Image Domain Research Articles

Protocol optimization for functional cardiac CT imaging using noise emulation in the raw data domain.

Four-dimensional (4D) wide coverage computed tomography (CT) is an effective imaging modality for measuring the mechanical function of the myocardium. However, repeated CT measurement across a number of heartbeats is still a concern. A projection-domain noise emulation method is presented to generate accurate low-dose (mA modulated) 4D cardiac CT scans from high-dose scans, enabling protocol optimization to deliver sufficient image quality for functional cardiac analysis while using a dose level that is as low as reasonably achievable (ALARA). Given a targeted low-dose mA modulation curve, the proposed noise emulation method injects both quantum and electronic noise of proper magnitude and correlation to the high-dose data in projection domain. A spatially varying (i.e., channel-dependent) detector gain term as well as its calibration method were proposed to further improve the noise emulation accuracy. To determine the ALARA dose threshold, a straightforward projection domain image quality (IQ) metric was proposed that is based on the number of projection rays that do not fall under the non-linear region of the detector response. Experiments were performed to validate the noise emulation method with both phantom and clinical data in terms of visual similarity, contrast-to-noise ratio (CNR), and noise-power spectrum (NPS). For both phantom and clinical data, the low-dose emulated images exhibited similar noise magnitude (CNR difference within 2%), artifacts, and texture to that of the real low-dose images. The proposed channel-dependent detector gain term resulted in additional increase in emulation accuracy. Using the proposed IQ metric, recommended kVp and mA settings were calculated for low dose 4D Cardiac CT acquisitions for patients of different sizes. A detailed method to estimate system-dependent parameters for a raw-data based low dose emulation framework was described. The method produced realistic noise levels, artifacts, and texture with phantom and clinical studies. The proposed low-dose emulation method can be used to prospectively select patient-specific minimal-dose protocols for functional cardiac CT.

Integrated approach of federated learning with transfer learning for classification and diagnosis of brain tumor

Brain tumor classification using MRI images is a crucial yet challenging task in medical imaging. Accurate diagnosis is vital for effective treatment planning but is often hindered by the complex nature of tumor morphology and variations in imaging. Traditional methodologies primarily rely on manual interpretation of MRI images, supplemented by conventional machine learning techniques. These approaches often lack the robustness and scalability needed for precise and automated tumor classification. The major limitations include a high degree of manual intervention, potential for human error, limited ability to handle large datasets, and lack of generalizability to diverse tumor types and imaging conditions.To address these challenges, we propose a federated learning-based deep learning model that leverages the power of Convolutional Neural Networks (CNN) for automated and accurate brain tumor classification. This innovative approach not only emphasizes the use of a modified VGG16 architecture optimized for brain MRI images but also highlights the significance of federated learning and transfer learning in the medical imaging domain. Federated learning enables decentralized model training across multiple clients without compromising data privacy, addressing the critical need for confidentiality in medical data handling. This model architecture benefits from the transfer learning technique by utilizing a pre-trained CNN, which significantly enhances its ability to classify brain tumors accurately by leveraging knowledge gained from vast and diverse datasets.Our model is trained on a diverse dataset combining figshare, SARTAJ, and Br35H datasets, employing a federated learning approach for decentralized, privacy-preserving model training. The adoption of transfer learning further bolsters the model’s performance, making it adept at handling the intricate variations in MRI images associated with different types of brain tumors. The model demonstrates high precision (0.99 for glioma, 0.95 for meningioma, 1.00 for no tumor, and 0.98 for pituitary), recall, and F1-scores in classification, outperforming existing methods. The overall accuracy stands at 98%, showcasing the model’s efficacy in classifying various tumor types accurately, thus highlighting the transformative potential of federated learning and transfer learning in enhancing brain tumor classification using MRI images.

