Ultrasound Kidney Images Research Articles

Ultrasound image smoothing based on adaptive and non-adaptive filters.

To model adaptive and non-adaptive filters to ensure smooth ultrasound images. The comparative study was conducted at Al-Yarmouk Teaching Hospital, Al Mustansiriyah University, Baghdad, Iraq, in 2019, and comprised ultrasound images of kidney (303x208 pixel) and foetus (111x109 pixel). These images were smoothed based on 8 filters; 1 non-adaptive (median), and 7 adaptive enhanced filters (Gamma, Wiener, Lee, Frost, Kuan, Adaptive Lee and Adaptive Frost). They were applied to the images by windows measuring 3x3, 5x5, 7x7. The additive noise and the multiplicative noise factor were calculated using histogram to determine the noise type for each image. Statistical criteria included mean square error, normalised absolute error and signalto- noise ratio. The relationship between noise ratio and filter type showed that Wiener was the best filter and the best sliding window was 3x3. The worst filters were Gamma, EFrost and Kuan. The relationship between sliding window size and noise ratio for all the smoothing filters clearly identified the best filter for the type of noise.

Cross-modal Transfer Learning Based on an Improved CycleGAN Model for Accurate Kidney Segmentation in Ultrasound Images

ObjectiveDeep-learning algorithms have been widely applied in the field of automatic kidney ultrasound (US) image segmentation. However, obtaining a large number of accurate kidney labels clinically is very difficult and time-consuming. To solve this problem, we have proposed an efficient cross-modal transfer learning method to improve the performance of the segmentation network on a limited labeled kidney US dataset. MethodsWe aim to implement an improved image-to-image translation network called Seg-CycleGAN to generate accurate annotated kidney US data from labeled abdomen computed tomography images. The Seg-CycleGAN framework primarily consists of two structures: (i) a standard CycleGAN network to visually simulate kidney US from a publicly available labeled abdomen computed tomography dataset; (ii) and a segmentation network to ensure accurate kidney anatomical structures in US images. Based on the large number of simulated kidney US images and small number of real annotated kidney US images, we then aimed to employ a fine-tuning strategy to obtain better segmentation results. ResultsTo validate the effectiveness of the proposed method, we tested this method on both normal and abnormal kidney US images. The experimental results showed that the proposed method achieved a segmentation accuracy of 0.8548 in dice similarity coefficient on all testing datasets and 0.7622 on the abnormal testing dataset. ConclusionsCompared with existing data augmentation and transfer learning methods, the proposed method improved the accuracy and generalization of the kidney US image segmentation network on a limited number of training datasets. It therefore has the potential to significantly reduce annotation costs in clinical settings.

DEEP LEARNING BASED KIDNEY DISEASE CLASSIFICATION SYSTEM

Kidney disease is a significant global health issue that requires timely and accurate diagnosis to improve patient outcomes. This project presents a deep learning-based kidney classification system designed to assist in the early detection and classification of various kidney conditions using medical imaging data. The proposed system leverages advanced convolutional neural networks (CNNs) to analyze and classify kidney ultrasound and MRI images, distinguishing between healthy kidneys and those affected by diseases such as kidney stones, cysts, and tumors. This methodology enhances the model's ability to learn intricate patterns and features associated with different kidney conditions, ensuring high classification accuracy and robustness. Keywords: Deep Learning, Kidney Classification, Medical Imaging, Convolutional Neural Networks (CNNs)

Automated grading of prenatal hydronephrosis severity from segmented kidney ultrasounds using deep learning

