A 3D-2D Hybrid U-Net Convolutional Neural Network Approach to Prostate Organ Segmentation of Multiparametric MRI.

OBJECTIVE: Prostate cancer is the most commonly diagnosed cancer in men in the United States with more than 200,000 new cases in 2018. Multiparametric MRI (mpMRI) is increasingly used for prostate cancer evaluation. Prostate organ segmentation is an essential step of surgical planning for prostate fusion biopsies. Deep learning convolutional neural networks (CNNs) are the predominant method of machine learning for medical image recognition. In this study, we describe a deep learning approach, a subset of artificial intelligence, for automatic localization and segmentation of prostates from mpMRI. MATERIALS AND METHODS: This retrospective study included patients who underwent prostate MRI and ultrasound-MRI fusion transrectal biopsy between September 2014 and December 2016. Axial T2-weighted images were manually segmented by two abdominal radiologists, which served as ground truth. These manually segmented images were used for training on a customized hybrid 3D-2D U-Net CNN architecture in a fivefold cross-validation paradigm for neural network training and validation. The Dice score, a measure of overlap between manually segmented and automatically derived segmentations, and Pearson linear correlation coefficient of prostate volume were used for statistical evaluation. RESULTS: The CNN was trained on 299 MRI examinations (total number of MR images = 7774) of 287 patients. The customized hybrid 3D-2D U-Net had a mean Dice score of 0.898 (range, 0.890-0.908) and a Pearson correlation coefficient for prostate volume of 0.974. CONCLUSION: A deep learning CNN can automatically segment the prostate organ from clinical MR images. Further studies should examine developing pattern recognition for lesion localization and quantification.

Intranodal Lymphangiography with Thoracic Duct Embolization for Treatment of Chyle Leak after Thoracic Outlet Decompression Surgery.

From 2015 to 2019, 9 patients underwent ultrasound-guided intranodal lymphangiography for the treatment of a chyle leak following thoracic outlet decompression surgery. Chyle leaks were identified by Lipiodol (Guerbet, Roissy, France) extravasation near the left supraclavicular surgical bed in all patients. The technical success rate of thoracic duct embolization was 67% (6 of 9), including fluoroscopic transabdominal antegrade access (n = 4) and ultrasound-guided retrograde access in the left neck (n = 2). Clinical success was achieved in 89% of patients (8 of 9). The mean interval from lymphangiography to drain removal was 6.6 days (range, 4-18 d). No patients had a chyle leak recurrence during clinical follow-up (mean, 304 d).

Successful Percutaneous Ultrasonic Lithotripsy of Gallstones.

Background: Acute calculous cholecystitis is the obstruction of the cystic duct by a gallstone that leads to inflammation of the gallbladder necessitating cholecystectomy. Case Series: We present the cases of 2 patients with acute calculous cholecystitis who were deemed ineligible candidates for cholecystectomy because of their complicating medical histories. Both patients initially underwent cholecystostomy and drain placement with interventional radiology for management of acute calculous cholecystitis. Their large gallstones remained refractory to attempts at removal by electrohydraulic lithotripsy via the cholecystostomy access. The patients' gallstones were successfully removed via percutaneous ultrasonic lithotripsy during a collaborative procedure with interventional radiology and urology. Conclusion: An interdisciplinary approach using percutaneous cholecystolithotomy with rigid ultrasonic lithotripsy is an effective method for removing challenging gallstones in patients for whom traditional approaches fail.

Impact of General Anesthesia on the Right Atrial Pressure During Transjugular Intrahepatic Portosystemic Shunt Creation: A Propensity Score Match Analysis.

