Kidney Segmentation Research Articles

EfficientNet family U-Net models for deep learning semantic segmentation of kidney tumors on CT images

IntroductionKidney tumors are common cancer in advanced age, and providing early detection is crucial. Medical imaging and deep learning methods are increasingly attractive for identifying and segmenting kidney tumors. Convolutional neural networks have successfully classified and segmented images, enabling clinicians to recognize and segment tumors effectively. CT scans of kidneys aid in tumor assessment and morphology study, using semantic segmentation techniques for pixel-level identification of kidney and surrounding anatomy. Accurate diagnostic procedures are crucial for early detection of kidney cancer.MethodsThis paper proposes an EfficientNet model for complex segmentation by linking the encoder stage EfficientNet with U-Net. This model represents a more successful system with improved encoder and decoder features. The Intersection over Union (IoU) metric quantifies model performance.Results and DiscussionThe EfficientNet models showed high IoU_Scores for background, kidney, and tumor segmentation, with mean IoU_Scores ranging from 0.976 for B0 to 0.980 for B4. B7 received the highest IoU_Score for segmenting kidneys, while B4 received the highest for segmenting tumors. The study utilizes the KiTS19 dataset for contrast-enhanced CT images. Using Semantic segmentation for EfficientNet Family U-Net Models, our method proved even more reliable and will aid doctors in accurate tumor detection and image classification for early diagnosis.

PSMA-PET improves deep learning-based automated CT kidney segmentation

For dosimetry of radiopharmaceutical therapies, it is essential to determine the volume of relevant structures exposed to therapeutic radiation. For many radiopharmaceuticals, the kidneys represent an important organ-at-risk. To reduce the time required for kidney segmentation, which is often still performed manually, numerous approaches have been presented in recent years to apply deep learning-based methods for CT-based automated segmentation. While the automatic segmentation methods presented so far have been based solely on CT information, the aim of this work is to examine the added value of incorporating PSMA-PET data in the automatic kidney segmentation. MethodsA total of 108 PET/CT examinations (53 [68Ga]Ga-PSMA-I&T and 55 [18F]F-PSMA-1007 examinations) were grouped to create a reference data set of manual segmentations of the kidney. These segmentations were performed by a human examiner. For each subject, two segmentations were carried out: one CT-based (detailed) segmentation and one PET-based (coarser) segmentation. Five different u-net based approaches were applied to the data set to perform an automated segmentation of the kidney: CT images only, PET images only (coarse segmentation), a combination of CT and PET images, a combination of CT images and a PET-based coarse mask, and a CT image, which had been pre-segmented using a PET-based coarse mask. A quantitative assessment of these approaches was performed based on a test data set of 20 patients, including Dice score, volume deviation and average Hausdorff distance between automated and manual segmentations. Additionally, a visual evaluation of automated segmentations for 100 additional (i.e., exclusively automatically segmented) patients was performed by a nuclear physician. ResultsOut of all approaches, the best results were achieved by using CT images which had been pre-segmented using a PET-based coarse mask as input. In addition, this method performed significantly better than the segmentation based solely on CT, which was supported by the visual examination of the additional segmentations. In 80% of the cases, the segmentations created by exploiting the PET-based pre-segmentation were preferred by the nuclear physician. ConclusionThis study shows that deep-learning based kidney segmentation can be significantly improved through the addition of a PET-based pre-segmentation. The presented method was shown to be especially beneficial for kidneys with cysts or kidneys that are closely adjacent to other organs such as the spleen, liver or pancreas. In the future, this could lead to a considerable reduction in the time required for dosimetry calculations as well as an improvement in the results.

Abstract P152: Chromatin Accessibility Landscape Of The Proximal Tubule And Thick Ascending Limb In Humans And Rats

Hypertension is a major risk factor for disease burden and mortality. A decrease in the kidney’sability to excrete sodium can lead to hypertension. Specifically, the proximal tubule (PT) and thickascending limb (TAL) of the nephron play crucial roles in the reabsorption of filtered sodium toregulate blood pressure (BP). Understanding the genetic and epigenetic factors that influence PTand TAL functions is essential for unraveling the molecular mechanisms underlying hypertension.GWAS identified over 1,000 BP-associated genomic loci comprising > 26,000 SNPs. Most (~90%)are in intronic and intergenic regions and may influence BP-relevant gene expression by yetunexplored epigenetic mechanisms. Better understanding of how non-coding genomic regionsregulate BP is critical to defining the genetic underpinnings of hypertension. To address this, weestablished chromatin state (open and closed chromatin) maps for PT and TAL from humans andinbred salt-sensitive rats (SS/JrHsdMcwi) using the Assay for Transposase-Accessible Chromatinwith Sequencing (ATAC-Seq). Linear dimensional reduction revealed that in both species, PT andTAL segments clearly separated, indicating their chromatin states were distinct. ATAC-seq “peaks”represent regions of open chromatin, and therefore likely to be within regulatory elements. Ahigher number of differential peaks was identified between human PT and TAL (65,823) comparedto rat (1,063). The number of PT and TAL enriched peaks also differs between species. Genomicdistribution of these differential peaks indicates the bulk fall within distal intergenic and intronicregions (human: 71.64%, rat: 88.81%) rather than the gene promoters or exonic regions for bothspecies. This indicates that the chromatin accessibility of regulatory regions is significantlydifferent between these two segments for both species. Our next steps are to identify upstreamregulators that differ between the kidney segments and species and integrate ATAC-Seq data withexpression data to define the association of non-coding regions with transcript levels. Our workhas generated chromatin state maps that may improve the understanding of the role of noncodinggenomic regions in the transcriptional regulation in PT and TAL in human and rat.

