted imaging (SWI) is the gold standard for visualizing cerebral microbleeds (CMBs) in the brain, the required phase data are not always available clinically. Having a postprocessing tool for generating SWI contrast from T2*-weighted magnitude images is therefore advantageous. PURPOSE: To create synthetic SWI images from clinical T2*-weighted magnitude images using deep learning and evaluate the resulting images in terms of similarity to conventional SWI images and ability to detect radiation-associated CMBs. STUDY TYPE: Retrospective. POPULATION: A total of 145 adults (87 males/58 females; 43.9 years old) with radiation-associated CMBs were used to train (16,093 patches/121 patients), validate (484 patches/4 patients), and test (2420 patches/20 patients) our networks. FIELD STRENGTH/SEQUENCE: 3D T2*-weighted, gradient-echo acquired at 3 T. ASSESSMENT: Structural similarity index (SSIM), peak signal-to-noise-ratio (PSNR), normalized mean-squared-error (nMSE), CMB counts, and line profiles were compared among magnitude, original SWI, and synthetic SWI images. Three blinded raters (J.E.V.M., M.A.M., B.B. with 8-, 6-, and 4-years of experience, respectively) independently rated and classified test-set images. STATISTICAL TESTS: Kruskall-Wallis and Wilcoxon signed-rank tests were used to compare SSIM, PSNR, nMSE, and CMB counts among magnitude, original SWI, and predicted synthetic SWI images. Intraclass correlation assessed interrater variability. P values <0.005 were considered statistically significant. RESULTS: SSIM values of the predicted vs. original SWI (0.972, 0.995, 0.9864) were statistically significantly higher than that of the magnitude vs. original SWI (0.970, 0.994, 0.9861) for whole brain, vascular structures, and brain tissue regions, respectively; 67% (19/28) CMBs detected on original SWI images were also detected on the predicted SWI, whereas only 10 (36%) were detected on magnitude images. Overall image quality was similar between the synthetic and original SWI images, with less artifacts on the former. CONCLUSIONS: This study demonstrated that deep learning can increase the susceptibility contrast present in neurovasculature and CMBs on T2*-weighted magnitude images, without residual susceptibility-induced artifacts. This may be useful for more accurately estimating CMB burden from magnitude images alone. EVIDENCE LEVEL: 3. TECHNICAL EFFICACY: Stage 2.

Sea Mine Detection Framework Using YOLO, SSD and EfficientDet Deep Learning Models.

In the context of new geopolitical tensions due to the current armed conflicts, safety in terms of navigation has been threatened due to the large number of sea mines placed, in particular, within the sea conflict areas. Additionally, since a large number of mines have recently been reported to have drifted into the territories of the Black Sea countries such as Romania, Bulgaria Georgia and Turkey, which have intense commercial and tourism activities in their coastal areas, the safety of those economic activities is threatened by possible accidents that may occur due to the above-mentioned situation. The use of deep learning in a military operation is widespread, especially for combating drones and other killer robots. Therefore, the present research addresses the detection of floating and underwater sea mines using images recorded from cameras (taken from drones, submarines, ships and boats). Due to the low number of sea mine images, the current research used both an augmentation technique and synthetic image generation (by overlapping images with different types of mines over water backgrounds), and two datasets were built (for floating mines and for underwater mines). Three deep learning models, respectively, YOLOv5, SSD and EfficientDet (YOLOv5 and SSD for floating mines and YOLOv5 and EfficientDet for underwater mines), were trained and compared. In the context of using three algorithm models, YOLO, SSD and EfficientDet, the new generated system revealed high accuracy in object recognition, namely the detection of floating and anchored mines. Moreover, tests carried out on portable computing equipment, such as Raspberry Pi, illustrated the possibility of including such an application for real-time scenarios, with the time of 2 s per frame being improved if devices use high-performance cameras.

A Review of Synthetic Image Data and Its Use in Computer Vision.

Development of computer vision algorithms using convolutional neural networks and deep learning has necessitated ever greater amounts of annotated and labelled data to produce high performance models. Large, public data sets have been instrumental in pushing forward computer vision by providing the data necessary for training. However, many computer vision applications cannot rely on general image data provided in the available public datasets to train models, instead requiring labelled image data that is not readily available in the public domain on a large scale. At the same time, acquiring such data from the real world can be difficult, costly to obtain, and manual labour intensive to label in large quantities. Because of this, synthetic image data has been pushed to the forefront as a potentially faster and cheaper alternative to collecting and annotating real data. This review provides general overview of types of synthetic image data, as categorised by synthesised output, common methods of synthesising different types of image data, existing applications and logical extensions, performance of synthetic image data in different applications and the associated difficulties in assessing data performance, and areas for further research.