Dual-domain faster Fourier convolution based network for MR image reconstruction

Deep learning methods for fast MRI have shown promise in reconstructing high-quality images from undersampled multi-coil k-space data, leading to reduced scan duration. However, existing methods encounter challenges related to limited receptive fields in dual-domain (k-space and image domains) reconstruction networks, rigid data consistency operations, and suboptimal refinement structures, which collectively restrict overall reconstruction performance. This study introduces a comprehensive framework that addresses these challenges and enhances MR image reconstruction quality. Firstly, we propose Faster Inverse Fourier Convolution (FasterIFC), a frequency domain convolutional operator that significantly expands the receptive field of k-space domain reconstruction networks. Expanding the information extraction range to the entire frequency spectrum according to the spectral convolution theorem in Fourier theory enables the network to easily utilize richer redundant long-range information from adjacent, symmetrical, and diagonal locations of multi-coil k-space data. Secondly, we introduce a novel softer Data Consistency (softerDC) layer, which achieves an enhanced balance between data consistency and smoothness. This layer facilitates the implementation of diverse data consistency strategies across distinct frequency positions, addressing the inflexibility observed in current methods. Finally, we present the Dual-Domain Faster Fourier Convolution Based Network (D2F2), which features a centrosymmetric dual-domain parallel structure based on FasterIFC. This architecture optimally leverages dual-domain data characteristics while substantially expanding the receptive field in both domains. Coupled with the softerDC layer, D2F2 demonstrates superior performance on the NYU fastMRI dataset at multiple acceleration factors, surpassing state-of-the-art methods in both quantitative and qualitative evaluations.

Research on Multi-Scale Fusion Method for Ancient Bronze Ware X-ray Images in NSST Domain

X-ray imaging is a valuable non-destructive tool for examining bronze wares, but the complexity of the coverings of bronze wares and the limitations of single-energy imaging techniques often obscure critical details, such as lesions and ornamentation. Therefore, multiple imaging is required to fully present the key information of bronze artifacts, which affects the complete presentation of information and increases the difficulty of analysis and interpretation. Using high-performance image fusion technology to fuse X-ray images of different energies into one image can effectively solve this problem. However, there is currently no specialized method for the fusion of images of bronze artifacts. Considering the special requirements for the restoration of bronze artifacts and the existing fusion framework, this paper proposes a new method. It is a novel multi-scale morphological gradient and local topology-coupled neural P systems approach within the Non-Subsampled Shearlet Transform domain. It addresses the absence of a specialized method for image fusion of bronze artifacts. The method proposed in this paper is compared with eight high-performance fusion methods and validated using a total of six evaluation metrics. The results demonstrate the significant theoretical and practical potential of this method for advancing the analysis and preservation of cultural heritage artifacts.

A Sparse Recovery Algorithm for Suppressing Multiple Linear Frequency Modulation Interference in the Synthetic Aperture Radar Image Domain.

In synthetic aperture radar (SAR) signal processing, compared with the raw data of level-0, level-1 SAR images are more readily accessible and available in larger quantities. However, an amount of level-1 images are affected by radio frequency interference (RFI), which typically originates from Linear Frequency Modulation (LFM) signals emitted by ground-based radars. Existing research on interference suppression in level-1 data has primarily focused on two methods: transforming SAR images into simulated echo data for interference suppression, or focusing interference in the frequency domain and applying notching filters to reduce interference energy. However, these methods overlook the effective utilization of the interference parameters or are confined to suppressing only one type of LFM interference at a time. In certain SAR images, multiple types of LFM interference manifest bright radiation artifacts that exhibit varying lengths along the range direction while remaining constant in the azimuth direction. It is necessary to suppress multiple LFM interference on SAR images when original echo data are unavailable. This article proposes a joint sparse recovery algorithm for interference suppression in the SAR image domain. In the SAR image domain, two-dimensional LFM interference typically exhibits differences in parameters such as frequency modulation rate and pulse width in the range direction, while maintaining consistency in the azimuth direction. Based on this observation, this article constructs a series of focusing operators for LFM interference in SAR images. These operators enable the sparse representation of dispersed LFM interference. Subsequently, an optimization model is developed that can effectively suppress multi-LFM interference and reduce image loss with the assistance of a regularization term in the image domain. Simulation experiments conducted in various scenarios validate the superior performance of the proposed method.

