B-scan Images Research Articles

Prediction and Detection of Glaucomatous Visual Field Progression Using Deep Learning on Macular Optical Coherence Tomography.

A deep learning model trained on macular OCT imaging studies detected clinically significant functional glaucoma progression and was also able to predict future progression. To use macular optical coherence tomography (OCT) imaging to predict the future and detect concurrent visual field progression, respectively, using deep learning. A retrospective cohort study. A pretraining data set was comprised of 7,702,201 B-scan images from 151,389 macular OCT studies. The progression detection task included 3902 macular OCT imaging studies from 1534 eyes of 828 patients with glaucoma, and the progression prediction task included 1346 macular OCT studies from 1205 eyes of 784. A novel deep learning method was developed to detect glaucoma progression and predict future progression using macular OCT, based on self-supervised pretraining of a vision transformer (ViT) model on a large, unlabeled data set of OCT images. Glaucoma progression was defined as a mean deviation (MD) rate of change of ≤ -0.5 dB/year over 5 consecutive Humphrey visual field tests, and rapid progression was defined as MD change ≤ -1 dB/year. Diagnostic performance of the ViT model for prediction of future visual field progression and detection of concurrent visual field progression using area under the receiver operating characteristic curve (AUC), sensitivity, and specificity. The model distinguished stable eyes from progressing eyes, achieving an AUC of 0.90 (95% CI, 0.88-0.91). Rapid progression was detected with an AUC of 0.92 (95% CI, 0.91-0.93). The model also demonstrated high predictive ability for forecasting future glaucoma progression, with an AUC of 0.85 (95% CI 0.83-0.87). Rapid progression was predicted with an AUC of 0.84 (95% CI 0.81-0.86). A deep learning model detected clinically significant functional glaucoma progression using macular OCT imaging studies and was also able to predict future progression. Early identification of patients undergoing glaucoma progression or at high risk for future progression may aid in clinical decision-making.

A novel classification method for GPR B-scan images based on weak-shot learning

Ground Penetrating Radar (GPR) is extensively employed for underground target detection, and the advancements in this domain have been hastened by the utilization of deep learning methodologies. Despite advancements in deep learning, practical engineering applications often face challenges in meeting the demanding requirements of large-scale data and high data quality. In this study, based on cross-domain transfer learning, we proposed a method for underground target classification of weak labeled GPR data. It combined strongly annotated source domain data with weakly annotated target domain data through adversarial learning to leverage the prior knowledge from existing large datasets and addressed the classification problem of small-sample GPR data. By employing adversarial learning to integrate strongly labeled source domain data with weakly labeled target domain data, we leverage prior knowledge from existing large datasets, thereby enhancing classification accuracy on a small GPR dataset. This approach reduces the requirements for annotation quality and sample quantity in the GPR B-scan dataset of underground targets. Additionally, it exhibits robustness to label noise, making it applicable to real-world GPR data classification tasks. Compared to some well-known networks, the proposed model achieved the best classification performance in multiclass classification problems, with an accuracy of 87.62% and an F1macro score of 87.60% on the test set. These results represent an improvement of 17.14 and 17.20 percentage points, respectively, over the best-performing classical classification networks. Additionally, the proposed model's reliability in cross-domain transfer learning, even with limited data, was substantiated through a comparison with other few-shot learning networks.

On-line measurement of nugget diameter in automatic resistance spot welding based on embedded ultrasound probe

The inspection of resistance spot welding (RSW) quality is crucial in the sheet metal joining industries. This paper proposes an enhanced approach for the automated real-time measurement of weld nugget diameter in welded joints, utilizing an ultrasonic probe integrated within the electrode. The ultrasonic signals received by the probe are sequentially arranged to generate consecutive ultrasonic B-scan images, which effectively capture the entire weld nugget growth and solidification process. The corresponding maximum time of flight (TOF) for the liquid nugget is extracted from the B-scan image. In the condition of qualified welding quality, a mathematic equation between maximum TOF and weld nugget diameter is established, while the opposite is alarmed. The on-line quality inspection of RSW can autonomously determine the qualification of welding quality, as well as quantitatively evaluate the weld nugget diameter in qualified weld joints. The research is applicable to various processing and manufacturing fields.