Shear-Wave Elastography-Based Radiomics Nomogram for the Prediction of Cardiovascular Disease in Patients with Diabetic Kidney Disease.

Diabetic kidney disease (DKD) patients have a high risk of suffering from cardiovascular disease (CVD), placing a heavy cost on the public health system. In this study, we intended to develop and validate a shear-wave elastography (SWE)-based radiomics nomogram for predicting the development of CVD in DKD patients. This approach allows extensive use of the valuable information contained in ultrasound images, thus helping clinicians to identify CVD in DKD patients. Totally 337 and 145 patients constituted the training and validation cohorts, respectively. The radiomics features of the segmented kidney in ultrasound images were extracted and selected to generate the rad-score of each patient. These rad-score, as well as the predictors of risk of CVD occurrence from the clinical characteristics, were included in the multivariate analysis to develop a nomogram. It was further assessed in the training and validation cohorts. Patients with CVD accounted for 30.9% (104/337) in the training cohort and 31.0% (45/145) in the validation cohort. The rad-score was calculated for each patient using 6 features extracted from the ultrasound images. The radiomics nomogram was built with the rad-score, age, systolic blood pressure (SBP), low-density lipoprotein cholesterol (LDL-C). It was superior to the clinical nomogram developed without the rad-score and demonstrated promising discrimination, calibration, and clinical utility in both training and validation cohorts. We developed and validated an SWE-based radiomics nomogram to predict CVD risk in patients with DKD. The model was demonstrated to have a promising prediction performance, showing its potential to identify CVD in DKD patients and assist decision-making for appropriate early intervention.

FPN-SE-ResNet Model for Accurate Diagnosis of Kidney Tumors Using CT Images

Kidney tumors are a significant health concern. Early detection and accurate segmentation of kidney tumors are crucial for timely and effective treatment, which can improve patient outcomes. Deep learning techniques, particularly Convolutional Neural Networks (CNNs), have shown great promise in medical image analysis, including identifying and segmenting kidney tumors. Computed tomography (CT) scans of kidneys aid in tumor assessment and morphology studies, employing semantic segmentation techniques for precise pixel-level identification of kidneys and surrounding anatomical structures. This paper proposes a Squeeze-and-Excitation-ResNet (SE-ResNet) model for segmentation by combining the encoder stage of SE-ResNet with the Feature Pyramid Network (FPN). The performance of the proposed SE-ResNet model is evaluated using the Intersection over Union (IoU) and F1-score metrics. Experimental results demonstrate that the SE-ResNet models achieve impressive IoU scores for background, kidney, and tumor segmentation, with mean IoU scores ranging from 0.988 to 0.981 for Seresnet50 and Seresnet18, respectively. Notably, Seresnet50 exhibits the highest IoU score for kidney segmentation. These findings suggest that SE-ResNet models accurately identify and segment regions of interest in CT images of renal carcinoma, with higher model versions generally exhibiting superior performance. The proposed Seresnet50 model is a good tool for accurate tumor detection and image classification, aiding medical professionals in early diagnosis and timely intervention.

Study on the therapeutic effect of glucocorticoids on acute kidney injury in rats exposed to diquat