PURPOSE: To evaluate the effect of general anesthesia on right atrial (RA) pressure measurements during transjugular intrahepatic portosystemic shunt (TIPS) placement using propensity score match analysis. MATERIALS: A single-institution database was used to identify 664 patients who underwent TIPS creation under either conscious sedation (CS) or general anesthesia (GA) between 2009 and 2018. A propensity-matched cohort was created using logistic regression of sedation method on demographics, liver disease status, and indications. Paired analyses were performed using mixed models for RA pressure and Cox proportional hazards model with robust standard errors for mortality. RESULTS: Of the 664 patients, 270 patients were matched based on similar characteristics (135 for GA and 135 for CS). Indications for TIPS creation included intractable ascites (n = 170, 63%), hepatic hydrothorax (n = 30, 11%), variceal bleeding (n = 43, 16%), and other (n = 27, 10%). Pre-TIPS RA pressure was greater in the matched GA group as compared to CS group by a mean of 4.2 mmHg (p < 0.0001). Similarly, post-TIPS RA pressure was greater in the matched GA group as compared to CS group by a mean of 3.3 mmHg (p < 0.0001). Pre- and post-procedure RA pressure was found to have no association with post-procedure mortality (0.8891, HR 1.077; p 0.917, HR 0.997; respectively). CONCLUSIONS: Utilization of GA during TIPS creation raises the intra-procedural RA pressure compared to CS. However, this elevated intra-procedural RA pressure does not appear to be predictive of mortality post-TIPS creation.

A report of 2 cases of the use of the Inari FlowTriever System in the treatment of pulmonary embolism.

The FlowTriever System (Inari Medical, Irvine, California) is the first FDA-approved mechanical thrombectomy device used for treatment of pulmonary embolism. This device enables nonsurgical removal of pulmonary blood clots without the use of thrombolytic medication and its associated risks. We report 2 cases of successful application of the Inari FlowTriever in treatment of pulmonary embolism and right atrial thrombus.

Adrenal vein sampling for ACTH-producing pheochromocytomas.

Adrenocorticotropic hormone (ACTH)-producing pheochromocytoma can cause a variety of clinical manifestations of excess catecholamine and corticosteroid. Anatomic localization of this source of ectopic ACTH is critical to facilitate unilateral adrenalectomy and prevent adrenal insufficiency due to bilateral adrenalectomy. Although nuclear scintigraphy remains the diagnostic gold standard, recent radiotracer supply shortages have necessitated alternative diagnostic paradigms to localize adrenal pheochromocytomas. We present a case where adrenal vein sampling (AVS) was utilized to lateralize an adrenal pheochromocytoma and discuss the approach and nuance as it differs from routine AVS for hyperaldosteronism or hypercortisolism.

Intranodal lymphangiography with thoracic duct embolization for the treatment of chyle leaks after head and neck cancer surgery.

BACKGROUND: Thoracic duct injury is a rare complication of head and neck surgery. Thoracic duct embolization (TDE) has been proposed to manage postoperative chyle leaks. METHODS: Twelve patients who underwent lymphangiography for a chyle leak after head and neck surgery (M:F = 5:7, mean 55 years) were retrospectively reviewed. Lymphangiographic findings, technical success, complications, and clinical outcomes were analyzed. RESULTS: Chyle leak was identified and TDE attempted in 11 of 12 patients. Three patients required repeat TDE. Technical success of TDE was 86% (12/14). Clinical success for patients with technically successful TDE was 90% (9/10). Median time until drain removal was 2.1 days in nine patients with clinical success. Two major complications were encountered, chylothorax after initial TDE, requiring additional TDE and in one case surgical TD ligation. CONCLUSIONS: TDE is a safe treatment for chyle leaks after head and neck surgery with high technical and clinical success rates.

Segmentation of the Prostate Transition Zone and Peripheral Zone on MR Images with Deep Learning.