Rebar Clutter Suppression and Road Defects Localization in GPR B-Scan Images Based on SuppRebar-GAN and EC-Yolov7 Networks

Rebar Clutter Suppression and Road Defects Localization in GPR B-Scan Images Based on SuppRebar-GAN and EC-Yolov7 Networks

GPR-Former: Detection and Parametric Reconstruction of Hyperbolas in GPR B-Scan Images With Transformers

GPR-Former: Detection and Parametric Reconstruction of Hyperbolas in GPR B-Scan Images With Transformers

Automated defect detection in precision forging ultrasonic images based on deep learning

Ultrasonic testing is a widely used non-destructive testing technique for precision forgings. However, assessing defects in ultrasonic B-scan images can be prone to errors, misses, and inefficiencies due to human judgment. To address these challenges, we propose a method based on deep learning to automate the evaluation of such images. We started by creating a dataset comprising 8000 images, each measuring 224 × 224 pixels. These images were cropped from ultrasonic B-scan images of 7 specimens, each featuring different sizes and locations of holes and crack defects. We then used state-of-the-art deep learning models to benchmark the dataset and identified YOLOv5s as the best-performing baseline model for our study. To address the challenges of deploying deep learning models and the issue of small defects being easily confused with the background in ultrasonic B-scan images, we made lightweight improvements to the deep learning model. Additionally, we enhanced the quality of data labels through data cleaning. Our experiments show that our method achieved a precision of 97.8%, a recall of 98.1%, mAP@0.5 of 99.0%, and mAP@.5:.95 of 67.6%, with a frames per second (FPS) of 74.5. Furthermore, the number of model parameters was reduced by 43.2%, while maintaining high detection accuracy. Overall, our proposed method offers a significant improvement over the original model, making it a more reliable and efficient tool for automated defect assessment in ultrasonic B-scan images.

Detection mechanism of delamination in thermal barrier coatings of turbine blade using a Rayleigh wave EMAT

Nondestructive testing (NDT) for damage in thermal barrier coatings (TBCs) is significant for the safety guarantee of gas turbine blades. As a new NDT technology, electromagnetic acoustic transducer (EMAT) is widely applied for NDT of conductive structural components due to its advantages of coupling-free and high adaptability. In this paper, numerical simulations are conducted to study the wave propagation and interaction with delamination defects in TBCs inspected with a Rayleigh wave EMAT of the Lorentz force mechanism. Based on the numerical results, the wave structure in TBC, wave conversion at delamination defect, time domain EMAT signals, and its B-scan images are evaluated and the feasibility of Rayleigh wave EMAT to inspect delamination in TBCs was theoretically clarified.

Deep Learning-Assisted Measurements of Photoreceptor Ellipsoid Zone Area and Outer Segment Volume as Biomarkers for Retinitis Pigmentosa.

The manual segmentation of retinal layers from OCT scan images is time-consuming and costly. The deep learning approach has potential for the automatic delineation of retinal layers to significantly reduce the burden of human graders. In this study, we compared deep learning model (DLM) segmentation with manual correction (DLM-MC) to conventional manual grading (MG) for the measurements of the photoreceptor ellipsoid zone (EZ) area and outer segment (OS) volume in retinitis pigmentosa (RP) to assess whether DLM-MC can be a new gold standard for retinal layer segmentation and for the measurement of retinal layer metrics. Ninety-six high-speed 9 mm 31-line volume scans obtained from 48 patients with RPGR-associated XLRP were selected based on the following criteria: the presence of an EZ band within the scan limit and a detectable EZ in at least three B-scans in a volume scan. All the B-scan images in each volume scan were manually segmented for the EZ and proximal retinal pigment epithelium (pRPE) by two experienced human graders to serve as the ground truth for comparison. The test volume scans were also segmented by a DLM and then manually corrected for EZ and pRPE by the same two graders to obtain DLM-MC segmentation. The EZ area and OS volume were determined by interpolating the discrete two-dimensional B-scan EZ-pRPE layer over the scan area. Dice similarity, Bland-Altman analysis, correlation, and linear regression analyses were conducted to assess the agreement between DLM-MC and MG for the EZ area and OS volume measurements. For the EZ area, the overall mean dice score (SD) between DLM-MC and MG was 0.8524 (0.0821), which was comparable to 0.8417 (0.1111) between two MGs. For the EZ area > 1 mm2, the average dice score increased to 0.8799 (0.0614). When comparing DLM-MC to MG, the Bland-Altman plots revealed a mean difference (SE) of 0.0132 (0.0953) mm2 and a coefficient of repeatability (CoR) of 1.8303 mm2 for the EZ area and a mean difference (SE) of 0.0080 (0.0020) mm3 and a CoR of 0.0381 mm3 for the OS volume. The correlation coefficients (95% CI) were 0.9928 (0.9892-0.9952) and 0.9938 (0.9906-0.9958) for the EZ area and OS volume, respectively. The linear regression slopes (95% CI) were 0.9598 (0.9399-0.9797) and 1.0104 (0.9909-1.0298), respectively. The results from this study suggest that the manual correction of deep learning model segmentation can generate EZ area and OS volume measurements in excellent agreement with those of conventional manual grading in RP. Because DLM-MC is more efficient for retinal layer segmentation from OCT scan images, it has the potential to reduce the burden of human graders in obtaining quantitative measurements of biomarkers for assessing disease progression and treatment outcomes in RP.