Background and motivationsAntenatal or prenatal hydronephrosis (AHN) is a common kidney complication in unborn children. While AHN is generally benign and resolves over time, this condition can inflict serious kidney damage or even organ failure due to excessive waste accumulation in severe cases. Regardless of severity, AHN must be clinically monitored for resolution, with treatment plans and medications being revised according to updated prognoses. Kidney ultrasound images are one of the most common methods of monitoring AHN, but grading of this condition is highly subjective and clinicians may select inappropriate therapies or surgical interventions as a result. New approaches are required to differentiate subjects who can be managed without surgical intervention from those who require life-saving operations. MethodsAn end-to-end deep machine learning framework was developed to sequentially detect ultrasound regions of interest, segment kidneys from US images, and classify AHN severity. We propose the novel Kidney Ultrasound Segmentation Network (KUSNet) for kidney segmentation from ultrasound images, and the Prenatal Hydronephrosis Classification Network (PHCNet) for hydronephrosis severity stratification according to the Society of Fetal Urology (SFU) standards. The ground truth kidney masks were generated by two radiologists with more than five years of working experience while the SFU-based annotations for the AHN severity were done by two senior radiologists and three senior urologists with more than ten years of domain expertise. At each stage, the performance of the proposed models was assessed both quantitatively and qualitatively against state-of-the-art networks in the respective fields. ResultsThe proposed KUSNet for ultrasound kidney segmentation achieved 97.6% accuracy, 97.4% precision, 97.6% recall or sensitivity, 97.5% F1-score, 95.5% IoU or Jaccard score, and 92.1% Dice score, beating several state-of-the-arts. On the other hand, the novel PHCNet reached 93.9% accuracy, 93.7% precision, 93.9% recall, 93.8% specificity, and 89.0% F1-score subject-wise when performing multiclass stratification of AHN severity on segmented kidney regions. ConclusionArtificial intelligence-based tools can reliably classify AHN severity to reduce inter- and intra-observer bias, thereby aiding clinicians in the rapid selection of appropriate treatments and surgeries. Moreover, segmenting kidney regions beforehand significantly boosts AHN severity classification performance.

Transformative Deep Neural Network Approaches in Kidney Ultrasound Segmentation: Empirical Validation with an Annotated Dataset.

Kidney ultrasound (US) images are primarily employed for diagnosing different renal diseases. Among them, one is renal localization and detection, which can be carried out by segmenting the kidney US images. However, kidney segmentation from US images is challenging due to low contrast, speckle noise, fluid, variations in kidney shape, and modality artifacts. Moreover, well-annotated US datasets for renal segmentation and detection are scarce. This study aims to build a novel, well-annotated dataset containing 44,880 US images. In addition, we propose a novel training scheme that utilizes the encoder and decoder parts of a state-of-the-art segmentation algorithm. In the pre-processing step, pixel intensity normalization improves contrast and facilitates model convergence. The modified encoder-decoder architecture improves pyramid-shaped hole pooling, cascaded multiple-hole convolutions, and batch normalization. The pre-processing step gradually reconstructs spatial information, including the capture of complete object boundaries, and the post-processing module with a concave curvature reduces the false positive rate of the results. We present benchmark findings to validate the quality of the proposed training scheme and dataset. We applied six evaluation metrics and several baseline segmentation approaches to our novel kidney US dataset. Among the evaluated models, DeepLabv3+ performed well and achieved the highest dice, Hausdorff distance 95, accuracy, specificity, average symmetric surface distance, and recall scores of 89.76%, 9.91, 98.14%, 98.83%, 3.03, and 90.68%, respectively. The proposed training strategy aids state-of-the-art segmentation models, resulting in better-segmented predictions. Furthermore, the large, well-annotated kidney US public dataset will serve as a valuable baseline source for future medical image analysis research.