Purpose To develop a deep learning model to delineate the transition zone (TZ) and peripheral zone (PZ) of the prostate on MR images. Materials and Methods This retrospective study was composed of patients who underwent a multiparametric prostate MRI and an MRI/transrectal US fusion biopsy between January 2013 and May 2016. A board-certified abdominal radiologist manually segmented the prostate, TZ, and PZ on the entire data set. Included accessions were split into 60% training, 20% validation, and 20% test data sets for model development. Three convolutional neural networks with a U-Net architecture were trained for automatic recognition of the prostate organ, TZ, and PZ. Model performance for segmentation was assessed using Dice scores and Pearson correlation coefficients. Results A total of 242 patients were included (242 MR images; 6292 total images). Models for prostate organ segmentation, TZ segmentation, and PZ segmentation were trained and validated. Using the test data set, for prostate organ segmentation, the mean Dice score was 0.940 (interquartile range, 0.930-0.961), and the Pearson correlation coefficient for volume was 0.981 (95% CI: 0.966, 0.989). For TZ segmentation, the mean Dice score was 0.910 (interquartile range, 0.894-0.938), and the Pearson correlation coefficient for volume was 0.992 (95% CI: 0.985, 0.995). For PZ segmentation, the mean Dice score was 0.774 (interquartile range, 0.727-0.832), and the Pearson correlation coefficient for volume was 0.927 (95% CI: 0.870, 0.957). Conclusion Deep learning with an architecture composed of three U-Nets can accurately segment the prostate, TZ, and PZ. Keywords: MRI, Genital/Reproductive, Prostate, Neural Networks Supplemental material is available for this article. © RSNA, 2021.

Outcomes of Artificial Intelligence Volumetric Assessment of Kidneys and Renal Tumors for Preoperative Assessment of Nephron-Sparing Interventions.

Background: Renal-cell carcinoma is the most common kidney cancer and the 13th most common cause of cancer death worldwide. Partial nephrectomy and percutaneous ablation, increasingly utilized to treat small renal masses and preserve renal parenchyma, require precise preoperative imaging interpretation. We sought to develop and evaluate a convolutional neural network (CNN), a type of deep learning (DL) artificial intelligence (AI), to act as a surgical planning aid by determining renal tumor and kidney volumes through segmentation on single-phase CT. Materials and Methods: After Institutional Review Board approval, the CT images of 319 patients were retrospectively analyzed. Two distinct CNNs were developed for (1) bounding cube localization of the right and left hemiabdomen and (2) segmentation of the renal parenchyma and tumor within each bounding cube. Training was performed on a randomly selected cohort of 269 patients. CNN performance was evaluated on a separate cohort of 50 patients using Sorensen-Dice coefficients (which measures the spatial overlap between the manually segmented and neural network-derived segmentations) and Pearson correlation coefficients. Experiments were run on a graphics processing unit-optimized workstation with a single NVIDIA GeForce GTX Titan X (12GB, Maxwell Architecture). Results: Median Dice coefficients for kidney and tumor segmentation were 0.970 and 0.816, respectively; Pearson correlation coefficients between CNN-generated and human-annotated estimates for kidney and tumor volume were 0.998 and 0.993 (p < 0.001), respectively. End-to-end trained CNNs were able to perform renal parenchyma and tumor segmentation on a new test case in an average of 5.6 seconds. Conclusions: Initial experience with automated DL AI demonstrates that it is capable of rapidly and accurately segmenting kidneys and renal tumors on single-phase contrast-enhanced CT scans and calculating tumor and renal volumes.

PI-RADS Version 2 Is an Excellent Screening Tool for Clinically Significant Prostate Cancer as Designated by the Validated International Society of Urological Pathology Criteria: A Retrospective Analysis.

OBJECTIVES: To assess the utility of multiparametric MRI in detecting clinically significant prostate cancer (csPCa) by comparing PI-RADSv2 scores with International Society of Urological Pathology (ISUP) pathologic grading criteria. METHODS: Data from 137 patients were retrospectively analyzed. PI-RADSv2 scores were compared with pathologic grade using ISUP criteria. Pathologic grades were divided into clinically significant (groups 3-5) and clinically insignificant lesions (groups 1-2). Chi-squared analysis was performed for to assess correlation. RESULTS: Sensitivity and specificity of PI-RADSv2 score 3-5 lesions for detecting csPCa was 100% and 18.5%, respectively. Negative predictive value (NPV) is 100% for these lesions. When considering only PI-RADSv2 score 4-5 lesions, sensitivity decreases to 90% and specificity increases to 67.5%, with a NPV of 98.5%. When only PI-RADSv2 score 5 lesions are considered, sensitivity decreases to 50% and specificity increases to 90%, with a NPV of 95%. CONCLUSIONS: Multiparametric MRI has excellent sensitivity for detecting csPCa. Specificity is poor for PI-RADSv2 score 3 lesions but improves significantly for PI-RADSv2 score 4 and 5 lesions. Overall, mpMRI is an excellent screening tool for csPCa, as designated by the recently validated ISUP criteria. ADVANCES IN KNOWLEDGE: Multiple limitations of the longstanding Gleason pathologic scoring system have led to the development of new ISUP pathologic criteria, which is more focused on the clinical significance of lesions. There are currently insufficient studies evaluating and validating the ISUP criteria with PIRADS v2 evaluation of the prostate.