Generative Adversarial Network Based Inversion of Cross-hole Radar Data

Cross-hole radar is a popular method to characterize subsurface structures, yet traditional data interpretation methods have limitations: tomography method has poor inversion accuracy, while the full waveform inversion method costs huge computing time. In order to solve the above problems, an automatic inversion algorithm based on generative adversarial networks (GAN) is developed in this paper to interpret the permittivity from cross-hole radar B-scan images. This algorithm uses a low-resolution GAN to extract the global features in the cross-hole radar data to reconstruct the dielectric constant distribution map with low resolution, and then adopts a high-resolution GAN to enhance the resolution of the inversion results. The algorithm is trained on 1000 pairs of cross-hole radar data obtained from the finite-difference time-domain (FDTD) method. Finally, 100 pairs of similar data which has never been shown in the network are used to verify the inversion performance of the algorithm. The results show that the inversion accuracy of is greater than 90%, and the structural similarity index measure (SSIM) of the reconstructed image reaches 0.9. In addition, the proposed method also has rapid computing speed.

Deep Learning Inversion of Multi-processed GPR Data to Characterize Subsurface Defects

Ground penetrating radar (GPR) is highly recommended for investing defects in concrete structures, yet interpreting GPR data is hindered by the fact that the images primarily display reflected electromagnetic waves rather than the actual shapes of the defects. In this paper, we introduce a deep learning network that aims to convert GPR B-scan images into permittivity maps of subsurface structures. This network takes three types of data as inputs and outputs the distribution of permittivity. These three types of data are the original data, the data obtained after applying migration to the original data, and the data obtained after processing the original data using a neural network. Each type of data is processed using different methods and contains distinct information, which is collectively used for inverting data. In this study, we validate the effectiveness of the proposed network by the finite-difference time-domain (FDTD) forward modeling data.

Safety Monitoring of Transportation Infrastructure Foundation: Intelligent Recognition of Subgrade Distresses Based on B-Scan GPR Images

The safety monitoring of transportation infrastructure foundation is crucial for the sustainable service of transportation systems. In recent years, the Ground Penetrating Radar (GPR) has become a powerful tool to identify and locate the subgrade distresses according to the different responses of wave characteristics, preliminarily realizing an intelligent nondestructive detection. To solve the problems like small sample size and unbalanced dataset, this study used a deep data augmentation method, e.g. WGAN-GP network, to augment the original limited B-Scan GPR data of subgrade, and then carried out supervised learning for classification task. The detailed computation steps include the image processing, data augmentation and intelligent analysis. First, the dataset was initially enlarged through the traditional methods after noise filtering, gamma transform and other processing methods. Then, the WGAN-GP network was adopted to generate new high-quality B-Scan images. Finally, the intelligent classification of subgrade distresses was realized by ResNet50 model with a satisfactory accuracy of 90.85%.