Detection of renal cell hydronephrosis in ultrasound kidney images: a study on the efficacy of deep convolutional neural networks

In the realm of medical imaging, the early detection of kidney issues, particularly renal cell hydronephrosis, holds immense importance. Traditionally, the identification of such conditions within ultrasound images has relied on manual analysis, a labor-intensive and error-prone process. However, in recent years, the emergence of deep learning-based algorithms has paved the way for automation in this domain. This study aims to harness the power of deep learning models to autonomously detect renal cell hydronephrosis in ultrasound images taken in close proximity to the kidneys. State-of-the-art architectures, including VGG16, ResNet50, InceptionV3, and the innovative Novel DCNN, were put to the test and subjected to rigorous comparisons. The performance of each model was meticulously evaluated, employing metrics such as F1 score, accuracy, precision, and recall. The results paint a compelling picture. The Novel DCNN model outshines its peers, boasting an impressive accuracy rate of 99.8%. In the same arena, InceptionV3 achieved a notable 90% accuracy, ResNet50 secured 89%, and VGG16 reached 85%. These outcomes underscore the Novel DCNN’s prowess in the realm of renal cell hydronephrosis detection within ultrasound images. Moreover, this study offers a detailed view of each model’s performance through confusion matrices, shedding light on their abilities to categorize true positives, true negatives, false positives, and false negatives. In this regard, the Novel DCNN model exhibits remarkable proficiency, minimizing both false positives and false negatives. In conclusion, this research underscores the Novel DCNN model’s supremacy in automating the detection of renal cell hydronephrosis in ultrasound images. With its exceptional accuracy and minimal error rates, this model stands as a promising tool for healthcare professionals, facilitating early-stage diagnosis and treatment. Furthermore, the model’s convergence rate and accuracy hold potential for enhancement through further exploration, including testing on larger and more diverse datasets and investigating diverse optimization strategies.

Point-of-care-ultrasound for the assessment of post-renal transplant recipients.

Postoperative imaging for deceased donor renal transplants is often delayed, as these surgeries occur after-hours. These delays can be critical in identifying immediate complications. To our knowledge, there are no formal training programs for point-of-care ultrasound (POCUS) in this setting; therefore, we aimed to develop and evaluate a feasible and practical POCUS curriculum for the assessment of a renal transplant graft. Urology and nephrology transplant physicians completed a three-hour online course, followed by a five-hour hands-on seminar for sonographic scanning. Simulated patients with transplanted kidneys were used. Course material was developed with licensed ultrasound technologists based on Sonography Canada national competency profiles. Pre- and post-course surveys focused on user confidence, while pre- and post-course multiple-choice questionnaires assessed theoretical knowledge. Twelve participants were included, six of whom were urologists. Theoretical knowledge in POCUS improved significantly (p<0.001). Confidence in manipulation of ultrasound controls, Doppler imaging, and POCUS of the transplant kidney also improved (all p<0.001, d>2.0). Participants indicated an increased likelihood of POCUS use in clinical practice and that training should be integrated into a transplant fellowship. We introduced a novel and guideline-based POCUS curriculum that leveraged local ultrasound educators and found improved theoretical knowledge and skill confidence in our cohort of transplant physicians. This course will serve as the first step toward a validated competency-based training system for POCUS use in the immediate post-renal transplant setting, and likely will be incorporated into the training of the modern transplant physician.