Efficient Classification on Remote Sensing Image Using Transfer Learning

With a rapid development in aerial technology, applications of Remote Sensing Images (RSI) have become more diverse. Remote sensing image classification plays a crucial role in analyzing and interpreting Earth observation data for various applications, such as land cover mapping, environmental monitoring, and urban planning. However, accurately classifying remote sensing images poses significant challenges due to their complex spatial and spectral characteristics. In recent years, transfer learning has emerged as a promising technique to improve the classification accuracy by leveraging the knowledge learned from pre-trained models on large-scale datasets. The proposed model explores different transfer learning strategies employed in remote sensing image classification, including fine-tuning, feature extraction, and domain adaptation. It discusses popular pre-trained models, such as VGG16, VGG19, and Inceptionv3, and their applicability to remote sensing datasets. The advantages and limitations of each strategy are analyzed, providing insights into their suitability for various remote sensing applications. A comparative study is done on all these techniques to evaluate the performance measures like Accuracy and Loss.

Enhancing online dose distribution reconstruction in radiotherapy through sparse sampled rotating fiber arrays and deep learning

A optical fiber array dosimeter based on radiation-induced luminescence can be used in two-dimensional dose distribution measurements. However, using rotating fiber arrays introduces challenges in terms of online monitoring and variations in detector performance due to the long data acquisition time. This study aims to propose and demonstrate a novel method for online fast reconstruction of two-dimensional dose distributions in radiation therapy using projection data obtained from sparsely sampled rotating fiber arrays. The projection data were processed in both the projection and image domains to achieve restoration and noise reduction by deep learning techniques. The performance of the proposed method was evaluated using simulation data obtained by Monte Carlo toolkit Geant4 and experimental data obtained from a medical accelerator. Results showed that the method can successfully reconstruct dose distribution through sparse sampling and reduce the measurement time to one-sixth of the original. The peak signal-to-noise ratio, root mean square error, and structural similarity index were 23.72, 2.68, and 0.95, respectively. Additionally, the 3%/3 mm gamma pass rate reached 98.4% under specific irradiation conditions with a dose of at least 10% of the maximum dose. The integration of sparse sampling techniques, rotating fiber arrays, and deep learning techniques offers new possibilities for the fast and efficient reconstruction of two-dimensional dose distributions, ensuring high-quality radiotherapy outcomes.

Effect of growth conditions on magnetization reversal and magnetic anisotropy in Co2Fe0.5Ti0.5Si quaternary Heusler alloy thin films

Ferromagnetic resonance and magneto-optic Kerr effect (MOKE) techniques are employed to unravel the nature of ‘in-plane’ (IP) magnetic anisotropy and magnetization reversal (MR) processes in magnetron-sputtered 100 nm Co2Fe0.5Ti0.5Si (CFTS) thin films, deposited (and subsequently annealed) at different substrate temperatures (Ts) ranging from 200 °C to 550 °C. By varying TS, the CFTS films are produced with different amounts of anti-site (AS) atomic disorder. Irrespective of the degree of AS disorder, the IP uniaxial magnetic anisotropy (UMA) is prevalent in all the CFTS films. The TS450 and TS500 films, deposited at T S = 450 °C and 500 °C, stand out as they have (i) the least AS disorder, (ii) lowest value (α = 0.0055) of the Gilbert damping constant, (iii) high saturation magnetization ( ≅770 G ) at 300 K, (iv) UMA energy density as high as ≅1.6×104erg/cc at 120 K dropping to ≅1.0×104erg/cc at 300 K, (v) spin-wave stiffness ( D ) at T = 0 K, D0≅175 meVÅ2 and, as in other CFTS films, the electron–magnon interaction is primarily responsible for the thermal renormalization of D . Furthermore, in the TS450 and TS500 films, MOKE hysteresis loops at various angles ( ψH ) that the field makes with the easy axis in the film plane, reveals two mutually exclusive UMAs, UMA1≫UMA2 , with easy axes perpendicular to each other. By contrast, only a single UMA is observed in all the remaining films. When ψH=0∘ , MOKE domain images reveal that, consistent with the uniaxial nature of magnetic anisotropy, an abrupt reversal of magnetization is accompanied by the formation of 180° domains. When ψH = 90°, due to the presence of two UMAs in TS450 and TS500 films, following the field reversal, first the reverse domains nucleate and then grow at the expense of the fundamental domains through field-induced domain wall motion.

