3. Producción
Browse
8 results
Search Results
- Some of the metrics are blocked by yourconsent settings
Item type:Publication, Standardizing Obstetric Ultrasound Segmentation Using Unpaired Domain Translation Techniques(IEEE Computer Society, 2025-01-01)Prenatal ultrasound is essential for fetal monitoring, yet access in low-resource settings remains limited by the shortage of trained personnel and variability in imaging equipment. Volume Sweep Imaging (VSI) enables acquisition by non-experts and provides diagnostically useful cine-loops for physician review. However, when extending VSI interpretation to artificial intelligence (AI) pipelines, cross-scanner variability introduces domain shift - scanner-dependent differences in contrast, speckle, and resolution that degrade model generalizability. In this study, we evaluate unpaired domain translation methods to standardize obstetric VSI across two scanners (Butterfly iQ+ and Mindray DP10). We implemented CycleGAN, denoising diffusion GANs, and a sequential (CG→Diff) approach. Performance was assessed with distribution similarity (PSNR, MI, BC, MAE) and structural similarity (SSIM, LNCC, CSS) metrics, alongside qualitative segmentation analysis. Results show that the hybrid method achieved the most balanced performance, improving PSNR (21.69) and LNCC (0.59) while preserving anatomical structures. These findings highlight the potential of adversarial-diffusion pipelines to mitigate domain shift and enable scalable AI-assisted obstetric ultrasound in low-resource environments.2 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Power Doppler-Based Shear Wave Speed Estimation via Spatial Interference Patterns(IEEE Computer Society, 2025-01-01)This paper presents the experimental implementation of a measurement system for convex optical surfaces based on Ronchi deflectometry. The experimental setup, data processing procedures, system calibration, and surface curvature parameters are described u1 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Fourier Synchrosqueezed Transform for Shear Wave Speed Estimation in Crawling Wave Sonoelastography Approach(Institute of Electrical and Electronics Engineers Inc., 2025-01-01)Crawling Wave Sonoelastography (CWS) is a quantitative elastography technique that employs two mechanical actuators to generate an interference pattern within the tissue. Ultrasound imaging is then used to capture the resulting wave fields, and the shear wave speed (SWS) is computed to produce an elastography image. In previous studies, different time-frequency techniques have been employed to estimate the SWS, but some limitations, such as lateral artifacts and blurred SWS maps, were reported. In this paper, a novel approach based on the Fourier Synchrosqueezed Transform (FSST) is presented. To assert the veracity of the results, previous datasets in homogeneous and heterogeneous phantoms with vibration frequencies between 200 and 360 Hz have been used. The proposed metrics for comparison were SWS mean value and standard variation, coefficient of variation (CV), Bias, R2080, and, contrast-to-noise ratio (CNR). The new estimator demonstrates marginally superior performance in SWS mean value (at 340 Hz, inclusion: 5.13±0.01 m/s, background: 3.42±0.02 m/s) CV (at 320 Hz, inclusion: 0.11%, background: 0%) and CNR (at 320 Hz, 104.7 dB), and better performance in Bias (at 320 Hz, inclusion: 0.6%, background: 0.84%) and R2080 (at 320 Hz, 0.5 mm) in comparison with previous time-frequency approaches.Clinical relevance - This investigation presents a new Shear Wave Speed estimator for Crawling Waves Sonoelastography approach, which is able to quantify stiffness tissue with great accuracy showing the potential of real-time time application to allow the characterization of tissue elasticity.3 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, First Exploration of H-scan Ultrasound Imaging in Diabetic Foot: A Feasibility Study(IEEE Computer Society, 2025-01-01)Diabetic foot complications are a leading cause of morbidity and lower-limb amputation worldwide, largely driven by structural and mechanical alterations of plantar soft tissues. Reverberant shear wave ultrasound elastography has shown potential in detecting increased stiffness in diabetic plantar tissue; however, stiffness alone does not fully capture microstructural remodeling at the scatterer level. H-scan ultrasound imaging is a scatterer-size-sensitive technique that encodes frequency-dependent backscatter information into color maps, providing a novel means of assessing tissue microarchitecture. In this feasibility study, we applied H-scan imaging to the plantar soft tissues of 10 diabetic patients and 3 healthy controls. Radiofrequency ultrasound data were acquired at clinically relevant sites (1st and 3rd metatarsal heads and heel), processed using a 256-filter Gaussian convolution algorithm, and analyzed with an automated region-of-interest detection method. The intensity-weighted percentage of red pixels (IWPred), representing the prevalence of larger scatterers, was extracted as a quantitative biomarker. Results showed significantly higher IWPred values in participants with diabetes at the 3rd metatarsal head for both feet (left: p ≤ 0.002, right: p ≤ 0.001), while no significant differences were observed at the 1st metatarsal head or heel. These findings suggest that H-scan imaging can detect microstructural alterations in diabetic plantar tissues, particularly at high-risk ulceration sites. This study provides the first evidence supporting the feasibility of H-scan ultrasound as a non-invasive, rapid, and clinically deployable tool for diabetic foot risk assessment.1 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Comparison of Reverberant Shear Wave Speed Estimators in Non-Ideal Fields(IEEE Computer Society, 2025-01-01)Reverberant shear wave elastography (RSWE) enables tissue stiffness estimation but implementations can be limited by non-ideal wave fields and directional artifacts. This study compares two robust shear wave speed (SWS) estimators: Angular Integration Autocorrelation (AIA) and Regularized Total Variation Phase Gradient (PG-TV) - across five experimental setups using three heterogeneous custom-made gelatin phantoms, a calibrated breast phantom, and ex vivo liver tissue, with vibrations from 100-900 Hz applied using 1-5 mini shakers. Results show that AIA