Ultrasound imaging of renal fibrosis using gold nanoparticles

Abstract Introduction Microbubble ultrasound contrast agents have limitations due to their short circulation half-life and poor acoustic attenuation. In our previous study, pH-responsive gold nanoparticles (2∼3 nm) were optimized for ultrasound fluorescence molecular imaging of mouse kidneys. Renal fibrosis is associated with increased expression of the integrin αvβ3 receptor in neovascular endothelial cells. To address this, gold nanoparticles (cRGD-GSH-AuNPs) were synthesized by introducing RGD peptide, which specifically binds to integrin. Purpose The purpose of this study was to evaluate the potential of cRGD-GSH-AuNPs as a contrast agent for targeted ultrasound imaging of renal fibrosis. Methods Cyclic arginine-glycine-aspartic (cRGD) targeting integrin αvβ3 receptors was linked with glutathione (GSH)-modified gold nanoparticles (AuNPs) to create cRGD-GSH-AuNPs. After characterizing their physical properties and verifying their biocompatibility, GSH-AuNPs and cRGD-GSH-AuNPs were injected into mice with the renal fibrosis model. The renal imaging effect was evaluated by ultrasound and fluorescence imaging, and the targeting of the particles was confirmed by pathological tissue staining. Results The average size of cRGD-GSH-AuNPs determined by electron microscopy was 1.7 ± 0.2 nm, and the particles maintained biological stability after modification. The average ultrasound intensity of cRGD-GSH-AuNPs in the renal fibrosis group (22 ± 2 dB) was significantly higher than that of GSH-AuNPs in the fibrosis group (3.5 ± 3 dB), cRGD-GSH-AuNPs in the control group (1.5 ± 1 dB), and GSH-AuNPs in the control group (2 ± 1.5 dB) (all, P &amp;lt; 0.05). Intravenous injection of cRGD-GSH-AuNPs significantly enhanced the ultrasound signal intensity in fibrotic kidneys (see Figure 1). Conclusions Gold nanoparticles targeting integrin αVβ3 receptors achieve specific non-invasive imaging of renal fibrosis, and are a promising ultrasound contrast agent for the evaluation of renal fibrosis.Figure 1

Image Search Strategy via Visual Servoing for Robotic Kidney Ultrasound Imaging

Ultrasound (US) imaging is beneficial for kidney diagnosis; however, it involves sophisticated tasks that must be performed by physicians to obtain the target image. We propose a target-image search strategy combining visual servoing and deep learning-based image evaluation for robotic kidney US imaging. The search strategy is designed by mimicking physicians’ motion axis of the US probe. By controlling the position of the US probe along each of the motion axes while evaluating the obtained US images based on an anatomical feature extraction method via instance segmentation with YOLACT++, we are able to search for an optimal target image. The proposed approach was validated through phantom studies. The results showed that the proposed approach could find the target kidney images with error rates of 2.88±1.76 mm and 2.75±3.36°. Thus, the proposed method enables the accurate identification of the target image, which highlights its potential for application in autonomous kidney US imaging.

Novel Solution for Using Neural Networks for Kidney Boundary Extraction in 2D Ultrasound Data.

Background and Objective: Kidney ultrasound (US) imaging is a significant imaging modality for evaluating kidney health and is essential for diagnosis, treatment, surgical intervention planning, and follow-up assessments. Kidney US image segmentation consists of extracting useful objects or regions from the total image, which helps determine tissue organization and improve diagnosis. Thus, obtaining accurate kidney segmentation data is an important first step for precisely diagnosing kidney diseases. However, manual delineation of the kidney in US images is complex and tedious in clinical practice. To overcome these challenges, we developed a novel automatic method for US kidney segmentation. Methods: Our method comprises two cascaded steps for US kidney segmentation. The first step utilizes a coarse segmentation procedure based on a deep fusion learning network to roughly segment each input US kidney image. The second step utilizes a refinement procedure to fine-tune the result of the first step by combining an automatic searching polygon tracking method with a machine learning network. In the machine learning network, a suitable and explainable mathematical formula for kidney contours is denoted by basic parameters. Results: Our method is assessed using 1380 trans-abdominal US kidney images obtained from 115 patients. Based on comprehensive comparisons of different noise levels, our method achieves accurate and robust results for kidney segmentation. We use ablation experiments to assess the significance of each component of the method. Compared with state-of-the-art methods, the evaluation metrics of our method are significantly higher. The Dice similarity coefficient (DSC) of our method is 94.6 ± 3.4%, which is higher than those of recent deep learning and hybrid algorithms (89.4 ± 7.1% and 93.7 ± 3.8%, respectively). Conclusions: We develop a coarse-to-refined architecture for the accurate segmentation of US kidney images. It is important to precisely extract kidney contour features because segmentation errors can cause under-dosing of the target or over-dosing of neighboring normal tissues during US-guided brachytherapy. Hence, our method can be used to increase the rigor of kidney US segmentation.