AimsTo preliminarily explore, whether glucocorticoids have a therapeutic effect on diquat-induced acute kidney injury in rats. Method150 Wistar rats were randomly divided into six groups: exposure model group (DQ group), dexamethasone control group (GC group), blank control group (Ctrl group), dexamethasone 2.1 mg/kg dose group (DQ+L-GC group), dexamethasone 4.2 mg/kg dose group (DQ+M-GC group), and dexamethasone 8.4 mg/kg dose group (DQ+H-GC group), with 25 rats in each group. Each group was further divided into five subgroups, 24 h, 3 d, 7 d, 14 d, and 21 d after exposure, according to the feeding time and the course of treatment, with five animals in each subgroup. The rats in DQ, DQ+L-GC, DQ+M-GC, and DQ+H-GC groups were administered 115.5 mg/kg diquat by gavage, respectively. Moreover, 30 min after gavage, rats in DQ+L-GC group, DQ+M-GC group, DQ+H-GC group and GC group were intragastric administered dexamethasone 2.1 mg/kg, 4.2 mg/kg, 8.4 mg/kg and 8.4 mg/kg, respectively. After 7 days, the intraperitoneal injection of dexamethasone was changed to 6.3 mg/kg prednisone by intragastric administration. Subsequently, 7 days later, it was changed to 3.15 mg/kg prednisone by intragastric administration until the end of the experiment on 21 days. After the start of the experiment, changes in the conditions of the rats in each group were observed at a fixed time every day, changes in the body weight of the rats were monitored at the same time, and the death of the rats was recorded at 24 h, 3 d, 7 d, 14 d, and 21 d after exposure. The rats were sacrificed by an intraperitoneal injection of 100 mg/kg sodium pentobarbital overdose. Blood was collected by puncture of the inferior vena cava, used to determine Cr and BUN. The upper segment of the left kidney was collected for histopathological examination. Elisa was used to detect neutrophil gelatinase-associated lipocalin (NGAL) and kidney injury molecule-1 (KIM-1) in the lower segment of left kidney. TLR4, Myd88, and NF-κB were detected in the right kidney. Results(1) After exposure, most rats in DQ group, DQ+L-GC group, DQ+M-GC group and DQ+H-GC group showed shortness of breath, oliguria, diarrhea, yellow hair and other symptoms. No symptoms and related signs were observed in Ctrl group and GC group.(2) The weight of rats in the Ctrl group and the GC group increased slowly during the test. the body weight of the rats in the DQ, DQ+L-GC, DQ+M-GC, and DQ+H-GC groups continued to decrease after self-infection. Body weight dropped to the lowest point at approximately 7 d, and gradually increased from 7 d to 21 d.(3) A small amount of capillary congestion in the medulla was observed after 7 days in the GC group. The DQ group showed tubular atrophy, edema of the epithelial cells, and over time, the tubules were seen dilated and became irregular in shape; large amount of capillary congestion was also observed in the renal cortex and medulla. The renal injury in the DQ+L-GC group was less than that in the DQ group. DQ+H-GC group had no obvious injury before 7 d, but more renal tubules were seen in the DQ+H-GC group from 7 d to 14 d.(4) Compared with the DQ group, there was no difference before 14 d, and at 14 d-21 d, DQ+L-GC group, DQ+M-GC group, DQ+H-GC group all had different degrees of decline. NGAL content: Compared with the DQ group, the content of NGAL and KIM-1 in kidney tissue of the DQ+L-GC, DQ+M-GC, and DQ+H-GC groups decreased compared with the DQ group at each time node.(5) Compared with the Ctrl group, the expression of TNF-α, TLR4, MyD88, NF-κB in the DQ, DQ+L-GC, DQ+M-GC, and DQ+H-GC groups at each time node increased in the renal tissue. The content of TNF-α, TLR4, MyD88, NF-κB in kidney tissue of the DQ+L-GC, DQ+M-GC, and DQ+H-GC groups at each time node was lower than that in the DQ group. Conclusion(1) Diquat can cause kidney damage in rats, mainly manifested as renal tubular atrophy, epithelial cell edema, capillary congestion and dilation, and the renal function damage indicators have been improved to varying degrees.(2) Glucocorticoids have therapeutic effects on acute kidney injury in rats exposed to diquat. During the treatment, the efficacy of glucocorticoids did not increase with increasing doses after reaching a dose of 4.2 mg/kg.(3) TLR4 receptor-mediated TLR4/Myd88/NF-κB signaling pathway is involved in the inflammatory response of acute kidney injury in diquat poisoning rats. Glucocorticoids can inhibit the inflammatory response, thereby affecting the expression of TLR4/Myd88/NF-κB signaling pathway-related proteins.