Lumbar Spine Computed Tomography to Magnetic Resonance Imaging Synthesis Using Generative Adversarial Network: Visual Turing Test.

(1) Introduction: Computed tomography (CT) and magnetic resonance imaging (MRI) play an important role in the diagnosis and evaluation of spinal diseases, especially degenerative spinal diseases. MRI is mainly used to diagnose most spinal diseases because it shows a higher resolution than CT to distinguish lesions of the spinal canals and intervertebral discs. When it is inevitable for CT to be selected instead of MR in evaluating spinal disease, evaluation of spinal disease may be limited. In these cases, it is very helpful to diagnose spinal disease with MR images synthesized with CT images. (2) Objective: To create synthetic lumbar magnetic resonance (MR) images from computed tomography (CT) scans using generative adversarial network (GAN) models and assess how closely the synthetic images resembled the true images using visual Turing tests (VTTs). (3) Material and Methods: Overall, 285 patients aged ≥ 40 years who underwent lumbar CT and MRI were enrolled. Based on axial CT and T2-weighted axial MR images from 285 patients, an image synthesis model using a GAN was trained using three algorithms (unsupervised, semi-supervised, and supervised methods). Furthermore, VTT to determine how similar the synthetic lumbar MR images generated from lumbar CT axial images were to the true lumbar MR axial images were conducted with 59 patients who were not included in the model training. For the VTT, we designed an evaluation form comprising 600 randomly distributed axial images (150 true and 450 synthetic images from unsupervised, semi-supervised, and supervised methods). Four readers judged the authenticity of each image and chose their first- and second-choice candidates for the true image. In addition, for the three models, structural similarities (SSIM) were evaluated and the peak signal to noise ratio (PSNR) was compared among the three methods. (4) Results: The mean accuracy for the selection of true images for all four readers for their first choice was 52.0% (312/600). The accuracies of determining the true image for each reader's first and first + second choices, respectively, were as follows: reader 1, 51.3% and 78.0%; reader 2, 38.7% and 62.0%, reader 3, 69.3% and 84.0%, and reader 4, 48.7% and 70.7%. In the case of synthetic images chosen as first and second choices, supervised algorithm-derived images were the most often selected (supervised, 118/600 first and 164/600 second; semi-supervised, 90/600 and 144/600; and unsupervised, 80/600 and 114/600). For image quality, the supervised algorithm received the best score (PSNR: 15.987 ± 1.039, SSIM: 0.518 ± 0.042). (5) Conclusion: This was the pilot study to apply GAN to synthesize lumbar spine MR images from CT images and compare training algorithms of the GAN. Based on VTT, the axial MR images synthesized from lumbar CT using GAN were fairly realistic and the supervised training algorithm was found to provide the closest image to true images.

Domain randomization for neural network classification.

Large data requirements are often the main hurdle in training neural networks. Convolutional neural network (CNN) classifiers in particular require tens of thousands of pre-labeled images per category to approach human-level accuracy, while often failing to generalized to out-of-domain test sets. The acquisition and labelling of such datasets is often an expensive, time consuming and tedious task in practice. Synthetic data provides a cheap and efficient solution to assemble such large datasets. Using domain randomization (DR), we show that a sufficiently well generated synthetic image dataset can be used to train a neural network classifier that rivals state-of-the-art models trained on real datasets, achieving accuracy levels as high as 88% on a baseline cats vs dogs classification task. We show that the most important domain randomization parameter is a large variety of subjects, while secondary parameters such as lighting and textures are found to be less significant to the model accuracy. Our results also provide evidence to suggest that models trained on domain randomized images transfer to new domains better than those trained on real photos. Model performance appears to remain stable as the number of categories increases.

Discovering Digital Tumor Signatures-Using Latent Code Representations to Manipulate and Classify Liver Lesions.