Automatic recognition of defects behind railway tunnel linings in GPR images using transfer learning

Ground-penetrating radar (GPR) is widely used for non-destructive detection of hidden defects behind railway tunnel linings, including non-compactness, air- and water-filled voids, fractures, and insufficient lining thickness. However, manual interpreting a substantial volume of GPR B-scan images is time-consuming, and the results heavily depend on the professional knowledge of the practitioner and the available prior information. This paper presents an automatic recognition algorithm for detecting hidden defects, as well as estimating lining thickness in GPR B-scan images of railway tunnels using a You Only Look Once for Panoptic driving perception network (YOLOP) model. The process mainly contains three steps. Firstly, a real GPR dataset of underground objects is employed for transfer learning to enhance the robustness of the prediction model. Secondly, a dataset containing 2,245 real GPR B-scan images of railway tunnel linings is created. Thirdly, the established dataset is used to retrain the YOLOP model and assess the prediction accuracy for identifying hidden defects and lining thickness in GPR images. Field experiment results demonstrate that the trained hidden defects recognition algorithm achieves an average accuracy of 80.6%, with a detection time of only 0.02 s for a GPR image sizing of 1024 × 512. Therefore, it is concluded that the proposed algorithm is efficient and reliable for automatic recognition of hidden defects behind railway tunnel linings.

Artificial intelligence-aided detection of rail defects based on ultrasonic imaging data

Railroads are one of society’s fundamental infrastructures, facilitating the transportation of passengers and goods over vast distances. Rail status data is immensely important for ensuring the safe and efficient operation of railroad networks. However, analyzing ultrasonic inspection data is a labor-intensive process and relies heavily on the expertise of experienced inspectors. To detect internal defects of the rail accurately and automatically, this paper proposes a customized image recognition method based on a convolutional neural network with limited B-scan rail image data collected within the industry. The proposed method uses EfficientNet-b7 as the backbone network to fully extract the B-scan rail image data features. With the help of transfer learning and data augmentation techniques, the backbone network is substantially enhanced so that it can understand high-level features of the object without being trained with large-scale B-scan image data. We establish a real-world internal rail defect dataset with 280 B-scan images and test our proposed method. The results reveal that the highest accuracy of the other mainstream CNN-based methods is 76.25% and the accuracy of the traditional method based on a support vector machine classifier trained with Tamura texture and LBP features is 60.00%. Our proposed EfficientNet-b7 model classifies rail defect B-scan images with an accuracy of 85.00%, precision of 84.71%, and recall of 85.00%. Compared to other rail internal defect detection methods, this method is more accurate. With the help of transfer learning and data augmentation, our proposed method achieves better performance and requires less data.

Cement pavement void detection algorithm based on GPR signal and continuous wavelet transform method

The dimension of the void area in pavement is crucial to its structural safety. However, there is no effective method to measure its geometric parameters. To address this issue, a void size extraction algorithm based on the continuous wavelet transform (CWT) method was proposed using ground-penetrating radar (GPR) signal. Firstly, the finite-difference time-domain (FDTD) method was used to investigate the GPR response of void areas with different shapes, sizes, and depths. Next, the GPR signal was processed using the CWT method, and a 3D image based on the CWT result was used to visualize the void area. Based on the differences between the void and normal pavement in the time and frequency domains, the signal with maximum energy from the CWT time–frequency result was extracted and combined to reconstruct the new B-scan image, where void areas have energy concentration phenomenon. Based on this, width and depth of void detection algorithm was proposed to recognize the void area. Finally, the detection algorithm was verified both in numerical model and physical lab model. The results indicated that the CWT time–frequency energy spectrum can be used to enhance the void feature, and the 3D CWT image can clearly visualize the void area with a highlighted energy area. After fully testing and validating in numerical and lab models, our proposed method achieved high accuracy in void width and depth detection, providing a precise method for estimating void dimension in pavement. This method can guide DOT departments to carry out pre-maintenance, thereby ensuring pavement safety.

Detecting Structural Changes in the Choroidal Layer of the Eye in Neurodegenerative Disease Patients through Optical Coherence Tomography Image Processing.