Advancing zero-shot semantic segmentation through attribute correlations

Zero-shot semantic segmentation aims to segment novel classes that have not been encountered during the training phase. Existing methods leverage available text features obtained from pretrained language models to produce semantic segmentation results for both base and novel classes. However, the text-based feature-producing paradigm only provides insufficient class correlations and limits the full exploitation of image features from base classes. Besides, there exists a non-negligible domain gap between the text and image domains, resulting in severe feature bias during feature production. Different from existing methods, we advance the zero-shot semantic segmentation through attribute correlations. Specifically, we introduce a set of shared-attribute labels, of which the design fully considers the structural relations between attributes and classes, to provide rational and sufficient attribute-class correlations. Besides, due to the minor intra-class variations of shared attributes, the text features are more easily mapped to image features, thereby alleviating the domain gap issue. Furthermore, we propose a hierarchical semantic segmentation framework incorporating an attribute prompt tuning method. This approach is designed to enhance the model’s adaptation to the attribute segmentation task and effectively leverage attribute features to produce better semantic segmentation results. Correspondingly, we construct a Visual Hierarchical Semantic Classes (VHSC) benchmark, meticulously annotating shared-attributes at the pixel level to conduct the experiments. Extensive experiments on the VHSC benchmark showcase the superior performance of our method compared to existing zero-shot semantic segmentation methods, achieving mIoU of 73.0% and FBIoU of 87.5%. The VHSC benchmark and our code are available at https://github.com/RuntongZHANG/ZeroShot-VHSC.

Leveraging diffusion models for unsupervised out-of-distribution detection on image manifold.

Out-of-distribution (OOD) detection is crucial for enhancing the reliability of machine learning models when confronted with data that differ from their training distribution. In the image domain, we hypothesize that images inhabit manifolds defined by latent properties such as color, position, and shape. Leveraging this intuition, we propose a novel approach to OOD detection using a diffusion model to discern images that deviate from the in-domain distribution. Our method involves training a diffusion model using in-domain images. At inference time, we lift an image from its original manifold using a masking process, and then apply a diffusion model to map it towards the in-domain manifold. We measure the distance between the original and mapped images, and identify those with a large distance as OOD. Our experiments encompass comprehensive evaluation across various datasets characterized by differences in color, semantics, and resolution. Our method demonstrates strong and consistent performance in detecting OOD images across the tested datasets, highlighting its effectiveness in handling images with diverse characteristics. Additionally, ablation studies confirm the significant contribution of each component in our framework to the overall performance.

Vectorial adaptive optics for advanced imaging systems

Vectorial adaptive optics (V-AO) is a cutting-edge technique extending conventional AO into the vectorial domain encompassing both polarization and phase feedback correction for optical systems. However, previous V-AO approaches focus on point correction. In this letter, we extend this AO approach into the imaging domain. We show how V-AO can benefit an aberrated imaging system to enhance not only scalar imaging but also the quality of vectorial information. Two important criteria, vectorial precision and uniformity are put forward and used in practice to evaluate the performance of the correction. These experimental validations pave the way for real-world imaging for V-AO technology and its applications.

A Statistical and AI Analysis of the Frequency Spectrum in the Measurement of the Center of Pressure Track in the Seated Position in Healthy Subjects and Subjects with Low Back Pain.