Applications of Artificial Intelligence to Prostate Multiparametric MRI (mpMRI): Current and Emerging Trends.

Prostate carcinoma is one of the most prevalent cancers worldwide. Multiparametric magnetic resonance imaging (mpMRI) is a non-invasive tool that can improve prostate lesion detection, classification, and volume quantification. Machine learning (ML), a branch of artificial intelligence, can rapidly and accurately analyze mpMRI images. ML could provide better standardization and consistency in identifying prostate lesions and enhance prostate carcinoma management. This review summarizes ML applications to prostate mpMRI and focuses on prostate organ segmentation, lesion detection and segmentation, and lesion characterization. A literature search was conducted to find studies that have applied ML methods to prostate mpMRI. To date, prostate organ segmentation and volume approximation have been well executed using various ML techniques. Prostate lesion detection and segmentation are much more challenging tasks for ML and were attempted in several studies. They largely remain unsolved problems due to data scarcity and the limitations of current ML algorithms. By contrast, prostate lesion characterization has been successfully completed in several studies because of better data availability. Overall, ML is well situated to become a tool that enhances radiologists' accuracy and speed.

Multiple renal capillary hemangiomas in a patient with end-stage renal disease.

Renal capillary hemangiomas are rare and benign vascular tumors which are typically incidentally discovered on imaging. Surgical excision is often performed, as imaging appearance is similar to malignant lesions. Renal hemangiomas are typically solitary and unilateral. We present a rare case of multiple renal capillary hemangiomas in a patient with end-stage renal disease. Two hemangiomas were detected on imaging and 2 smaller hemangiomas were detected upon pathological evaluation, suggesting there may be a wider prevalence of smaller, radiographically-occult renal hemangiomas.

Utility of quantitative apparent diffusion coefficient measurements and normalized apparent diffusion coefficient ratios in the diagnosis of clinically significant peripheral zone prostate cancer.

OBJECTIVE: The aim of this study is to evaluate the utility of quantitative apparent diffusion coefficient (ADC) measurements and normalized ADC ratios in multiparametric MRI for the diagnosis of clinically significant peripheral zone (PZ) prostate cancer particularly among equivocally suspicious prostate lesions. METHODS: A retrospective analysis of 95 patients with PZ lesions by PI-RADSv2 criteria, and who underwent subsequent MRI-US fusion biopsy, was approved by an institutional review board. Two radiologists independently measured ADC values in regions of interest (ROIs) of PZ lesions and calculated normalized ADC ratio based on ROIs in the bladder lumen. Diagnostic performance was evaluated using ROC. Inter observer variability was assessed using intraclass correlation coefficient (ICC). RESULTS: Mean ADC and normalized ADC ratios for clinically significant and non-clinically significant lesions were 0.763 × 10-3 mm2 s-1, 29.8%; and 1.135 × 10-3 mm2 s-1, 47.2% (p < 0.001), respectively. Area under the ROC curve (AUC) was 0.880 [95% CI (0.816-0.944) and 0.885 (95% CI (0.814-0.955)] for ADC and ADC ratio, respectively. Optimal AUC threshold for ADC was 0.843 × 10-3 mm2 s-1 (Sn 70.5%, Sp 88.2%) and for normalized ADC was 33.1% (Sn 75.0%, Sp 95.7%). intraclass correlation coefficient was high at 0.889. CONCLUSION: Quantitative ADC measurement in PZ prostate lesions demonstrates excellent diagnostic performance in differentiating clinically significant from non-clinically significant prostate cancer with high inter observer correlation. Advances In knowledge: Quantitative ADC is presented as an additional method to evaluate lesions in mpMRI of the prostate. This technique may be incorporated in new and existing methods to improve detection and discrimination of clinically significant prostate cancer.