Modern generative deep learning (DL) architectures allow for unsupervised learning of latent representations that can be exploited in several downstream tasks. Within the field of oncological medical imaging, we term these latent representations "digital tumor signatures" and hypothesize that they can be used, in analogy to radiomics features, to differentiate between lesions and normal liver tissue. Moreover, we conjecture that they can be used for the generation of synthetic data, specifically for the artificial insertion and removal of liver tumor lesions at user-defined spatial locations in CT images. Our approach utilizes an implicit autoencoder, an unsupervised model architecture that combines an autoencoder and two generative adversarial network (GAN)-like components. The model was trained on liver patches from 25 or 57 inhouse abdominal CT scans, depending on the experiment, demonstrating that only minimal data is required for synthetic image generation. The model was evaluated on a publicly available data set of 131 scans. We show that a PCA embedding of the latent representation captures the structure of the data, providing the foundation for the targeted insertion and removal of tumor lesions. To assess the quality of the synthetic images, we conducted two experiments with five radiologists. For experiment 1, only one rater and the ensemble-rater were marginally above the chance level in distinguishing real from synthetic data. For the second experiment, no rater was above the chance level. To illustrate that the "digital signatures" can also be used to differentiate lesion from normal tissue, we employed several machine learning methods. The best performing method, a LinearSVM, obtained 95% (97%) accuracy, 94% (95%) sensitivity, and 97% (99%) specificity, depending on if all data or only normal appearing patches were used for training of the implicit autoencoder. Overall, we demonstrate that the proposed unsupervised learning paradigm can be utilized for the removal and insertion of liver lesions at user defined spatial locations and that the digital signatures can be used to discriminate between lesions and normal liver tissue in abdominal CT scans.

Deep convolutional generative adversarial network for Alzheimer's disease classification using positron emission tomography (PET) and synthetic data augmentation.

With the evolution of deep learning technologies, computer vision-related tasks achieved tremendous success in the biomedical domain. For supervised deep learning training, we need a large number of labeled datasets. The task of achieving a large number of label dataset is a challenging. The availability of data makes it difficult to achieve and enhance an automated disease diagnosis model's performance. To synthesize data and improve the disease diagnosis model's accuracy, we proposed a novel approach for the generation of images for three different stages of Alzheimer's disease using deep convolutional generative adversarial networks. The proposed model out-perform in synthesis of brain positron emission tomography images for all three stages of Alzheimer disease. The three-stage of Alzheimer's disease is normal control, mild cognitive impairment, and Alzheimer's disease. The model performance is measured using a classification model that achieved an accuracy of 72% against synthetic images. We also experimented with quantitative measures, that is, peak signal-to-noise (PSNR) and structural similarity index measure (SSIM). We achieved average PSNR score values of 82 for AD, 72 for CN, and 73 for MCI and SSIM average score values of 25.6 for AD, 22.6 for CN, and 22.8 for MCI.

Evaluation of microcalcification contrast in clinical images for digital mammography and synthetic mammography.

PURPOSE: The aim of this work was to compare, in a clinical study, digital mammography and synthetic mammography imaging by evaluating the contrast in microcalcifications of different sizes. METHODS: A retrospective review of microcalcifications from 46 patients was undertaken. A Hologic 3-Dimensions mammography system and a HD Combo protocol was used for simultaneous acquisition of the digital and synthetic images. Microcalcifications were classified in accordance with their size, and patient breast images were classified in accordance with their density as adipose, moderately dense and dense. The contrast of the microcalcifications was measured and the contrast ratio between synthetic and digital images was compared. An additional qualitative assessment of the images was presented to correlate the conspicuity of the microcalcifications with the suppression of the structure noise. RESULTS: Microcalcifications in adipose background always exhibit a comparable or better contrast on synthetic images, regardless their size. For moderately dense background, synthetic images show a better contrast in 91.2 % of cases for small microcalcifications and in 90.9 % of cases for large microcalcifications. For a dense background, better contrast is seen in 89.5 % of cases for small microcalcifications, and in 85.7 % of cases for large microcalcifications. The contrast ratio increases with increasing breast glandularity. The suppression of structure noise also contributes to the enhancement of microcalcifications in the synthetic images. CONCLUSIONS: Synthetic mammography imaging is superior to digital mammography imaging in terms of microcalcification contrast, regardless their size and breast density.

Cross-modality deep learning: Contouring of MRI data from annotated CT data only.