To evaluate alterations of the choroid in patients with a neurodegenerative disease versus healthy controls, a custom algorithm based on superpixel segmentation was used. A cross-sectional study was conducted on data obtained in a previous cohort study. Swept-source optical coherence tomography (OCT) B-scan images obtained using a Triton (Topcon, Japan) device were compiled according to current OSCAR IB and APOSTEL OCT image quality criteria. Images were included from three cohorts: multiple sclerosis (MS) patients, Parkinson disease (PD) patients, and healthy subjects. Only patients with early-stage MS and PD were included. In total, 104 OCT B-scan images were processed using a custom superpixel segmentation (SpS) algorithm to detect boundary limits in the choroidal layer and the optical properties of the image. The algorithm groups pixels with similar structural properties to generate clusters with similar meaningful properties. SpS selects and groups the superpixels in a segmented choroidal area, computing the choroidal optical image density (COID), measured as the standard mean gray level, and the total choroidal area (CA), measured as px2. The CA and choroidal density (CD) were significantly reduced in the two neurodegenerative disease groups (higher in PD than in MS) versus the healthy subjects (p < 0.001); choroidal area was also significantly reduced in the MS group versus the healthy subjects. The COID increased significantly in the PD patients versus the MS patients and in the MS patients versus the healthy controls (p < 0.001). The SpS algorithm detected choroidal tissue boundary limits and differences optical density in MS and PD patients versus healthy controls. The application of the SpS algorithm to OCT images potentially acts as a non-invasive biomarker for the early diagnosis of MS and PD.

OIMHS: An Optical Coherence Tomography Image Dataset Based on Macular Hole Manual Segmentation

Macular holes, one of the most common macular diseases, require timely treatment. The morphological changes on optical coherence tomography (OCT) images provided an opportunity for direct observation of the disease, and accurate segmentation was needed to identify and quantify the lesions. Developments of such algorithms had been obstructed by a lack of high-quality datasets (the OCT images and the corresponding gold standard macular hole segmentation labels), especially for supervised learning-based segmentation algorithms. In such context, we established a large OCT image macular hole segmentation (OIMHS) dataset with 3859 B-scan images of 119 patients, and each image provided four segmentation labels: retina, macular hole, intraretinal cysts, and choroid. This dataset offered an excellent opportunity for investigating the accuracy and reliability of different segmentation algorithms for macular holes and a new research insight into the further development of clinical research for macular diseases, which included the retina, lesions, and choroid in quantitative analyses.

Simulations and experimentation of ultrasonic wave propagation and flaw characterisation for underwater concrete structures

ABSTRACT Ultrasonic non-destructive evaluation of concrete submerged in water is a complex phenomenon that has important implications for the durability and safety of marine structures such as bridges, piers and offshore platforms. The scattering of ultrasonic waves inside concrete materials with aggregate size comparable to the ultrasound wavelength often limits the interpretation of inspection results. To enhance the understanding of ultrasonic wave propagation inside the underwater concrete, numerical simulations using the finite element method (FEM) with the Transient Acoustics-Piezoelectric Interaction application mode of Multiphysics software package have been developed. This study focuses on developing a 2D numerical model to reproduce and illustrate the propagation of ultrasound in seven types of underwater concrete test blocks with various types of inclusions and voids embedded inside it. The B-Scan 2D images of the concrete test blocks were acquired using the simulated data and compared with experimental results acquired using an in-house developed four-channel ultrasonic imaging system. Both simulation and experimental results showed a good resemblance, indicating that B-Scan imaging can be applied to identify and characterise various types of defects and inclusions in underwater concrete structures, providing better assessment of the material than conventional ultrasonic pulse velocity (UPV) method.

Ultrasound Assessment of Gaze-induced Posterior Eyewall Deformation in Highly Myopic Eyes.