Validation of Prostate Imaging-Reporting and Data System Version 2: A Retrospective Analysis.

PURPOSE: Use of magnetic resonance imaging (MRI)/transrectal ultrasound fusion biopsies to determine the accuracy of multiparametric MRI (mpMRI), using Prostate Imaging-Reporting and Data System version 2 (PI-RADSv2), for detecting clinically significant prostate cancer in the overall gland and specifically the peripheral zone (PZ) and transitional zone (TZ). METHODS: A retrospective analysis of patients who underwent fusion biopsy identified 137 men with 231 prostate lesions was approved by the Institutional Review Board. Subjects initially classified under PI-RADSv1 criteria were regraded using PI-RADSv2 by a radiologist blinded to PI-RADSv1 score and biopsy results. Spearman correlation, chi-squared, and logistic regression analysis were performed. RESULTS: There was positive correlation between PI-RADSv2 and Gleason scores (P < 0.001). In the PZ, mpMRI demonstrated 100% sensitivity, 100% negative predictive value, and 35.9% positive predictive value, compared to 100%, 100%, and 27.1%, respectively, for TZ lesions. When predicting clinically significant prostate cancer, the PI-RADSv2 area under the curve for TZ lesions was 0.844 (95% CI: 0.753-0.935, P < 0.001) and 0.769 (95% CI: 0.684-0.854, P < 0.001) for PZ lesions. Combining PI-RADSv2 with additional risk factors (body mass index, prostate-specific antigen density, digital rectal examination) improved the area under curve. CONCLUSIONS: PI-RADSv2 achieves excellent sensitivity and negative predictive value for both PZ and TZ lesions.

Further evidence of emotional allodynia in unmedicated young adults with major depressive disorder.

BACKGROUND: Recent evidence suggests that sensitivity to the emotional sequela of experimental thermal pain(measured by emotional unpleasantness) is heightened in individuals with major depressive disorder(MDD), a phenomenon we termed "emotional allodynia". The aim of this study was to examine whether acute happy and sad mood induction alters emotional allodynia in MDD. We hypothesized that emotional allodynia will be a robust characteristic of individuals with MDD compared to healthy controls. Thus, it would remain following acute mood induction, independent of valence. METHODS: Twenty-one subjects with current MDD and 21 well-matched healthy subjects(HC) received graded brief temperature stimuli following happy and sad mood inductions procedures(MIP). All subjects rated the intensity and affect(pleasantness/unpleasantness) of each stimulus. Sensory(pain intensity) and affective(pain unpleasantness) thresholds were determined by methods of constant stimuli. RESULTS: The MIPs reliably induced happy and sad mood and the resulting induced mood and subjective arousal were not different between the groups at the time of temperature stimulation. Compared to HC, MDD individuals demonstrated emotional allodynia. We found significantly decreased affective pain thresholds whereby significantly lower temperatures became unpleasant in the MDD compared to the HC group. This was not observed for the sensory pain thresholds. Within the MDD, the affective pain thresholds were significantly lower than the corresponding pain intensity thresholds, whereby non-painful temperatures were already unpleasant for the MDD irrespective of the induced mood. This was not observed for the HC groups where the affective and pain intensity thresholds were comparable. CONCLUSIONS: These findings suggest that emotional allodynia may be a chronic characteristic of current MDD. Future studies should determine if emotional allodynia persists after psychological or pharmacological interventions. Finally, longitudinal work should examine whether emotional allodynia is a result of or vulnerability for depression and the role it plays in the increased susceptibility for pain complaints in this disorder.