A Primer for Utilizing Deep Learning and Abdominal MRI Imaging Features to Monitor Autosomal Dominant Polycystic Kidney Disease Progression.

Abdominal imaging of autosomal dominant polycystic kidney disease (ADPKD) has historically focused on detecting complications such as cyst rupture, cyst infection, obstructing renal calculi, and pyelonephritis; discriminating complex cysts from renal cell carcinoma; and identifying sources of abdominal pain. Many imaging features of ADPKD are incompletely evaluated or not deemed to be clinically significant, and because of this, treatment options are limited. However, total kidney volume (TKV) measurement has become important for assessing the risk of disease progression (i.e., Mayo Imaging Classification) and predicting tolvaptan treatment's efficacy. Deep learning for segmenting the kidneys has improved these measurements' speed, accuracy, and reproducibility. Deep learning models can also segment other organs and tissues, extracting additional biomarkers to characterize the extent to which extrarenal manifestations complicate ADPKD. In this concept paper, we demonstrate how deep learning may be applied to measure the TKV and how it can be extended to measure additional features of this disease.

Quantitative susceptibility mapping for detection of kidney stones, hemorrhage differentiation, and cyst classification in ADPKD.

BACKGROUND AND PURPOSE: The objective is to demonstrate feasibility of quantitative susceptibility mapping (QSM) in autosomal dominant polycystic kidney disease (ADPKD) patients and to compare imaging findings with traditional T1/T2w magnetic resonance imaging (MRI). METHODS: Thirty-three consecutive patients (11 male, 22 female) diagnosed with ADPKD were initially selected. QSM images were reconstructed from the multiecho gradient echo data and compared to co-registered T2w, T1w, and CT images. Complex cysts were identified and classified into distinct subclasses based on their imaging features. Prevalence of each subclass was estimated. RESULTS: QSM visualized two renal calcifications measuring 9 and 10 mm and three pelvic phleboliths measuring 2 mm but missed 24 calcifications measuring 1 mm or less and 1 larger calcification at the edge of the field of view. A total of 121 complex T1 hyperintense/T2 hypointense renal cysts were detected. 52 (43%) Cysts appeared hyperintense on QSM consistent with hemorrhage; 60 (49%) cysts were isointense with respect to simple cysts and normal kidney parenchyma, while the remaining 9 (7%) were hypointense. The presentation of the latter two complex cyst subtypes is likely indicative of proteinaceous composition without hemorrhage. CONCLUSION: Our results indicate that QSM of ADPKD kidneys is possible and uniquely suited to detect large renal calculi without ionizing radiation and able to identify properties of complex cysts unattainable with traditional approaches.

Test Retest Reproducibility of Organ Volume Measurements in ADPKD Using 3D Multimodality Deep Learning.

RATIONALE AND OBJECTIVES: Following autosomal dominant polycystic kidney disease (ADPKD) progression by measuring organ volumes requires low measurement variability. The objective of this study is to reduce organ volume measurement variability on MRI of ADPKD patients by utilizing all pulse sequences to obtain multiple measurements which allows outlier analysis to find errors and averaging to reduce variability. MATERIALS AND METHODS: In order to make measurements on multiple pulse sequences practical, a 3D multi-modality multi-class segmentation model based on nnU-net was trained/validated using T1, T2, SSFP, DWI and CT from 413 subjects. Reproducibility was assessed with test-re-test methodology on ADPKD subjects (n = 19) scanned twice within a 3-week interval correcting outliers and averaging the measurements across all sequences. Absolute percent differences in organ volumes were compared to paired students t-test. RESULTS: Dice similarlity coefficient > 97%, Jaccard Index > 0.94, mean surface distance < 1 mm and mean Hausdorff Distance < 2 cm for all three organs and all five sequences were found on internal (n = 25), external (n = 37) and test-re-test reproducibility assessment (38 scans in 19 subjects). When averaging volumes measured from five MRI sequences, the model automatically segmented kidneys with test-re-test reproducibility (percent absolute difference between exam 1 and exam 2) of 1.3% which was better than all five expert observers. It reliably stratified ADPKD into Mayo Imaging Classification (area under the curve=100%) compared to radiologist. CONCLUSION: 3D deep learning measures organ volumes on five MRI sequences leveraging the power of outlier analysis and averaging to achieve 1.3% total kidney test-re-test reproducibility.