PURPOSE: Online adaptive radiotherapy would greatly benefit from the development of reliable auto-segmentation algorithms for organs-at-risk and radiation targets. Current practice of manual segmentation is subjective and time-consuming. While deep learning-based algorithms offer ample opportunities to solve this problem, they typically require large datasets. However, medical imaging data are generally sparse, in particular annotated MR images for radiotherapy. In this study, we developed a method to exploit the wealth of publicly available, annotated CT images to generate synthetic MR images, which could then be used to train a convolutional neural network (CNN) to segment the parotid glands on MR images of head and neck cancer patients. METHODS: Imaging data comprised 202 annotated CT and 27 annotated MR images. The unpaired CT and MR images were fed into a 2D CycleGAN network to generate synthetic MR images from the CT images. Annotations of axial slices of the synthetic images were generated by propagating the CT contours. These were then used to train a 2D CNN. We assessed the segmentation accuracy using the real MR images as test dataset. The accuracy was quantified with the 3D Dice similarity coefficient (DSC), Hausdorff distance (HD), and mean surface distance (MSD) between manual and auto-generated contours. We benchmarked the approach by a comparison to the interobserver variation determined for the real MR images, as well as to the accuracy when training the 2D CNN to segment the CT images. RESULTS: The determined accuracy (DSC: 0.77±0.07, HD: 18.04±12.59mm, MSD: 2.51±1.47mm) was close to the interobserver variation (DSC: 0.84±0.06, HD: 10.85±5.74mm, MSD: 1.50±0.77mm), as well as to the accuracy when training the 2D CNN to segment the CT images (DSC: 0.81±0.07, HD: 13.00±7.61mm, MSD: 1.87±0.84mm). CONCLUSIONS: The introduced cross-modality learning technique can be of great value for segmentation problems with sparse training data. We anticipate using this method with any nonannotated MRI dataset to generate annotated synthetic MR images of the same type via image style transfer from annotated CT images. Furthermore, as this technique allows for fast adaptation of annotated datasets from one imaging modality to another, it could prove useful for translating between large varieties of MRI contrasts due to differences in imaging protocols within and between institutions.

Using deep learning to generate synthetic B-mode musculoskeletal ultrasound images.

BACKGROUND AND OBJECTIVE: Deep learning approaches are common in image processing, but often rely on supervised learning, which requires a large volume of training images, usually accompanied by hand-crafted labels. As labelled data are often not available, it would be desirable to develop methods that allow such data to be compiled automatically. In this study, we used a Generative Adversarial Network (GAN) to generate realistic B-mode musculoskeletal ultrasound images, and tested the suitability of two automated labelling approaches. METHODS: We used a model including two GANs each trained to transfer an image from one domain to another. The two inputs were a set of 100 longitudinal images of the gastrocnemius medialis muscle, and a set of 100 synthetic segmented masks that featured two aponeuroses and a random number of 'fascicles'. The model output a set of synthetic ultrasound images and an automated segmentation of each real input image. This automated segmentation process was one of the two approaches we assessed. The second approach involved synthesising ultrasound images and then feeding these images into an ImageJ/Fiji-based automated algorithm, to determine whether it could detect the aponeuroses and muscle fascicles. RESULTS: Histogram distributions were similar between real and synthetic images, but synthetic images displayed less variation between samples and a narrower range. Mean entropy values were statistically similar (real: 6.97, synthetic: 7.03; p = 0.218), but the range was much narrower for synthetic images (6.91 - 7.11 versus 6.30 - 7.62). When comparing GAN-derived and manually labelled segmentations, intersection-over-union values- denoting the degree of overlap between aponeurosis labels- varied between 0.0280 - 0.612 (mean ± SD: 0.312 ± 0.159), and pennation angles were higher for the GAN-derived segmentations (25.1° vs. 19.3°; p < 0.001). For the second segmentation approach, the algorithm generally performed equally well on synthetic and real images, yielding pennation angles within the physiological range (13.8-20°). CONCLUSIONS: We used a GAN to generate realistic B-mode ultrasound images, and extracted muscle architectural parameters from these images automatically. This approach could enable generation of large labelled datasets for image segmentation tasks, and may also be useful for data sharing. Automatic generation and labelling of ultrasound images minimises user input and overcomes several limitations associated with manual analysis.

SpeckleGAN: a generative adversarial network with an adaptive speckle layer to augment limited training data for ultrasound image processing.