Measuring postural control in an upright standing position is the standard method. However, this diagnostic method has floor or ceiling effects and its implementation is only possible to a limited extent. Assessing postural control directly on the trunk in a sitting position and consideration of the results in the spectrum in conjunction with an AI-supported evaluation could represent an alternative diagnostic method quantifying neuromuscular control. In a prospective cross-sectional study, 188 subjects aged between 18 and 60 years were recruited and divided into two groups: "LowBackPain" vs. "Healthy". Subsequently, measurements of postural control in a seated position were carried out for 60 s using a modified balance board. A spectrum per trail was calculated using the measured CoP tracks in the range from 0.01 to 10 Hz. Various algorithms for data classification and prediction of these classes were tested for the parameter combination with the highest proven static influence on the parameter pain. The best results were found in a frequency spectrum of 0.001 Hz and greater than 1 Hz. After transforming the track from the time domain to the image domain for representation as power density, the influence of pain was highly significant (effect size 0.9). The link between pain and gender (p = 0.015) and pain and height (p = 0.012) also demonstrated significant results. The assessment of postural control in a seated position allows differentiation between "LowBackPain" and "Healthy" subjects. Using the AI algorithm of neural networks, the data set can be correctly differentiated into "LowBackPain" and "Healthy" with a probability of 81%.

Characterization of pulsations in the brain and cerebrospinal fluid using ultra-high field magnetic resonance imaging.

The development of innovative non-invasive neuroimaging methods and biomarkers is critical for studying brain disease. Imaging of cerebrospinal fluid (CSF) pulsatility may inform the brain fluid dynamics involved in clearance of cerebral metabolic waste. In this work, we developed a methodology to characterize the frequency and spatial localization of whole brain CSF pulsations in humans. Using 7 Tesla (T) human magnetic resonance imaging (MRI) and ultrafast echo-planar imaging (EPI), in-vivo images were obtained to capture pulsations of the CSF signal. Physiological data were simultaneously collected and compared with the 7 T MR data. The primary components of signal pulsations were identified using spectral analysis, with the most evident frequency bands identified around 0.3, 1.2, and 2.4 Hz. These pulsations were mapped spatially and temporally onto the MR image domain and temporally onto the physiological measures of electrocardiogram and respiration. We identified peaks in CSF pulsations that were distinct from peaks in grey matter and white matter regions. This methodology may provide novel in vivo biomarkers of disrupted brain fluid dynamics.

GMIM: Self-supervised pre-training for 3D medical image segmentation with adaptive and hierarchical masked image modeling

Self-supervised pre-training and fully supervised fine-tuning paradigms have received much attention to solve the data annotation problem in deep learning fields. Compared with traditional pre-training on large natural image datasets, medical self-supervised learning methods learn rich representations derived from unlabeled data itself thus avoiding the distribution shift between different image domains. However, nowadays state-of-the-art medical pre-training methods were specifically designed for downstream tasks making them less flexible and difficult to apply to new tasks. In this paper, we propose grid mask image modeling, a flexible and general self-supervised method to pre-train medical vision transformers for 3D medical image segmentation. Our goal is to guide networks to learn the correlations between organs and tissues by reconstructing original images based on partial observations. The relationships are consistent within the human body and invariant to disease type or imaging modality. To achieve this, we design a Siamese framework consisting of an online branch and a target branch. An adaptive and hierarchical masking strategy is employed in the online branch to (1) learn the boundaries or small contextual mutation regions within images; (2) to learn high-level semantic representations from deeper layers of the multiscale encoder. In addition, the target branch provides representations for contrastive learning to further reduce representation redundancy. We evaluate our method through segmentation performance on two public datasets. The experimental results demonstrate our method outperforms other self-supervised methods. Codes are available at https://github.com/mobiletomb/Gmim.