Clinical Quality Control of MRI Total Kidney Volume Measurements in Autosomal Dominant Polycystic Kidney Disease.

Total kidney volume measured on MRI is an important biomarker for assessing the progression of autosomal dominant polycystic kidney disease and response to treatment. However, we have noticed that there can be substantial differences in the kidney volume measurements obtained from the various pulse sequences commonly included in an MRI exam. Here we examine kidney volume measurement variability among five commonly acquired MRI pulse sequences in abdominal MRI exams in 105 patients with ADPKD. Right and left kidney volumes were independently measured by three expert observers using model-assisted segmentation for axial T2, coronal T2, axial single-shot fast spin echo (SSFP), coronal SSFP, and axial 3D T1 images obtained on a single MRI from ADPKD patients. Outlier measurements were analyzed for data acquisition errors. Most of the outlier values (88%) were due to breathing during scanning causing slice misregistration with gaps or duplication of imaging slices (n = 35), slice misregistration from using multiple breath holds during acquisition (n = 25), composing of two overlapping acquisitions (n = 17), or kidneys not entirely within the field of view (n = 4). After excluding outlier measurements, the coefficient of variation among the five measurements decreased from 4.6% pre to 3.2%. Compared to the average of all sequences without errors, TKV measured on axial and coronal T2 weighted imaging were 1.2% and 1.8% greater, axial SSFP was 0.4% greater, coronal SSFP was 1.7% lower and axial T1 was 1.5% lower than the mean, indicating intrinsic measurement biases related to the different MRI contrast mechanisms. In conclusion, MRI data acquisition errors are common but can be identified using outlier analysis and excluded to improve organ volume measurement consistency. Bias toward larger volume measurements on T2 sequences and smaller volumes on axial T1 sequences can also be mitigated by averaging data from all error-free sequences acquired.

Effect of Averaging Measurements From Multiple MRI Pulse Sequences on Kidney Volume Reproducibility in Autosomal Dominant Polycystic Kidney Disease.

BACKGROUND: Total kidney volume (TKV) is an important biomarker for assessing kidney function, especially for autosomal dominant polycystic kidney disease (ADPKD). However, TKV measurements from a single MRI pulse sequence have limited reproducibility, ± ~5%, similar to ADPKD annual kidney growth rates. PURPOSE: To improve TKV measurement reproducibility on MRI by extending artificial intelligence algorithms to automatically segment kidneys on T1-weighted, T2-weighted, and steady state free precession (SSFP) sequences in axial and coronal planes and averaging measurements. STUDY TYPE: Retrospective training, prospective testing. SUBJECTS: Three hundred ninety-seven patients (356 with ADPKD, 41 without), 75% for training and 25% for validation, 40 ADPKD patients for testing and 17 ADPKD patients for assessing reproducibility. FIELD STRENGTH/SEQUENCE: T2-weighted single-shot fast spin echo (T2), SSFP, and T1-weighted 3D spoiled gradient echo (T1) at 1.5 and 3T. ASSESSMENT: 2D U-net segmentation algorithm was trained on images from all sequences. Five observers independently measured each kidney volume manually on axial T2 and using model-assisted segmentations on all sequences and image plane orientations for two MRI exams in two sessions separated by 1-3 weeks to assess reproducibility. Manual and model-assisted segmentation times were recorded. STATISTICAL TESTS: Bland-Altman, Schapiro-Wilk (normality assessment), Pearson's chi-squared (categorical variables); Dice similarity coefficient, interclass correlation coefficient, and concordance correlation coefficient for analyzing TKV reproducibility. P-value < 0.05 was considered statistically significant. RESULTS: In 17 ADPKD subjects, model-assisted segmentations of axial T2 images were significantly faster than manual segmentations (2:49 minute vs. 11:34 minute), with no significant absolute percent difference in TKV (5.9% vs. 5.3%, P = 0.88) between scans 1 and 2. Absolute percent differences between the two scans for model-assisted segmentations on other sequences were 5.5% (axial T1), 4.5% (axial SSFP), 4.1% (coronal SSFP), and 3.2% (coronal T2). Averaging measurements from all five model-assisted segmentations significantly reduced absolute percent difference to 2.5%, further improving to 2.1% after excluding an outlier. DATA CONCLUSION: Measuring TKV on multiple MRI pulse sequences in coronal and axial planes is practical with deep learning model-assisted segmentations and can improve TKV measurement reproducibility more than 2-fold in ADPKD. EVIDENCE LEVEL: 2 TECHNICAL EFFICACY: Stage 1.