AI-generated R.E.N.A.L.+ Score Surpasses Human-generated Score in Predicting Renal Oncologic Outcomes

To determine whether we can surpass the traditional R.E.N.A.L. nephrometry score (H-score) prediction ability of pathologic outcomes by creating artificial intelligence (AI)-generated R.E.N.A.L.+score (AI+ score) with continuous rather than ordinal components. We also assessed the AI+ score components' relative importance with respect to outcome odds. This is a retrospective study of 300 consecutive patients with preoperative computed tomography scans showing suspected renal cancer at a single institution from 2010 to 2018. H-score was tabulated by three trained medical personnel. Deep neural network approach automatically generated kidney segmentation masks of parenchyma and tumor. Geometric algorithms were used to automatically estimate score components as ordinal and continuous variables. Multivariate logistic regression of continuous R.E.N.A.L. components was used to generate AI+ score. Predictive utility was compared between AI+, AI, and H-scores for variables of interest, and AI+ score components' relative importance was assessed. Median age was 60years (interquartile range 51-68), and 40% were female. Median tumor size was 4.2cm (2.6-6.12), and 92% were malignant, including 27%, 37%, and 23% with high-stage, high-grade, and necrosis, respectively. AI+ score demonstrated superior predictive ability over AI and H-scores for predicting malignancy (area under the curve [AUC] 0.69 vs 0.67 vs 0.64, respectively), highstage (AUC 0.82 vs 0.65 vs 0.71, respectively), highgrade (AUC 0.78 vs 0.65 vs 0.65, respectively), pathologic tumor necrosis (AUC 0.81 vs 0.72 vs 0.74, respectively), and partial nephrectomy approach (AUC 0.88 vs 0.74 vs 0.79, respectively). Of AI+ score components, the maximal tumor diameter ("R") was the most important outcomes predictor. AI+ score was superior to AI-score and H-score in predicting oncologic outcomes. Time-efficient AI+ score can be used at the point of care, surpassing validated clinical scoring systems.

Renal ultrasound image segmentation method based on channel attention and GL-UNet11

Background and objectiveKidney tumor is one of the common malignant tumors of urinary system. However, rapid and accurate segmentation of kidney tissue is still a challenging problem in the medical image processing field on account of its obscure onset and uncertain and discontinuous boundary of its anatomical structure. We proposed a deep learning method globa-local UNet11 (GL-UNet11) for ultrasonic image segmentation of kidney, renal parenchyma and renal sinus, aiming at the inaccuracy of internal and marginal segmentation in ultrasonic image segmentation. MethodologyWe deepen the UNet network to make it more expressive, and propose a new channel attention network, Global Local network (GL-UNet), which takes into account the influence of global channel and local channel on predicting the importance of each channel, so that the network can better focus on important information. At the same time, the proposed global-local network is added to the convolutional block of the deepened UNet subsampling part, which effectively enhances the features of the important channels. ResultsThe Dice coefficient and the intersection over union (IOU) of our method reached 96.25% and 92.78% in the task of kidney segmentation. In the task of segmentating renal parenchyma, Dice coefficient reached 92.90% and IOU reached 86.75%. In the task of segmentating renal sinus, Dice coefficient reached 90.18% and IOU reached 82.12%, both of which were superior to other deep learning methods. ConclusionThe GL-UNet11 network proposed in this study compared with the previously proposed UNet and Segnet methods and SE-UNet, ECA-UNet and GL-UNet methods proposed in this paper, Great improvement has been made in both quantitative and qualitative parts. It has achieved good segmentation effect on the kidney segmentation data set, which is of great significance for the early detection and prognosis of kidney patients in the future.

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.

Automatic analysis framework based on 3D-CT multi-scale features for accurate prediction of Ki67 expression levels in substantial renal cell carcinoma

PurposeTo investigate the effectiveness of an automatic analysis framework based on 3D-CT multi-scale features in predicting Ki67 expression levels in substantial renal cell carcinoma (RCC).MethodsThis retrospective study was conducted using multi-center cohorts consisting of 588 participants with pathologically confirmed RCC. The participants were divided into an internal training set (n = 485) and an external testing set (n = 103) from four and one local hospitals, respectively. The proposed automatic analytic framework comprised a 3D kidney and tumor segmentation model constructed by 3D UNet, a 3D-CT multi-scale features extractor based on the renal–tumor feature, and a low or high Ki67 prediction classifier using XGBoost. The framework was validated using a fivefold cross-validation strategy. The Shapley additive explanation (SHAP) method was used to determine the contribution of each feature.ResultsIn the prediction of low or high Ki67, the combination of renal and tumor features achieved better performance than any single features. Internal validation using a fivefold cross-validation strategy yielded AUROC values of 0.75 ± 0.1, 0.75 ± 0.1, 0.83 ± 0.1, 0.77 ± 0.1, and 0.87 ± 0.1, respectively. The optimal model achieved an AUROC of 0.87 ± 0.1 and 0.82 ± 0.1 for low vs. high Ki67 prediction in the internal validation and external testing sets, respectively. Notably, the tumor first-order-10P was identified as the most influential feature in the model decision.ConclusionsOur study suggests that the proposed automatic analysis framework based on 3D-CT multi-scale features has great potential for accurately predicting Ki67 expression levels in substantial RCC.Critical relevance statementAutomatic analysis framework based on 3D-CT multi-scale features provides reliable predictions for Ki67 expression levels in substantial RCC, indicating the potential usage of clinical applications.Graphical abstract