Heterogeneity in Acute Kidney Injury Management in Critically Ill Patients: National Survey

We conducted a descriptive survey study with 365 multidisciplinary critical care providers to examine the heterogeneity in using tools for diagnosing acute kidney injury (AKI) in critical care settings. Results indicated that 52% of respondents used point-of-care ultrasound, followed by 15% using novel biomarkers and 8% using formal kidney ultrasound imaging for diagnostic testing. Cystatin C was the most commonly used novel biomarker (72% of respondents). Only 6% used prediction models and an integrated clinical decision support system to identify patients at high-risk for developing AKI. We confirmed that significant variation exists in the diagnosis of critically ill patients with AKI.

A Comparative Study of Deep Learning Methods for Multi-Class Semantic Segmentation of 2D Kidney Ultrasound Images.

Ultrasound (US) imaging is a widely used medical imaging modality for the diagnosis, monitoring, and surgical planning for kidney conditions. Thus, accurate segmentation of the kidney and internal structures in US images is essential for the assessment of kidney function and the detection of pathological conditions, such as cysts, tumors, and kidney stones. Therefore, there is a need for automated methods that can accurately segment the kidney and internal structures in US images. Over the years, automatic strategies were proposed for such purpose, with deep learning methods achieving the current state-of-the-art results. However, these strategies typically ignore the segmentation of the internal structures of the kidney. Moreover, they were evaluated in different private datasets, hampering the direct comparison of results, and making it difficult to determination the optimal strategy for this task. In this study, we perform a comparative analysis of 7 deep learning networks for the segmentation of the kidney and internal structures (Capsule, Central Echogenic Complex (CEC), Cortex and Medulla) in 2D US images in an open access multi-class kidney US dataset. The dataset includes 514 images, acquired in multiple clinical centers using different US machines and protocols. The dataset contains the annotation of two experts, but 321 images with complete segmentation of all 4 classes were used. Overall, the results demonstrate that the DeepLabV3+ network outperformed the inter-rater variability with a Dice score of 78.0% compared to 75.6% for inter-rater variability. Specifically, DeepLabV3Plus achieved mean Dice scores of 94.2% for the Capsule, 85.8% for the CEC, 62.4% for the Cortex, and 69.6% for the Medulla. These findings suggest the potential of deep learning-based methods in improving the accuracy of kidney segmentation in US images.Clinical Relevance- This study shows the potential of DL for improving accuracy of kidney segmentation in US, leading to increased diagnostic efficiency, and enabling new applications such as computer-aided diagnosis and treatment, ultimately resulting in improved patient outcomes and reduced healthcare costs.1.

Value of ultrasound imaging-omics features in the diagnosis of primary nephrotic syndrome