Pleural Effusions on MRI in Autosomal Dominant Polycystic Kidney Disease.

Autosomal dominant polycystic kidney disease (ADPKD) has cystic fluid accumulations in the kidneys, liver, pancreas, arachnoid spaces as well as non-cystic fluid accumulations including pericardial effusions, dural ectasia and free fluid in the male pelvis. Here, we investigate the possible association of ADPKD with pleural effusion. ADPKD subjects (n = 268) and age-gender matched controls without ADPKD (n = 268) undergoing body magnetic resonance imaging from mid-thorax down into the pelvis were independently evaluated for pleural effusion by 3 blinded expert observers. Subjects with conditions associated with pleural effusion were excluded from both populations. Clinical and laboratory data as well as kidney, liver and spleen volume, pleural fluid volume, free pelvic fluid and polycystic kidney disease genotype were evaluated. Pleural effusions were observed in 56 of 268 (21%) ADPKD subjects compared with 21 of 268 (8%) in controls (p < 0.0001). In a subpopulation controlling for renal function by matching estimated glomerular filtration rate (eGFR), 28 of 110 (25%) ADPKD subjects had pleural effusions compared to 5 of 110 (5%) controls (p < 0.001). Pleural effusions in ADPKD subjects were more prevalent in females (37/141; 26%) than males (19/127,15%; p = 0.02) and in males were weakly correlated with the presence of free pelvic fluid (r = 0.24, p = 0.02). ADPKD subjects with pleural effusions were younger (48 ± 14 years old vs. 43 ± 14 years old) and weighed less (77 vs. 70 kg; p ≤ 0.02) than those without pleural effusions. For ADPKD subjects with pleural effusions, the mean volume of fluid layering dependently in the posterior−inferior thorax was 19 mL and was not considered to be clinically significant. Pleural effusion is associated with ADPKD, but its role in the pathogenesis of ADPKD requires further evaluation.

Deep Learning Automation of Kidney, Liver, and Spleen Segmentation for Organ Volume Measurements in Autosomal Dominant Polycystic Kidney Disease.

Organ volume measurements are a key metric for managing ADPKD (the most common inherited renal disease). However, measuring organ volumes is tedious and involves manually contouring organ outlines on multiple cross-sectional MRI or CT images. The automation of kidney contouring using deep learning has been proposed, as it has small errors compared to manual contouring. Here, a deployed open-source deep learning ADPKD kidney segmentation pipeline is extended to also measure liver and spleen volumes, which are also important. This 2D U-net deep learning approach was developed with radiologist labeled T2-weighted images from 215 ADPKD subjects (70% training = 151, 30% validation = 64). Additional ADPKD subjects were utilized for prospective (n = 30) and external (n = 30) validations for a total of 275 subjects. Image cropping previously optimized for kidneys was included in training but removed for the validation and inference to accommodate the liver which is closer to the image border. An effective algorithm was developed to adjudicate overlap voxels that are labeled as more than one organ. Left kidney, right kidney, liver and spleen labels had average errors of 3%, 7%, 3%, and 1%, respectively, on external validation and 5%, 6%, 5%, and 1% on prospective validation. Dice scores also showed that the deep learning model was close to the radiologist contouring, measuring 0.98, 0.96, 0.97 and 0.96 on external validation and 0.96, 0.96, 0.96 and 0.95 on prospective validation for left kidney, right kidney, liver and spleen, respectively. The time required for manual correction of deep learning segmentation errors was only 19:17 min compared to 33:04 min for manual segmentations, a 42% time saving (p = 0.004). Standard deviation of model assisted segmentations was reduced to 7, 5, 11, 5 mL for right kidney, left kidney, liver and spleen respectively from 14, 10, 55 and 14 mL for manual segmentations. Thus, deep learning reduces the radiologist time required to perform multiorgan segmentations in ADPKD and reduces measurement variability.