Unsupervised low-dose CT denoising using bidirectional contrastive network

Background and Objective:Low-dose computed tomography (LDCT) scans significantly reduce radiation exposure, but introduce higher levels of noise and artifacts that compromise image quality and diagnostic accuracy. Supervised learning methods have proven effective in denoising LDCT images, but are hampered by the need for large, paired datasets, which pose significant challenges in data acquisition. This study aims to develop a robust unsupervised LDCT denoising method that overcomes the reliance on paired LDCT and normal-dose CT (NDCT) samples, paving the way for more accessible and practical denoising techniques. Methods:We propose a novel unsupervised network model, Bidirectional Contrastive Unsupervised Denoising (BCUD), for LDCT denoising. This model innovatively combines a bidirectional network structure with contrastive learning theory to map the precise mutual correspondence between the noisy LDCT image domain and the clean NDCT image domain. Specifically, we employ dual encoders and discriminators for domain-specific data generation, and use unique projection heads for each domain to adaptively learn customized embedded representations. We then align corresponding features across domains within the learned embedding spaces to achieve effective noise reduction. This approach fundamentally improves the model’s ability to match features in latent space, thereby improving noise reduction while preserving fine image detail. Results:Through extensive experimental validation on the AAPM-Mayo public dataset and real-world clinical datasets, the proposed BCUD method demonstrated superior performance. It achieved a peak signal-to-noise ratio (PSNR) of 31.387 dB, a structural similarity index measure (SSIM) of 0.886, an information fidelity criterion (IFC) of 2.305, and a visual information fidelity (VIF) of 0.373. Notably, subjective evaluation by radiologists resulted in a mean score of 4.23, highlighting its advantages over existing methods in terms of clinical applicability. Conclusions:This paper presents an innovative unsupervised LDCT denoising method using a bidirectional contrastive network, which greatly improves clinical applicability by eliminating the need for perfectly matched image pairs. The method sets a new benchmark in unsupervised LDCT image denoising, excelling in noise reduction and preservation of fine structural details.

Watermarking algorithm based on phase-only CGH in fractional Hartley domain for DICOM images

In this manuscript, we proposed a watermarking algorithm based on phase-only computer-generated holography (CGH) in the fractional Hartley domain for digital imaging and communications in medicine (DICOM) images. The proposed algorithm improves the security of the CGH-based algorithm. The cascaded use of fractional Hartley transform and attenuation factor β increases the keyspace of the proposed watermarking algorithm. The robustness and effectiveness of the proposed watermarking algorithm is validated using simulations on DICOM images. The effectiveness of the proposed watermarking algorithm is assessed using statistical tools in terms of mean-squared error, information entropy, correlation coefficient, histogram, and mesh plots. The robustness is evaluated by testing the proposed algorithm’s performance under real-time threats, including contamination and data loss attacks. Furthermore, the security of the proposed algorithm is also tested for existing cryptographic attacks, such as chosen-plaintext attacks, and known-plaintext attacks. The simulation results indicate that the proposed watermarking algorithm is robust and effective.

FPM-WSI: Fourier ptychographic whole slide imaging via feature-domain backdiffraction

Fourier ptychographic microscopy (FPM) theoretically provides a solution to the trade-off between spatial resolution and field of view (FOV), and has promising prospects in digital pathology. However, block reconstruction and then stitching has become an unavoidable procedure for reconstruction of large FOV due to vignetting artifacts. This introduces digital stitching artifacts, as the existing image-domain optimization algorithms are highly sensitive to systematic errors. Such obstacles significantly impede the advancement and practical implementation of FPM, explaining why, despite a decade of development, FPM has not gained widespread recognition in the field of biomedicine. We report a feature-domain FPM (FD-FPM) based on the structure-aware forward model to realize stitching-free, full-FOV reconstruction. The loss function is uniquely formulated in the feature domain of images, which bypasses the troublesome vignetting effect and algorithmic vulnerability via feature-domain backdiffraction. Through massive simulations and experiments, we show that FD-FPM effectively eliminates vignetting artifacts for full-FOV reconstruction, and still achieves impressive reconstructions despite the presence of various systematic errors. We also found it has great potential in recovering the data with a lower spectrum overlapping rate, and in realizing digital refocusing without a prior defocus distance. With FD-FPM, we achieved full-color and high-throughput imaging (4.7 mm diameter FOV, 336 nm resolution in the blue channel) free of blocking-and-stitching procedures on a self-developed Fourier ptychographic microscopy whole slide imaging platform. The reported FD-FPM shows the value of FPM for various experimental circumstances, and offers physical insights useful for the developments of models for other computational imaging techniques. The reported platform demonstrates high-quality, high-speed imaging and low cost, and could find applications in many fields of biomedical research, as well as in clinical applications.