BackgroundPrimary nephrotic syndrome (PNS) is a common autoimmune systemic suppressive disease that, if left untreated, can develop into end-stage renal disease in severe cases, seriously threatening the health of patients. This disease is gradually attracting attention, and ultrasound imaging histology is being tried for PNS as a noninvasive examination method; however, ultrasound imaging histology has not yet seen clinical application in the diagnosis of PNS; therefore, this study investigates the value of ultrasound imaging histological features in the diagnosis of PNS. MethodsA total of 126 PNS patients hospitalized in the Nephrology Department of Hangzhou First People's Hospital from January 2016 to November 2020 were collected in the PNS group, and 11 7 volunteers from the physical examination center were in the normal control group (NC group). The 7:3 stratified sampling was divided into training set (171 cases) and validation set (72 cases), 396 imaging omics parameters of the right kidney ultrasound image were extracted, statistical different parameters were selected after dimension reduction, the prediction model was constructed by Logistics regression, and the ROC curve was drawn to evaluate the efficacy of PNS. Results 19 imaging omics parameters were obtained in Least Absolute Shrinkage and Selection Operator (Least Absolute Shrinkage and Selection Operator, LASSO) after dimension reduction, and the constructed model had good efficacy for predicting PNS. The area under the curve of training set was 0.82, the accuracy, specificity, sensitivity, positive predictive value and negative predictive value were 0.75,0.77,0.78 and 0.73; the validation set was 0.84, accuracy, specificity, sensitivity, positive, and ne gative predictive values were 0.76,0.80,0.72,0.79 and 0.74, respectively. ConclusionThe imaging omics parameters of the right kidney have the potential as non-invasive biological markers for PNS diagnosis, and can provide more accurate imaging basis for the diagnosis and treatment of PNS patients.PNS has a rapid onset and severe symptoms, especially the prevalence and incidence of PNS in children (Alfakeekh et al., 2019; Nephrology Group Pediatric Branch of Chinese Medical Association, 2017). Because the kidneys of PNS patients have no specific signs during routine imaging examination, puncture examination is mostly used clinically to determine the pathological type, but puncture is a traumatic examination and a secondary injury to the kidney, so seeking the noninvasive biological markers of PNS is the focus of current research. Imaging omics is a rapidly developing discipline in recent years. The imaging omics studies on kidney at home and abroad mainly include benign and malignant classification of kidney space, pathological classification of clear cell carcinoma, kidney stone research and so on (Farber et al., 2015; Homayounieh et al., 2021; Ursprung et al., 2020a, 2020b), However, imaging studies based on the kidney itself of nephrotic syndrome are still blank. In this study, the author conducted a retrospective study on the imaging omics analysis of the right kidney of PNS patients, and constructed a predictive model, aiming to provide a new method for the accurate diagnosis of PNS.

Enhanced Grey Wolf Optimization based Hyper-parameter optimized Convolution Neural Network for Kidney Image Classification

Over the last few years, Convolution Neural Networks (CNN) have shown dominant performance over real world applications due to their ability to find good solutions and deal with image data. However their performance is highly dependent on the network architecture and methods for optimizing their hyper parameters especially number and size of filters. Designing a good CNN architecture requires human expertise and domain knowledge. So, it is difficult in CNN to find sufficient number and size of filters for classification problems. The standard GWO algorithm used for any optimization purpose suffers from some issues such as slow convergence speed, trapping in local minima and unable to maintain balance between exploration and exploitation. In order to have proper balance between these phases, two modifications in GWO are introduced in this paper. A technique for finding optimum CNN architecture using methods based on Enhanced Grey Wolf Optimization (E-GWO) is proposed. The paper presents optimization of hyper parameters (numbers and size of filters in convolution layer) of CNN using E-GWO to improve the performance of the model. Kidney ultrasound images dataset collected from ultrasound centre is used to evaluate the performance of the proposed algorithm. Experimental results showed that optimization of CNN with E-GWO outperformed CNN optimized with traditional GA, PSO and GWO and conventional CNN yielding 97.01% accuracy. At last, the obtained results are statistically validated using t-test.

Modified B-QUIET scoring of kidney images acquired by veterinary students in live canines with versus without sequential practice on ultrasound phantoms.

Ultrasound phantoms are training tools that can help students learn basic ultrasound principles. The purpose of this prospective cohort design study was to determine whether preclinical veterinary students in a curriculum with more phantom training sessions acquire better-quality ultrasound images of kidneys in live canines compared with students in a curriculum without sequential phantom training sessions. In clinical skills labs, 132 second-year (2VM) and 130 third-year (3VM) veterinary students obtained sagittal and transverse images of the left kidney of healthy, student-owned dogs. Images were graded on proper identification/orientation, technique, and image anatomy using a modified Brightness Mode Quality Ultrasound Imaging Examination Technique, a modified standardized ultrasound interpretation scale. A two-sample t-test was used to compare 2VM and 3VM performance. 2VM students were inaugural members of a redesigned curriculum and had previously participated in eight clinical skills labs involving hands-on ultrasound practice using phantoms and live animals prior to this study. The 3VM students were the final members of the prior curriculum and had previously participated in a single ultrasound lab using phantoms and a single ultrasound lab using live animals. For Identification/Orientation categories, 2VM students acquired slightly but statistically significantly better transverse images (P=0.04). There were no significant differences between identification/orientation tasks for sagittal images or for technique and image anatomy categories. The findings indicate that future studies assessing more sensitive evaluation tools and serial evaluation of students may be beneficial in monitoring student competency and assist in evaluating the role of phantoms in ultrasound training in the veterinary curriculum.