To establish a quantitative metric of posterior eyewall deformability in different directions of gaze in highly myopic eyes with and without posterior staphyloma. A prospective study was performed on 53 highly myopic patients (106 eyes). Ultrasound scans were acquired in primary, up, downward, nasal, and temporal gazes. A validated intensity-based segmentation algorithm was used to quantify the posterior eyewall geometry on digitalized B-scan images. Posterior eyewall local curvature (K) and distance (L) to the transducer were calculated. The associations between directions of gaze, axial length (AL), and presence of staphyloma with the K and L parameters were assessed. A total of 53 participants (106 eyes) were studied. Multivariate regression analysis demonstrated that, after accounting for longer AL, and presence of staphyloma, eccentric gaze was often independently associated with various K and L parameters. Specifically, downward gaze was associated with increased posterior eyewall concavity as reflected in the maximum of K (KMax) (β = 0.050, P < 0.001) and absolute value of KMax (β = 0.041, P = 0.011). Both downward gaze and upgaze were independently associated with increase in the derivative of absolute KMax (which is consistent with more apparent, steeper staphyloma ridges), local KMax (which detects KMax at smaller intervals), and Kstd (which represents likelihood of staphyloma presence) and decrease in maximum of L (which represents movement of the staphyloma apex) with all P < 0.05. The β coefficients for downward gaze were consistently greater in magnitude compared with those in upgaze. After accounting for AL and presence of staphyloma, horizontal gazes were independently associated only with decrease in the standard deviation of L (which also represents likelihood of staphyloma presence) and maximum of L. Downward gaze results in a significant increase in posterior eyewall concavity in highly myopic eyes after accounting for AL and staphyloma presence. In comparison with downward gaze, upgaze resulted in a lower magnitude, but significant changes in staphyloma ridge steepness and the likelihood of staphyloma presence. Horizontal gazes seemed to be associated with less posterior eyewall geometric parameters. Studies are required to further assess the association between downward gaze during near work on posterior eyewall concavity and possible effects on myopia development and progression.

Evaluation of epiretinal membrane formation after scleral buckling for treating rhegmatogenous retinal detachment: En face optical coherence tomography image-based study.

To assess epiretinal membrane (ERM) formation, severity, and the associated risk factors after scleral buckling using en face optical coherence tomography (OCT) images. Medical records of 61 consecutive patients (66 eyes) with rhegmatogenous retinal detachment who underwent scleral buckling were retrospectively reviewed. Posterior vitreous detachment (PVD) was determined based on B-scan OCT images. En face OCT images were used to visualize the ERM and retinal folds. ERM formation was identified by comparing en face images pre- and post-surgery. The maximum depth of the retinal folds (MDRF) was measured using en face imaging to objectively assess traction strength. ERM formation occurred in 15 (22.7%) eyes at the final visit; the foveal pit was preserved in all cases. Parafoveal retinal folds were present in 5 (7.6%) eyes, with a mean MDRF of 21.8 ± 12.6µm. No significant difference was observed in best-corrected visual acuity (logarithm of the minimal angle of resolution) between the ERM formation (-0.019 ± 0.128) and non-ERM formation (-0.001 ± 0.213) groups at the final visit (P = 0.593; Mann-Whitney U test). Multivariate logistic regression analysis revealed that older age and the presence of PVD were significant risk factors for ERM formation (odds ratio 1.07, 95% confidence interval 1.01-1.14, P = 0.032; odds ratio 5.26, 95% confidence interval 1.06-26.10, P = 0.042; respectively). ERM occurred in 22.7% of cases but was mild and did not affect visual acuity. Older age and the presence of PVD are risk factors for ERM formation.

Birefringence-derived artifact in optical coherence tomography imaging of the lamina cribrosa in eyes with glaucoma

We investigated birefringence-derived artifacts that potentially mimic focal defects of the lamina cribrosa (focal LC defects) in optical coherence tomography (OCT) imaging of eyes with glaucoma. This study included 74 eyes of 48 patients with glaucoma. Five horizontal line B-scan images of the optic disc were obtained using commercial swept-source OCT. From a dataset of prototype swept-source polarization-diversity OCT, we calculated the following types of OCT images: polarization-dependent, polarization-dependent attenuation-coefficient, polarization-independent, and polarization-independent attenuation-coefficient. We assessed the commercial OCT images for the presence of birefringence-derived artifacts by comparison with the polarization-diversity OCT images. Commercial OCT showed suggestive findings of focal LC defects in 17 of 74 eyes. Reevaluation using polarization-independent OCT revealed that the focal LC defects in one of 17 eyes (5.9%) were actually birefringence-derived artifacts. This study demonstrated the existence of birefringence-derived artifacts mimicking focal LC defects in commercial OCT imaging and indicated that polarization-diversity OCT is an effective tool to evaluate the presence of these artifacts.