consistently achieved an average lower error across setups (27.1%) compared to PG-TV (38.2%), providing greater robustness to noise and directional bias, particularly in background regions at higher frequencies. Nevertheless, PG-TV provided greater CNR overall, with an average of 2.78 vs. 2.11 from the AIA estimator, showing sharper boundaries, especially in heterogeneous media. Both estimators showed reduced performance at low frequencies due to standing waves. Notably, increasing the number of actuators did not guarantee improved estimations. Results suggest that AIA and PG-TV offer complementary strengths and highlight the need for adaptive metrics to assess field diffusivity and estimator reliability.3 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Comparison of Training Strategies Using YOLOv8 for Automatic Region of Interest Selection and Landmark Identification in Foot Ultrasound(Springer Science and Business Media Deutschland GmbH, 2025-01-01)This study explores the use of foot ultrasound technique for assessing the recognition of regions of interest (ROI’s) in the foot, so it can help streamline processes for detecting abnormal patterns, specifically in diabetic foot using RSWE elastography. It focuses on the heel’s microchamber and calcaneus zones as key ROI’s. Traditionally, ROI detection relied on conventional algorithms; however, this paper introduces a novel approach using deep learning (DL) for automatic ROI detection in heel segmentation employing YOLOv8 algorithms, refined with data from prior research. A detailed comparative analysis was conducted between single-categorical and multicategorical models to classify the heel’s critical areas. Results showed that single-categorical models achieved an average precision of 98%, recall of 88%, mAP50 of 93%, and mAP50–95 of 65%. In contrast, multicategorical models showed a balanced performance with 90% precision and recall, 92% mAP50, and 58% mAP50–95%. These findings underscore the potential of deploying DL techniques for diabetic foot assessment, marking a significant advancement in automated and precise medical diagnostics for early diabetes detection through foot condition analysis.2 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, Automated Analysis of Fetal Heart Rate from VSI-Based Ultrasound Using Segmentation-Guided Optical Flow(IEEE Computer Society, 2025-01-01)Volume Sweep Imaging (VSI) enables standardized obstetric ultrasound acquisition by non-experts in low-resource settings; helping to the assessment of diverse fetal variables. While Doppler is used to detect fetal heart changes; due to the limited constraints is not possible to apply this method, and VSI has not yet been applied for fetal heart rate (FHR) measurement without Doppler capability. This study proposes a segmentation-guided optical flow method to estimate FHR from VSI-based B-mode cine-loops. Third-trimester VSI sweeps where fetal heart was visible were manually segmented frame-by-frame by a physician. Dense optical flow was computed within the heart mask, and the mean motion angle over time was bandpass-filtered (90-200 BPM) with variable Butterworth order. A confidence score (1-5) was proposed based on the filter order that was correlated to the quantity of frames with visible heart. Out of 114 cine-loops analyzed, all clips with confidence score ≤ 3 (n=49) were non-measurable by both the algorithm and physicians. Among high-confidence clips (≥ 4; n=65), 18 were excluded due to incomplete heart visualization, leaving 47 for Bland-Altman analysis. The mean bias was -8.63 BPM with 95% limits of agreement [-52.26,69.52] BPM and no statistically significant difference (p > 0.05). These findings demonstrate the feasibility of integrating automated FHR estimation into existing VSI obstetric protocols without Doppler, enabling functional cardiac assessment in settings with limited access to specialized sonography.1 - Some of the metrics are blocked by yourconsent settings
Item type:Publication, An AI-enabled comprehensive breast ultrasound diagnostic system for low-resource settings without a sonographer or a radiologist(Nature Portfolio, 2026-12-01)Breast cancer is the most common non-skin related malignancy and the leading cause of cancer death in women. Mammography remains the gold standard for early detection; however, its accessibility is limited in low-resource settings due to cost and technical complexity. Ultrasound (US) is a viable alternative, but its implementation is hindered by the scarcity of trained radiologists and sonographers. Volume sweep imaging (VSI) has addressed the issue of US acquisition by enabling non-specialists to perform standardized scans. However, these still require expert interpretation, limiting their impact. To overcome this barrier, we propose a fully automated Breast VSI (VSI-B) system integrating artificial intelligence (AI) for segmentation and classification of breast lesions, aiming to provide an accessible diagnostic tool for low-resource environments. This study developed an AI-driven diagnostic system for VSI-B, combining a segmentation model (Attention U-Net 3D) with a classification model for lesion detection. A total of 98 patients with palpable breast lumps were included in the study. The dataset consisted of 392 VSI-B US videos and 2,100 classified frames. A new method was implemented to enhance mass identification by selecting key frames for analysis. A majority voting algorithm was used to optimize lesion classification. The system's performance was assessed based on sensitivity, specificity, and accuracy. Following a detection step that achieved 100% sensitivity and 93.6% specificity for cancer and no cancer patients, as well as 95.0% sensitivity and 63.0% specificity for mass and no mass patients, the DenseNet classification model reached 87% accuracy, 100% sensitivity, and 83% specificity. A majority voting algorithm optimized classification, yielding an AUC of 0.91. This study highlights the potential of an AI-enabled VSI-B system as a reliable diagnostic tool in low-resource settings. By integrating multi-modal segmentation and classification, the system automates breast lesion detection and stratification, reducing reliance on radiologists. The results suggest that this approach could enhance early breast cancer diagnosis and guide clinical decision-making, particularly in underserved regions.1