Ultrasound Kidney Images with IKNN-Dependent FPGA Abnormality Classification

Ultrasound imaging is commonly used to diagnose internal anomalies. Imaging for abnormality detection is a challenging process in today’s world. Even though there is an advancement in technology, tele-radiographers face difficulty in the accurate diagnosis of abnormalities. In order to resolve this issue, tele-radiology has paved a new way for doctors around the world to access the Internet to share the radiological images from one location to another. But frequent online access is one of the bottleneck issues. In order to overcome this drawback, Computer Assisted Diagnosis (CAD) is preferred in this proposed study and it uses VIRTEX-6 FPGA to clearly identify abnormality in the platform and also manual control is minimized in this condition. The proposed algorithm includes five steps: pre-processing, segmentation, feature extraction, selection and classification. The classification is performed using the Iterative K-Nearest Neighbor (IKNN) classifier based on the selected features. Unlike popular KNN, the proposed IKNN algorithm performs the similarity measurement on selective neighbors for a number of times where the number of neighbors has been dynamically selected at each iteration. Also, at each iteration, the method would select a subset of features in a random way. For the features selected and with the neighbors selected, the method computes the similarity value of Hist-sim which is being measured according to the features selected from the histogram features where the method computes the Haralick similarity with the features selected from the Haralick features. Using the features selected, the method computes the value of cumulative class drive similarity (CCDS). At each iteration the class with maximum similarity is selected and finally, the class being selected for the most number of times is selected as a result of classification. This improves the performance of classification. While comparing with the existing algorithms such as Support Vector Machine (SVM) with the linear, Radial Basis Function (RBF) and polynomial kernels, greater accuracy is achieved via IKNN classification. The specificity is found to be 95, 80 and 75 for normal, cystic and stone kidneys.

Asymmetric U-shaped network with hybrid attention mechanism for kidney ultrasound images segmentation

Kidney ultrasound (KUS) images segmentation is one of the key steps in computer-aided diagnosis. The perturbation of heterogeneous structure, similar intensity distribution, and kidney morphology pose challenges for the segmentation of KUS images. In this paper, we proposed an asymmetric U-shaped network based on the U-net core architecture to segment KUS images accurately and reliably. Specifically, the architecture mainly consists of a dense residual connection encoder, a multi-step up-sampling decoder with the hybrid attention module, and a side-out deep supervision module. The design of the dense residual connection encoder can capture sufficient kidney feature information to improve the representation ability of the network. The development of the hybrid attention module can further guide the network to pay more attention to the representation of the kidney. In addition, the introduction of the side-out deep supervision model can help the network obtain segmentation results that are closer to the ground-truth masks. Moreover, to reduce network parameters, we proposed a multi-step up-sampling optimization strategy to simplify the design of the network. We compare with several state-of-the-art medical image segmentation methods on the same KUS dataset using seven quantitative metrics. The results of our method on Jaccard, Dice, Accuracy, Recall, Precision, ASSD and AUC are 89.95%, 94.59%, 98.65%, 94.47%, 95.07%, 0.3006 and 0.9703, respectively. Experimental results demonstrate that the proposed method achieves the most competitive segmentation performance on KUS images.