Major hemorrhage and mortality in COVID-19 patients on therapeutic anticoagulation for venous thromboembolism.

We observed multiple fatal intracranial hemorrhages shortly after initiating therapeutic anticoagulation for treatment of venous thromboembolism (VTE) in COVID-19 patients suggesting increased anticoagulation risk associated with COVID-19. The objective of this study is to quantify risk of major hemorrhage in hospitalized COVID-19 patients on therapeutic anticoagulation for deep venous thrombosis (DVT) or pulmonary embolism (PE). Hospitalized patients with COVID-19 receiving therapeutic anticoagulation for DVT, PE or both at four New York City hospitals were evaluated for hemorrhagic complications. These were categorized as major (including fatal) or clinically relevant non-major according to the criteria of the International Society of Thrombosis and Haemostasis. Hemorrhagic complications were correlated with clinical and laboratory data, ICD-10 code diagnoses and type of anticoagulation treatment. Minor hemorrhages were excluded. Major/clinically relevant hemorrhages occurred in 36 of 170 (21%) hospitalized COVID-19 patients being treated with therapeutic anticoagulation for VTE including 4 (2.4%) fatal hemorrhages. Hemorrhage was 3.4 times more likely with unfractionated heparin 27/76 (36%) compared to 8/81 (10%) with low molecular weight heparin (p = 0.002). Multivariate analysis showed that major hemorrhage was associated with intubation (p = 0.04) and elevated serum LDH (p < 0.001) and low fibrinogen (p = 0.05). Increased risk of hemorrhagic complications in treating VTE in hospitalized COVID-19 patients should be considered especially when using unfractionated heparin, in intubated patients, with low fibrinogen and/or elevated LDH. Checking serum fibrinogen and LDH before initiating therapeutic anticoagulation and monitoring coagulation parameters frequently may reduce bleeding complications.

Deployed Deep Learning Kidney Segmentation for Polycystic Kidney Disease MRI.

This study develops, validates, and deploys deep learning for automated total kidney volume (TKV) measurement (a marker of disease severity) on T2-weighted MRI studies of autosomal dominant polycystic kidney disease (ADPKD). The model was based on the U-Net architecture with an EfficientNet encoder, developed using 213 abdominal MRI studies in 129 patients with ADPKD. Patients were randomly divided into 70% training, 15% validation, and 15% test sets for model development. Model performance was assessed using Dice similarity coefficient (DSC) and Bland-Altman analysis. External validation in 20 patients from outside institutions demonstrated a DSC of 0.98 (IQR, 0.97-0.99) and a Bland-Altman difference of 2.6% (95% CI: 1.0%, 4.1%). Prospective validation in 53 patients demonstrated a DSC of 0.97 (IQR, 0.94-0.98) and a Bland-Altman difference of 3.6% (95% CI: 2.0%, 5.2%). Last, the efficiency of model-assisted annotation was evaluated on the first 50% of prospective cases (n = 28), with a 51% mean reduction in contouring time (P < .001), from 1724 seconds (95% CI: 1373, 2075) to 723 seconds (95% CI: 555, 892). In conclusion, our deployed artificial intelligence pipeline accurately performs automated segmentation for TKV estimation of polycystic kidneys and reduces expert contouring time. Keywords: Convolutional Neural Network (CNN), Segmentation, Kidney ClinicalTrials.gov identification no.: NCT00792155 Supplemental material is available for this article. © RSNA, 2022.

Pericardial Effusion on MRI in Autosomal Dominant Polycystic Kidney Disease.