Diff-IF: Multi-modality image fusion via diffusion model with fusion knowledge prior

Multi-modality image fusion (MMIF) aims to aggregate the complementary information from diverse source image domains. As the generative adversarial network-based methods have been the primary choice and demonstrated satisfactory fusion performance, they suffer from the unstable training and mode collapse. To tackle this challenge, we propose a novel diffusion model for MMIF incorporating fusion knowledge prior, termed as Diff-IF. Diff-IF proposes a trainable diffusion model paradigm for multi-modality image fusion, resolving the issue of lacking the ground truth for the diffusion model in image fusion tasks. It decomposes the diffusion-based image fusion method into conditional diffusion model and fusion knowledge prior with the targeted search to derive the prior distribution for the specific image fusion task. In particular, the forward diffusion process is guided by the fusion knowledge prior distribution through targeted search, while the reverse diffusion process is designed to generate high-quality fused images. Extensive experiments demonstrate that Diff-IF achieves outstanding performance, including exemplary visual preservation, and good preservation of weak textures, across various MMIF tasks such as infrared-visible image fusion and medical image fusion. The code will be available at https://github.com/XunpengYi/Diff-IF.

HOG-CNN based evaluation of cement integrity using 2D dispersion curves from an experimental through tubing logging setup

In oil wells, adequate cementation is essential to guarantee that the well has support for the casing and is able to isolate the well from groundwater. Operations known as Plugging & Abandonment (P&A) must carry out inspections in the cement layer of the well to ensure that the bond quality of the cement with the formation and the hydraulic isolation is adequate before well sealing. The interpretation of data logs during the decommissioning processes is made exclusively for a petrophysicists team, which makes the process susceptible to human errors. Furthermore, it requires a lot of analysis time, increasing the search for alternative solutions that help reduce errors in the interpretation of data logs. So far, the use of machine learning has proven to be a strong candidate in this search. This paper deals with frequency domain images (dispersion curves and Cepstrum) that are acquired from acoustic pressure signals obtained in the time domain and along the vertical axis of an experimental setup. These images are used to train machine learning models to test and compare the accuracies obtained. An adjustment of the hyperparameters was carried out to find those that best describe the simulated experimental characteristics. Furthermore, the convolution neural network (CNN) model was combined with the oriented gradient histogram technique (HOG) to verify its efficiency by analyzing experimental data. Two different databases were used in this work, one with 130 samples relating to the average results of the four hydrophones used in the experiment and one with 520 samples relating to all results. The results found demonstrated that when using the dispersion curves as input with the CNN the accuracy hits 100%. Other combinations of models were also tested hitting similar results. This article contributes to the current research scenario as it presents an innovative combination of machine learning techniques applied to the oil and gas field (HOG-CNN) with the proposal to compare different types of databases, such as using Cepstrum instead of dispersion curves and image processing.

Stereo matching from monocular images using feature consistency

AbstractSynthetic images facilitate stereo matching. However, synthetic images may suffer from image distortion, domain bias, and stereo mismatch, which would significantly restrict the widespread use of stereo matching models in the real world. The first goal in this paper is to synthesize real‐looking images for minimizing the domain bias between the synthesized and real images. For this purpose, sharpened disparity maps are produced from a mono real image. Then, stereo image pairs are synthesized using these imperfect disparity maps and the single real image in the proposed pipeline. Although the synthesized images are as realistic as possible, the domain styles of the synthesized images are always very different from the real images. Thus, the second goal is to enhance the domain generalization ability of the stereo matching network. For that, the feature extraction layer is replaced with a teacher–student model. Then, a constraint of binocular contrast features is imposed on the output of the model. When tested on the KITTI, ETH3D, and Middlebury datasets, the accuracy of the method outperforms traditional methods by at least 30%. Experiments demonstrate that the approaches are general and can be conveniently embedded into existing stereo networks.