PURPOSE: In the field of medical image analysis, deep learning methods gained huge attention over the last years. This can be explained by their often improved performance compared to classic explicit algorithms. In order to work well, they need large amounts of annotated data for supervised learning, but these are often not available in the case of medical image data. One way to overcome this limitation is to generate synthetic training data, e.g., by performing simulations to artificially augment the dataset. However, simulations require domain knowledge and are limited by the complexity of the underlying physical model. Another method to perform data augmentation is the generation of images by means of neural networks. METHODS: We developed a new algorithm for generation of synthetic medical images exhibiting speckle noise via generative adversarial networks (GANs). Key ingredient is a speckle layer, which can be incorporated into a neural network in order to add realistic and domain-dependent speckle. We call the resulting GAN architecture SpeckleGAN. RESULTS: We compared our new approach to an equivalent GAN without speckle layer. SpeckleGAN was able to generate ultrasound images with very crisp speckle patterns in contrast to the baseline GAN, even for small datasets of 50 images. SpeckleGAN outperformed the baseline GAN by up to 165 % with respect to the Fréchet Inception distance. For artery layer and lumen segmentation, a performance improvement of up to 4 % was obtained for small datasets, when these were augmented with images by SpeckleGAN. CONCLUSION: SpeckleGAN facilitates the generation of realistic synthetic ultrasound images to augment small training sets for deep learning based image processing. Its application is not restricted to ultrasound images but could be used for every imaging methodology that produces images with speckle such as optical coherence tomography or radar.

Active dynamic thermography to detect the presence of stenosis in the carotid artery.

Unlike passive infrared (IR) thermal imaging/thermography, where no external stimulation is applied, active dynamic thermography (ADT) results in a high contrast thermal image. In ADT, transient thermal images of the skin surface are captured using an IR thermal camera while the skin surface is stimulated externally, followed by a recovery phase. Upon the application of external stimulation, the presence of stenosis in the carotid artery is expected to differ the recovery rate of the external neck skin surface from the case with no stenosis. In this prospective study, using an external cooling stimulation, the ADT procedure was performed on a total of 54 (N) samples (C: N = 19, 0% stenosis; D1: N = 17, 10%-29% stenosis; D2: N = 18, ≥30% stenosis using Duplex Ultrasound). Analyzing the ADT sequence with a parameter called tissue activity ratio (TAR), the samples were classified using a cut-off value: C versus (D1+D2) and (C + D1) versus D2. As the degree of stenosis increases, the value of the TAR parameter depreciates with a significant difference among the sample groups (C:0.97 ± 0.05, D1:0.80 ± 0.04, D2:0.75 ± 0.02; p < 0.05). Under the two classification scenarios, classification accuracies of 90% and 85%, respectively, were achieved. This study suggests the potential of screening CAS with the proposed ADT procedure.

Image generation by GAN and style transfer for agar plate image segmentation.

BACKGROUND AND OBJECTIVES: Deep learning models and specifically Convolutional Neural Networks (CNNs) are becoming the leading approach in many computer vision tasks, including medical image analysis. Nevertheless, the CNN training usually requires large sets of supervised data, which are often difficult and expensive to obtain in the medical field. To address the lack of annotated images, image generation is a promising method, which is becoming increasingly popular in the computer vision community. In this paper, we present a new approach to the semantic segmentation of bacterial colonies in agar plate images, based on deep learning and synthetic image generation, to increase the training set size. Indeed, semantic segmentation of bacterial colony is the basis for infection recognition and bacterial counting in Petri plate analysis. METHODS: A convolutional neural network (CNN) is used to separate the bacterial colonies from the background. To face the lack of annotated images, a novel engine is designed - which exploits a generative adversarial network to capture the typical distribution of the bacterial colonies on agar plates - to generate synthetic data. Then, bacterial colony patches are superimposed on existing background images, taking into account both the local appearance of the background and the intrinsic opacity of the bacterial colonies, and a style transfer algorithm is used for further improve visual realism. RESULTS: The proposed deep learning approach has been tested on the only public dataset available with pixel-level annotations for bacterial colony semantic segmentation in agar plates. The role of including synthetic data in the training of a segmentation CNN has been evaluated, showing how comparable performances can be obtained with respect to the use of real images. Qualitative results are also reported for a second public dataset in which the segmentation annotations are not provided. CONCLUSIONS: The use of a small set of real data, together with synthetic images, allows obtaining comparable results with respect to using a complete set of real images. Therefore, the proposed synthetic data generator is able to address the scarcity of biomedical data and provides a scalable and cheap alternative to human ground-truth supervision.