Kidney Segmentations Using CNN models

For medical diagnostic tests, kidney segmentation from high-volume imagery is an important major. Since 3D medical images need a lot of GPU memory, slices and patches are used for training and inference in traditional neural network variant architectures, which necessarily slows down contextual learning. In this research, Mobile Net and Efficient Net CNN models were trained for segmenting human kidney images generated from The Human Biomolecular Atlas Program (HuBMAP). The purpose of this work is to evaluate the effectiveness of different strategies for Glomeruli identification in order to solve the issue. The high size images were decoded to be fitted and trained in the models first, then the CNN models were trained. The CNN models result show that the Efficient Net has the highest accuracy rate with 99.49 %, and Mobile Net with 99.33 %.

Kidney Segmentation from Dynamic Contrast-Enhanced Magnetic Resonance Imaging Integrating Deep Convolutional Neural Networks and Level Set Methods.

The dynamic contrast-enhanced magnetic resonance imaging (DCE-MRI) technique has taken on a significant and increasing role in diagnostic procedures and treatments for patients who suffer from chronic kidney disease. Careful segmentation of kidneys from DCE-MRI scans is an essential early step towards the evaluation of kidney function. Recently, deep convolutional neural networks have increased in popularity in medical image segmentation. To this end, in this paper, we propose a new and fully automated two-phase approach that integrates convolutional neural networks and level set methods to delimit kidneys in DCE-MRI scans. We first develop two convolutional neural networks that rely on the U-Net structure (UNT) to predict a kidney probability map for DCE-MRI scans. Then, to leverage the segmentation performance, the pixel-wise kidney probability map predicted from the deep model is exploited with the shape prior information in a level set method to guide the contour evolution towards the target kidney. Real DCE-MRI datasets of 45 subjects are used for training, validating, and testing the proposed approach. The valuation results demonstrate the high performance of the two-phase approach, achieving a Dice similarity coefficient of 0.95 ± 0.02 and intersection over union of 0.91 ± 0.03, and 1.54 ± 1.6 considering a 95% Hausdorff distance. Our intensive experiments confirm the potential and effectiveness of that approach over both UNT models and numerous recent level set-based methods.

#3390 ADPKD PREDICTOR: A CLOUD-BASED PROGNOSTIC TOOL FOR AUTOSOMAL DOMINANT POLYCYSTIC KIDNEY DISEASE

Abstract Background and Aims ADPKD is a progressively debilitating genetic disease characterized by the growth of numerous cysts in the kidneys, leading eventually to end-stage renal disease. Total Kidney Volume (TKV) is accepted by FDA and EMA as a prognostic biomarker and quantitative predictor of kidney function decline, and is currently used to select patients to receive Tolvaptan treatment. However, TKV calculation from manual tracing of medical images is labor-intensive; for better accuracy, current available methods involve the injection of an iodinated contrast medium for CT scan, with important limitations in patients with impaired renal function. We developed ADPKD Predictor, a user-friendly cloud-based tool for fast and accurate estimation of disease classification and progression, based on automated kidney and cysts segmentation from MRI data. Method An online tool was designed on Microsoft Azure Cloud to automatize the set-up and running of a previously developed algorithm implemented in MATLAB to automatically detect kidneys and cysts contours from MRI data (T2-weighted turbo-spin-echo SPIR), based on advanced image processing techniques [1]. Through the web interface, the user is requested to upload the MRI data and select one point inside kidney's parenchyma in the central slice (Figure 1). Then, TKV is automatically calculated and eGFR based on CKD-EPI equation, ADPKD Imaging Classification and future eGFR [2], estimated Tolvaptan treatment effect [3], and GFR Category based on KDIGO CKD staging system are obtained (Figure 2). The MRI dataset is anonymized before upload to the cloud; data and results are stored in a secure and reliable environment controlled by the user. Results The proposed solution is very fast and precise compared to manual segmentation of medical images (absolute mean error 2.4% ± 2.7%) [1]. Moreover, it is faster and more accurate than the commonly used ellipsoid-based method, resulting in a manifold reduction of misclassification error (2.5%) [1] and therefore potential therapeutic consequences. Another advantage is its usability, with no specific computational expertise, numerical software or dedicated hardware required, since all computations are run remotely in the cloud. Conclusion ADPKD Predictor provides a fast and reproducible assessment of risk classification and disease progression, based on precise morphologic classification of the renal and cysts volume of patient. The proposed solution represents an extremely useful tool for researchers and clinicians to easily obtain an accurate estimation of risk classification potentially helping in a correct and effective stratification of patients, and monitor patient's disease progression, hence supporting a correct and effective therapy administration. Also, it would represent a great benefit for the patient, since the tool analyzes medical images obtained without the use of contrast medium.