Autosomal dominant polycystic kidney disease (ADPKD) has been associated with cardiac abnormalities including mitral valve prolapse and aneurysmal dilatation of the aortic root. Herein, we investigated the potential association of pericardial effusion with ADPKD. Subjects with ADPKD (n = 117) and control subjects without ADPKD matched for age, gender and renal function (n = 117) undergoing MRI including ECG-gated cine MRI of the aorta and heart were evaluated for pericardial effusion independently by three observers measuring the maximum pericardial effusion thickness in diastole using electronic calipers. Pericardial effusion thickness was larger in ADPKD subjects compared to matched controls (Mann-Whitney p = 0.001) with pericardial effusion thickness >5 mm observed in 24 of 117 (21%) ADPKD subjects compared to 4 of 117 (3%) controls (p = 0.00006). Pericardial effusion thickness in ADPKD was associated with female gender patients (1.2 mm greater than in males, p = 0.03) and pleural effusion thickness (p < 0.001) in multivariate analyses. No subjects exhibited symptoms related to pericardial effusion or required pericardiocentesis. In conclusion, pericardial effusion appears to be more prevalent in ADPKD compared to controls. Although in this retrospective cross-sectional study we did not identify clinical significance, future investigations of pericardial effusion in ADPKD subjects may help to more fully understand its role in this disease.

Pulmonary Embolism in Hospitalized Patients with COVID-19: A Multicenter Study.

Background Pulmonary embolism (PE) commonly complicates SARS-CoV-2 infection, but incidence and mortality reported in single-center studies, along with risk factors, vary. Purpose To determine the incidence of PE in patients with COVID-19 and its associations with clinical and laboratory parameters. Materials and Methods In this HIPAA-compliant study, electronic medical records were searched retrospectively for demographic, clinical, and laboratory data and outcomes among patients with COVID-19 admitted at four hospitals from March through June 2020. PE found at CT pulmonary angiography and perfusion scintigraphy was correlated with clinical and laboratory parameters. The d-dimer level was used to predict PE, and the obtained threshold was externally validated among 85 hospitalized patients with COVID-19 at a fifth hospital. The association between right-sided heart strain and embolic burden was evaluated in patients with PE undergoing echocardiography. Results A total of 413 patients with COVID-19 (mean age, 60 years ± 16 [standard deviation]; age range, 20-98 years; 230 men) were evaluated. PE was diagnosed in 102 (25%; 95% CI: 21, 29) of 413 hospitalized patients with COVID-19 who underwent CT pulmonary angiography or perfusion scintigraphy. PE was observed in 21 (29%; 95% CI: 19, 41) of 73 patients in the intensive care unit (ICU) versus 81 (24%; 95% CI: 20, 29) of 340 patients who were not in the ICU (P = .37). PE was associated with male sex (odds ratio [OR], 1.74; 95% CI: 1.1, 2.8; P = .02); smoking (OR, 1.86; 95% CI: 1.0, 3.4; P = .04); and increased d-dimer (P < .001), lactate dehydrogenase (P < .001), ferritin (P = .001), and interleukin-6 (P = .02) levels. Mortality in hospitalized patients was similar between patients with PE and those without PE (14% [13 of 102]; 95% CI: 8, 22] vs 13% [40 of 311]; 95% CI: 9, 17; P = .98), suggesting that diagnosis and treatment of PE were not associated with excess mortality. The d-dimer levels greater than 1600 ng/mL [8.761 nmol/L] helped predict PE with 100% sensitivity and 62% specificity in an external validation cohort. Embolic burden was higher in patients with right-sided heart strain among the patients with PE undergoing echocardiography (P = .03). Conclusion Pulmonary embolism (PE) incidence was 25% in patients hospitalized with COVID-19 suspected of having PE. A d-dimer level greater than 1600 ng/mL [8.761 nmol/L] was sensitive for identification of patients who needed CT pulmonary angiography. © RSNA, 2021 Online supplemental material is available for this article. See also the editorial by Ketai in this issue.

Hemorrhagic Cysts and Other MR Biomarkers for Predicting Renal Dysfunction Progression in Autosomal Dominant Polycystic Kidney Disease.

