CME Radiology Grand Rounds
Where:
Li Ka Shing Learning and Knowledge Center or James H. Clark Center & Zoom.
2024 - 2025 Schedule
Friday, April 11, 2025
12:00-1:00PM | LK120 & Zoom
Etta K. Moskowitz Fund Resident Research Awards
Michael Ghijsen
CT indicators of fluid overload serve as opportunistic predictors of clinical outcome
Over the past two decades, transcatheter aortic valve replacement (TAVR) has emerged as a safe and effective alternative to surgical valve replacement. Despite continued advances in valve technology and operator skill, appropriate candidate selection remains critical to patient outcomes. Recently, several studies have demonstrated increased risk of adverse outcomes associated with fluid overload measured using a point of care technique known as bioelectrical impedance spectroscopy (BIS). Because all patients undergo computed tomographic (CT) imaging prior to TAVR, it is possible to evaluate for fluid overload opportunistically. In this work, TAVR-planning CT studies were analyzed for the presence or absence of fluid overload by evaluating for pulmonary edema, pleural effusion, anasarca, and ascites. Cohorts with and without fluid overload were compared using Kaplan-Meier survival analysis and the Cox proportional hazards mode. Kaplan Meier analysis demonstrated significant differences in survival between the two groups, and the Cox model produced a hazard ratio of 2.93 with a p-value of 0.01. Overall, this work shows that CT evidence of fluid overload prior to TAVR is associated with increased mortality.
Aakash Gupta
Liver Venous Deprivation Versus Portal Vein Embolization: A focus on perihilar cholangiocarcinoma
Post-hepatectomy liver failure from insufficient remnant liver following curative intent major hepatectomy is a significant source of morbidity and mortality in the post-operative period, reported in up to 10% of cases. Since 1920, several different surgical and image-guided techniques have emerged to hypertrophy the future liver remnant prior to hepatic resection. Due to a minimally invasive approach, portal vein embolization has largely surpassed surgical portal vein ligation as a pre-operative technique to redirect portal venous flow from the tumor-bearing hepatic lobe to the contralateral hepatic lobe. More recently, single-session hepatic vein embolization and portal vein embolization, termed liver venous deprivation, has demonstrated a greater degree of future liver remnant hypertrophy compared to portal vein embolization alone. However, whether liver venous deprivation is safe and efficacious across the broad range of primary and secondary hepatic malignancies is still unknown. Here, we present our preliminary work analyzing safety and progression to planned surgical resection of liver venous deprivation compared to portal vein embolization. We focus on the safety of these techniques in patients with perihilar cholangiocarcinoma who have histories of cholestasis, cholangitis, and biliary instrumentation. This work has implications for maximizing a patient's chance of undergoing curative intent surgical hepatic resection.
Preya Shah
A Clinical and Imaging Fused Deep Learning Model Matches Expert Clinician Prediction of 90-Day Stroke Outcome
Predicting long-term clinical outcome in acute ischemic stroke (AIS) is beneficial for setting patient expectations, guiding treatment and rehabilitation strategies, and designing clinical trials. Although some studies have attempted to predict long-term functional outcomes, these traditional methodologies rely on manually crafted imaging features [1-3]. This study used a deep learning-based predictive model to predict 90-day modified Rankin Scale (mRS) outcomes and compared model predictions with those made by clinicians.
Jessica Wen
Y-90 Radioembolization for Unresectable Hepatocellular Carcinoma: Correlation of Alpha-Fetoprotein Response Patterns with Disease Progression and Survival
Over the past decade, transarterial radioembolization with Yttrium-90 (Y-90) has become a valuable treatment strategy for hepatocellular carcinoma (HCC). Y-90 radioembolization can be performed as a bridge to liver transplant, optimization for resection, or primary/combination therapy for unresectable or advanced HCC. Currently, there are no consensus guidelines for predicting survival and progression outcomes in patients with HCC after Y-90 radioembolization. The aims of this study were to evaluate AFP response patterns in patients after Y-90 treatment, correlate response patterns with overall survival, distant progression-free survival, and progression-free survival, and finally to describe patterns of progression to extrahepatic metastases after treatment.
Friday, April 25, 2025
12:00-1:00PM | CAM Grand Rounds & Zoom
Mirabela Rusu, PhD, MS
Assistant Professor, Stanford University
Bridging the gap between ultrasound and MRI for cancer detection
Clinical care is inherently multimodal, with medical image data collected throughout the patient’s journey. For example, a patient at risk of cancer will undergo an ultrasound-guided biopsy, and when available with MRI revealing regions to be targeted due to higher risk to harbor aggressive disease. This biopsy procedure seeks to collect tissue samples for pathology and will inform treatment strategies for best outcomes. This common scenario provides unique opportunities for Artificial Intelligence (AI) methods to effectively integrate multimodal data, and learn imaging signatures in patients with known outcomes, to enable early cancer detection for patients at risk. This presentation focuses on showcasing some of our AI methods that bridge the gap between highly informative modalities, e.g., MRI, and lower resolution modalities, e.g., ultrasound. These methods rely on multimodal image registration, image feature fusion, or integration of patient-specific data and population-specific information and rely on AI approaches for effective integration. While the learning is done with multiple imaging modalities, the inference requires only the low-resolution modality, e.g., ubiquitous conventional ultrasound, with applications in low-resource settings.
Curtis Langlotz, MD, PhD, FACMI, FSIIM
Professor, Stanford University
The Future of Radiology in the AI Era
Artificial intelligence (AI) is an incredibly powerful tool for building computer vision systems that support the work of radiologists. Over the last decade, artificial intelligence methods have revolutionized the analysis of digital images, leading to high interest and explosive growth in the use of AI and machine learning methods to analyze clinical images and text. Deep learning methods are now being developed for image reconstruction, imaging quality assurance, imaging triage, computer-aided detection, computer-aided classification, and radiology report drafting. The systems have the potential to provide real-time assistance to radiologists and other imaging professionals, thereby reducing diagnostic errors, improving patient outcomes, and reducing costs. We will review the origins of AI and its applications to medical imaging and associated text, define key terms, and show examples of real-world applications that suggest how AI and large language models may change the practice of medicine. We will also review key shortcomings and challenges that may limit the application of these new methods.