#5276 DEVELOPMENT OF A MULTIMODAL “KIDNEY AGE” PREDICTION BASED ON AUTOMATIC SEGMENTATION CT IMAGE IN PATIENTS WITH NORMAL RENAL FUNCTION

Abstract Background and Aims Renal function, for decades, has been of interest to clinicians and researchers to describe. For example, serum creatinine (Scr) and estimated glomerulus filtration rate (eGFR) is familiar but also limited in many circumstances. Meanwhile, the physiological volumes of the kidney cortex and medulla are presumed to change with age and have been proven to change with decreasing kidney function. Method We recruited 182 patients with normal Scr levels and contrasted CT images between Oct. 2021 and Feb 2022 in Peking Union Medical College Hospital (PUMCH) with their demographic and clinical data. The automatic segmentation method was modified from U-NET and used for both cortex and medullary separation and their volume calculation, respectively. We combined the kidney volume and clinical data as multimodal features of the machine learning model. All the data were separated into a training dataset (PUMCH training set, 80%) and a test dataset (PUMCH test set, 20%). Besides, we included patients with the same inclusion criteria but with diabetes (PUMCH-DM test set) and diabetic nephropathy (PUMCH-DN test set) for internal comparison to verify the possible clinical value of “KIDNEY AGE” (K-AGE). Data from DongZhiMen Hospital (DZMH test set) was used as separate external validation sets to evaluate model generalizability. Results The PUMCH training set included 146 participants with a mean age of 47.5±7.4 years. 58.9% were female, and the mean Scr is 63.5±12.3 μmol/L. The PUMCH test set included 36 participants with a mean age of 47.1±7.9 years, 52.7% were female, and the mean Scr was 66.9±13.0 μmol/L. For segmented kidneys, volume differences between non-contrast-enhanced CT and non-contrast-enhanced CT in the left and right kidneys were not evident. The multimodal neural network predicted: K-AGE approximately close to the patient's actual physiological age, with 92% prediction within the 95% confidential interval (Figure 1). We also compared the K-AGE prediction in PUMCH, PUMCH-DM, and PUMCH-DN test sets by applying U-NET-nonCon segmentation algorithm in non-contrast CT images. The mean absolute error increases along with the disease process (control group 5.00, diabetes group 6.99, diabetic nephropathy group 9.32) (Figure 2). Conclusion We established a machine learning model for predicting the K-AGE of normal Scr patients.

An automated surgical decision-making framework for partial or radical nephrectomy based on 3D-CT multi-level anatomical features in renal cell carcinoma

ObjectivesTo determine whether 3D-CT multi-level anatomical features can provide a more accurate prediction of surgical decision-making for partial or radical nephrectomy in renal cell carcinoma.MethodsThis is a retrospective study based on multi-center cohorts. A total of 473 participants with pathologically proved renal cell carcinoma were split into the internal training and the external testing set. The training set contains 412 cases from five open-source cohorts and two local hospitals. The external testing set includes 61 participants from another local hospital. The proposed automatic analytic framework contains the following modules: a 3D kidney and tumor segmentation model constructed by 3D-UNet, a multi-level feature extractor based on the region of interest, and a partial or radical nephrectomy prediction classifier by XGBoost. The fivefold cross-validation strategy was used to get a robust model. A quantitative model interpretation method called the Shapley Additive Explanations was conducted to explore the contribution of each feature.ResultsIn the prediction of partial versus radical nephrectomy, the combination of multi-level features achieved better performance than any single-level feature. For the internal validation, the AUROC was 0.93 ± 0.1, 0.94 ± 0.1, 0.93 ± 0.1, 0.93 ± 0.1, and 0.93 ± 0.1, respectively, as determined by the fivefold cross-validation. The AUROC from the optimal model was 0.82 ± 0.1 in the external testing set. The tumor shape Maximum 3D Diameter plays the most vital role in the model decision.ConclusionsThe automated surgical decision framework for partial or radical nephrectomy based on 3D-CT multi-level anatomical features exhibits robust performance in renal cell carcinoma. The framework points the way towards guiding surgery through medical images and machine learning.Clinical relevance statementWe proposed an automated analytic framework that can assist surgeons in partial or radical nephrectomy decision-making. The framework points the way towards guiding surgery through medical images and machine learning.Key Points• The 3D-CT multi-level anatomical features provide a more accurate prediction of surgical decision-making for partial or radical nephrectomy in renal cell carcinoma.• The data from multicenter study and a strict fivefold cross-validation strategy, both internal validation set and external testing set, can be easily transferred to different tasks of new datasets.• The quantitative decomposition of the prediction model was conducted to explore the contribution of each extracted feature.