BACKGROUND: Screening for rapidly progressing autosomal dominant polycystic kidney disease (ADPKD) is necessary for assigning and monitoring therapies. Height-adjusted total kidney volume (ht-TKV) is an accepted biomarker for clinical prognostication, but represents only a small fraction of information on abdominal MRI. PURPOSE: To investigate the utility of other MR features of ADPKD to predict progression. STUDY TYPE: Single-center retrospective. POPULATION: Longitudinal data from 186 ADPKD subjects with baseline serum creatinine, PKD gene testing, abdominal MRI measurements, and ≥2 follow-up serum creatinine were reviewed. FIELD STRENGTH/SEQUENCE: 1.5T, T2 -weighted single-shot fast spin echo, T1 -weighted 3D spoiled gradient echo (liver accelerated volume acquisition) and 2D cine velocity encoded gradient echo (phase contrast MRA). ASSESSMENT: Ht-TKV, renal blood flow (RBF), number and fraction of renal and hepatic cysts, bright T1 hemorrhagic renal cysts, and liver and spleen volumes were independently assessed by three observers blinded to estimated glomerular filtration rate (eGFR) data. STATISTICAL TESTS: Linear mixed-effect models were applied to predict eGFR over time using MRI features at baseline adjusted for confounders. Validation was performed in 158 patients who had follow-up MRI using receiver operator characteristic, sensitivity, and specificity. RESULTS: Hemorrhagic cysts, fraction of renal and hepatic cysts, height-adjusted liver and spleen volumes were significant independent predictors of future eGFR (final prediction model R2 = 0.88 P < 0.05). The number of hemorrhagic cysts significantly improved the prediction compared to ht-TKV in predicting future eGFR (area under the curve [AUC] = 0.94, 95% confidence interval [CI]: 0.9-0.94 vs. R2 = 0.9, 95% CI: 0.85-0.9, P = 0.045). For baseline eGFR ≥60 ml/min/1.73m2 , sensitivity for predicting eGFR<45 ml/min/1.73m2 by ht-TKV alone was 29%. Sensitivity increased to 72% with all MRI variables in the model (P < 0.05 = 0.019), whereas specificity was unchanged, 100% vs. 99%. DATA CONCLUSION: Combining multiple MR features including hemorrhagic renal cysts, renal cyst fraction, liver and spleen volume, hepatic cyst fraction, and renal blood flow enhanced sensitivity for predicting eGFR decline in ADPKD compared to the standard model including only ht-TKV. Level of Evidence 2 Technical Efficacy Stage 2 J. MAGN. RESON. IMAGING 2021;53:564-576.

Multivariate analysis of CT imaging, laboratory, and demographical features for prediction of acute kidney injury in COVID-19 patients: a Bi-centric analysis.

PURPOSE: To develop and externally validate a multivariate prediction model for the prediction of acute kidney injury (AKI) in COVID-19, based on baseline renal perfusion from contrast-enhanced CT together with clinical and laboratory parameters. METHODS: In this retrospective IRB-approved study, we identified COVID-19 patients who had a standard-of-care contrast-enhanced abdominal CT scan within 5 days of their COVID-19 diagnosis at our institution (training set; n = 45, mean age 65 years, M/F 23/22) and at a second institution (validation set; n = 41, mean age 61 years, M/F 22/19). The CT renal perfusion parameter, cortex-to-aorta enhancement index (CAEI), was measured in both sets. A multivariate logistic regression model for predicting AKI was constructed from the training set with stepwise feature selection with CAEI together with demographical and baseline laboratory/clinical data used as input variables. Model performance in the training and validation set was evaluated with ROC analysis. RESULTS: AKI developed in 16 patients (35.6%) of the training set and in 6 patients (14.6%) of the validation set. Baseline CAEI was significantly lower in the patients that ultimately developed AKI (P = 0.003). Logistic regression identified a model combining baseline CAEI, blood urea nitrogen, and gender as most significant predictor of AKI. This model showed excellent diagnostic performance for prediction of AKI in the training set (AUC = 0.89, P < 0.001) and good performance in the validation set (AUC 0.78, P = 0.030). CONCLUSION: Our results show diminished renal perfusion preceding AKI and a promising role of CAEI, combined with laboratory and demographic markers, for prediction of AKI in COVID-19.