MBDSNet: Automatic segmentation of kidney ultrasound images using a multi-branch and deep supervision network

Ultrasound imaging has been widely used in the screening of kidney diseases. The segmentation of the kidney in ultrasound images can aid in the diagnosis of clinical diseases. However, segmenting the kidney from ultrasound images is challenging, due to the effects of heterostructure, image quality, and other factors. In this paper, a multi-branch and deep supervision network (MBDSNet) is proposed to segment the kidney from ultrasound images. Specifically, the architecture consists of two parts: multi-branch encoding network and refined decoding network. The multi-branch encoding network consists of a multi-scale feature input pyramid and three branch encoders. The multi-branch encoding network can reduce the sensitivity of the network to the input data through the multi-scale feature inputs pyramid, and utilize the information exchange between the three branch encoders can improve the generalization ability of the network. In the refined decoding network, the introduction of the side-out deep supervision module and the kidney boundary detection module can further guide the network to effectively segment the kidney with complete contour. Objective comparisons with several state-of-the-art segmentation methods on six commonly used metrics are performed. The method proposed in this paper has achieved the best results in terms of evaluation indicators and segmentation results.

Deep learning imaging features derived from kidney ultrasounds predict chronic kidney disease progression in children with posterior urethral valves.

We sought to use deep learning to extract anatomic features from postnatal kidney ultrasounds and evaluate their performance in predicting the risk and timing of chronic kidney disease (CKD) progression for boys with posterior urethral valves (PUV). We hypothesized that these features would predict CKD progression better than clinical characteristics such as nadir creatinine alone. We performed a retrospective cohort study of boys with PUV treated at two pediatric health systems from 1990 to 2021. Features of kidneys were extracted from initial postnatal kidney ultrasound images using a deep learning model. Three time-to-event prediction models were built using random survival forests. The Imaging Model included deep learning imaging features, the Clinical Model included clinical data, and the Ensemble Model combined imaging features and clinical data. Separate models were built to include time-dependent clinical data that were available at 6months, 1year, 3years, and 5years. Two-hundred and twenty-five patients were included in the analysis. All models performed well with C-indices of 0.7 or greater. The Clinical Model outperformed the Imaging Model at all time points with nadir creatinine driving the performance of the Clinical Model. Combining the 6-month Imaging Model (C-index 0.7; 95% confidence interval [CI] 0.6, 0.79) with the 6-month Clinical Model (C-index 0.79; 95% CI 0.71, 0.86) resulted in a 6-month Ensemble Model that performed better (C-index 0.82; 95% CI 0.77, 0.88) than either model alone. Deep learning imaging features extracted from initial postnatal kidney ultrasounds may improve early prediction of CKD progression among children with PUV. A higher resolution version of the Graphical abstract is available as Supplementary information.

Hydronephrotic Kidney Assessment via Ultrasonography

Objective: To assess hydronephrotic kidney assessment via ultrasonography. Study Design: Prospective study. Place and Duration of Study: Department of Radiology, Chandka Medical College Hospital, Larkana from 1st August 2020 to 31st September/2021. Methodology: Two hundred patients visited for the complaint and suspicion of hydronephrosis and within them 40 patients were diagnosed with hydronephrosis. The both kidneys were examined in the longitudinal as well as transvers planes of scanning using colored Doppler technique in ultrasonography. In older age patients a curved-array transducer having centered frequencies (3-6 MHz) was applied while in case of younger patients’ linear array-transducer was used with high entre frequencies. Results: The age of the patients was between 12-72 years with more elderly patients than youngsters. Most of the patients were males rather than females. Ureteropelvic obstruction in junction, primary megaureter as well as urethral valve involvement was observed in ultrasound imaging of hydronephrotic pediatric kidney. End stage hydronephrosis was presented in adult patients with cortical thining. Conclusion: Ureter and kidney stone appeared to be a common reason of hydronephrosis. Keywords: Hydronephrosis, Ultrasonography, Blockage, Calculi