Robust and efficient abdominal CT segmentation using shape constrained multi-scale attention network

PurposeAlthough many deep learning-based abdominal multi-organ segmentation networks have been proposed, the various intensity distributions and organ shapes of the CT images from multi-center, multi-phase with various diseases introduce new challenges for robust abdominal CT segmentation. To achieve robust and efficient abdominal multi-organ segmentation, a new two-stage method is presented in this study. MethodsA binary segmentation network is used for coarse localization, followed by a multi-scale attention network for the fine segmentation of liver, kidney, spleen, and pancreas. To constrain the organ shapes produced by the fine segmentation network, an additional network is pre-trained to learn the shape features of the organs with serious diseases and then employed to constrain the training of the fine segmentation network. ResultsThe performance of the presented segmentation method was extensively evaluated on the multi-center data set from the Fast and Low GPU Memory Abdominal oRgan sEgmentation (FLARE) challenge, which was held in conjunction with International Conference on Medical Image Computing and Computer Assisted Intervention (MICCAI) 2021. Dice Similarity Coefficient (DSC) and Normalized Surface Dice (NSD) were calculated to quantitatively evaluate the segmentation accuracy and efficiency. An average DSC and NSD of 83.7% and 64.4% were achieved, and our method finally won the second place among more than 90 participating teams. ConclusionsThe evaluation results on the public challenge demonstrate that our method shows promising performance in robustness and efficiency, which may promote the clinical application of the automatic abdominal multi-organ segmentation.

2.5D MFFAU-Net: a convolutional neural network for kidney segmentation

BackgroundKidney tumors have become increasingly prevalent among adults and are now considered one of the most common types of tumors. Accurate segmentation of kidney tumors can help physicians assess tumor complexity and aggressiveness before surgery. However, segmenting kidney tumors manually can be difficult because of their heterogeneity.MethodsThis paper proposes a 2.5D MFFAU-Net (multi-level Feature Fusion Attention U-Net) to segment kidneys, tumors and cysts. First, we propose a 2.5D model for learning to combine and represent a given slice in 2D slices, thereby introducing 3D information to balance memory consumption and model complexity. Then, we propose a ResConv architecture in MFFAU-Net and use the high-level and low-level feature in the model. Finally, we use multi-level information to analyze the spatial features between slices to segment kidneys and tumors.ResultsThe 2.5D MFFAU-Net was evaluated on KiTS19 and KiTS21 kidney datasets and demonstrated an average dice score of 0.924 and 0.875, respectively, and an average Surface dice (SD) score of 0.794 in KiTS21.ConclusionThe 2.5D MFFAU-Net model can effectively segment kidney tumors, and the results are comparable to those obtained with high-performance 3D CNN models, and have the potential to serve as a point of reference in clinical practice.

A PPM-based UNet for Tumour and Kidney Segmentation in CTScans

ABSTRACT Kidney cancer is one of the leading causes of cancer death worldwide. The high rates of mortality and occurrences of cancer show the importance of research and the development of means for the early diagnosis. The correct segmentation of kidney becomes an essential step to help with the case’s analysis and degree of severity. Computational methods have been used for this purpose, emerging as alternatives to manual segmentation and as mechanisms to reduce fatigue during diagnosis. Fully convolutional neural networks have presented great prominence in automatic segmentation in medical images. This work aims to build a novel network based on the U-Net architecture, using the concept of Pyramid Pooling Module (PPM) blocks that we call PPM-UNet. We propose this model to force learning features in multiple resolutions during their encoding phase and emphasise differences between similar regions. The hypothesis is that in cases where the texture of the images is very similar, the PPM block will be able to represent information with greater precision. In the experiments, we verified that the PPM-UNet obtained promising results of 0.9466 of Dice for the kidney and 0.9383 of Dice for tumour regions when the method was applied to the Kits19 database.

Automatic Kidney Segmentation Method Based on an Enhanced Generative Adversarial Network

When deciding on a kidney tumor's diagnosis and treatment, it is critical to take its morphometry into account. It is challenging to undertake a quantitative analysis of the association between kidney tumor morphology and clinical outcomes due to a paucity of data and the need for the time-consuming manual measurement of imaging variables. To address this issue, an autonomous kidney segmentation technique, namely SegTGAN, is proposed in this paper, which is based on a conventional generative adversarial network model. Its core framework includes a discriminator network with multi-scale feature extraction and a fully convolutional generator network made up of densely linked blocks. For qualitative and quantitative comparisons with the SegTGAN technique, the widely used and related medical image segmentation networks U-Net, FCN, and SegAN are used. The experimental results show that the Dice similarity coefficient (DSC), volumetric overlap error (VOE), accuracy (ACC), and average surface distance (ASD) of SegTGAN on the Kits19 dataset reach 92.28%, 16.17%, 97.28%, and 0.61 mm, respectively. SegTGAN outscores all the other neural networks, which indicates that our proposed model has the potential to improve the accuracy of CT-based kidney segmentation.

PD08-04 VIRTUAL RENAL MASS BIOPSY: PREDICTING RENAL TUMOR HISTOLOGY ON ABDOMINAL CT IMAGES USING MACHINE LEARNING

You have accessJournal of UrologyCME1 Apr 2023PD08-04 VIRTUAL RENAL MASS BIOPSY: PREDICTING RENAL TUMOR HISTOLOGY ON ABDOMINAL CT IMAGES USING MACHINE LEARNING Abhinav Khanna, Vidit Sharma, Adriana Gregory, H. Chase Gottlich, Cole J. Cook, Jason Klug, Christine Lohse, Theodora Potretzke, Aaron Potretzke, Stephen A. Boorjian, R. Houston Thompson, Naoki Takahashi, Bradley Erickson, John Cheville, Timothy Kline, and Bradley Leibovich Abhinav KhannaAbhinav Khanna More articles by this author , Vidit SharmaVidit Sharma More articles by this author , Adriana GregoryAdriana Gregory More articles by this author , H. Chase GottlichH. Chase Gottlich More articles by this author , Cole J. CookCole J. Cook More articles by this author , Jason KlugJason Klug More articles by this author , Christine LohseChristine Lohse More articles by this author , Theodora PotretzkeTheodora Potretzke More articles by this author , Aaron PotretzkeAaron Potretzke More articles by this author , Stephen A. BoorjianStephen A. Boorjian More articles by this author , R. Houston ThompsonR. Houston Thompson More articles by this author , Naoki TakahashiNaoki Takahashi More articles by this author , Bradley EricksonBradley Erickson More articles by this author , John ChevilleJohn Cheville More articles by this author , Timothy KlineTimothy Kline More articles by this author , and Bradley LeibovichBradley Leibovich More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003239.04AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Up to 20% of renal tumors are benign and may not require treatment. However, benign versus malignant renal tumors cannot be distinguished using cross-sectional imaging. Renal mass biopsy is a possible solution, but biopsy is invasive and has notable non-diagnostic and false negative rates. As a result, many patients proceed directly to treatment, including some who undergo extirpative surgery for benign tumors. We aim to develop a radiomics and machine learning model for distinguishing between oncocytoma versus malignant renal neoplasms based on abdominal CT images. METHODS: Our institutional registry was queried for patients who underwent surgical treatment of renal tumors from 2000-2018. All surgical specimens underwent pathology re-review by an expert genitourinary pathologist. A total of 843 images from 609 patients (434 images of oncocytoma and 409 images of malignant renal tumor) were included. A previously developed artificial intelligence algorithm for segmentation of kidney, cyst, and tumor area was applied. Images were preprocessed by resampling via linear interpolation to 0.8 mm x 0.8 mm x 5 mm, window/level=440/40, and intensity normalized [0 255]. Features were extracted for the tumor region using PyRadiomics with a fixed bin width of 16. Both unsupervised (PCA, tSNE) and supervised (LR, SVM) machine learning approaches were explored. Data was split 80:20, and cross validation was used in the training/validation set parameters. The best performing model based on F1 macro was then calibrated using an isotonic calibration approach. RESULTS: Dimensionality reduction showed adequate class separation (Figure 1A). The top 10 radiomic features showed a mixture of first and second order features (Figure 1B). The final model reached an accuracy of 0.90 (Figure 1C). Of the 168 images in the testing set, only 2 benign cases were predicted to be malignant, and 15 malignant cases were predicted to be benign. The AUC for oncocytoma versus malignant histology prediction was 0.96 with CI [0.93 0.98] (Figure 1D). CONCLUSIONS: We developed a machine learning model for accurately distinguishing benign oncocytoma versus malignant renal tumors based on CT images alone. This may provide an opportunity for non-invasive risk stratification of solid renal neoplasms, which could potentially reduce overtreatment of renal tumors. Source of Funding: None © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e233 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Abhinav Khanna More articles by this author Vidit Sharma More articles by this author Adriana Gregory More articles by this author H. Chase Gottlich More articles by this author Cole J. Cook More articles by this author Jason Klug More articles by this author Christine Lohse More articles by this author Theodora Potretzke More articles by this author Aaron Potretzke More articles by this author Stephen A. Boorjian More articles by this author R. Houston Thompson More articles by this author Naoki Takahashi More articles by this author Bradley Erickson More articles by this author John Cheville More articles by this author Timothy Kline More articles by this author Bradley Leibovich More articles by this author Expand All Advertisement PDF downloadLoading ...