Fireside Chat

Abstract coming soon.

Seeing the Full Spectrum: Gender Inclusive Care in Radiology

LGBTQIA+ people face lower healthcare utilization rates due to discrimination, poor experiences in healthcare, and barriers to accessing care. There is an increasing need to improve care and reduce health care disparities for the LGBTQIA+ population. The medical community can begin by educating themselves on LGBTQIA+ terminology, using inclusive language and developing cultural competence in clinical settings. In order to achieve this it is first important to understand that Sex and Gender are distinct and that both Sex and Gender exist on continuums. By thoughtfully incorporating this knowledge into our research and clinical practice, the radiology community will enhance the healthcare experiences of all patients.

Learning Objectives: 

1. Define concepts and terminology regarding gender diversity within a gender expansive (non-binary) framework. 
2. Recognize the unique health care experiences of a gender diverse individuals and identify barriers to equitable health care. 
3. Propose strategies and solutions to create a more inclusive and welcoming environment for gender diverse individuals in healthcare.

Multiparametric Ultrasound for Chronic Liver Disease

Chronic liver disease is a worldwide epidemic with estimates of 25-30% people affected. MAFLD (NAFLD) is now the major cause of chronic liver disease. With the large number of affected individuals an inexpensive, safe, rapid, accurate method of evaluation is need to not only make the diagnosis but also to follow-up with patients on treatment. Ultrasound is uniquely able to fulfill these requirements. This lecture will review the use of multiparametric ultrasound including liver stiffness estimation, liver fat estimation, and new techniques that may provide some information on inflammation.

Learning Objectives:

1. Review the use of ARFI based techniques for estimation of liver stiffness
2. Discuss the use of quantitative ultrasound techniques for liver fat estimation
3. Present how to use Multiparametric Ultrasound in Clinical Practice

Reflector (Scout) localization at the time of breast biopsy instead of a biopsy clip: an academic center experience

Preoperative image-guided localization of non-palpable breast findings is pivotal in reducing the need for reoperation by ensuring a negative resection margin and yielding improved cosmetic and patient satisfaction outcomes. Wire-guided localization (WGL) has historically been the gold standard, but emerging alternatives, such as wireless-guided localization (Wi-GL), offer distinct advantages.

Wi-GL stands out by decoupling the localization procedure from surgery, allowing for more flexible surgical scheduling. While radioactive seed was the first commercially available Wi-GL option, it requires strict regulatory control, with careful tracking from seed deployment to retrieval. Radar localization devices, such as the reflector (Scout®) provide additional convenience without the regulatory burden. FDA approval in 2017 of the Scout for long-term use enabled the placement of the reflector at the time of biopsy, eliminating the need for a separate localization procedure for surgically intervened lesions. However, considerations arise regarding the permanent presence of the reflector if surgery does not proceed, impacting future imaging (e.g., MRI) and localization. Also, if the patient undergoes a mastectomy, then localization is essentially not needed, given the relatively higher cost of Wi-GL as compared to WGL or no localization.

Currently, there is little guidance on choosing between wireless devices and biopsy clips during breast biopsy. The study retrospectively evaluates eventual surgical utilization of reflectors placed at the time of biopsy, aiming to identify factors influencing successful utilization. The forthcoming presentation will shed light on these factors and contribute to refining the decision-making process for choosing reflector devices or biopsy clips during breast biopsy.

Mentor: Dr. Brittany Dashevsky MD, DPhil 

Eye-tracking assisted target tumor quantification on FDG-PET/CT scans.

FDG-PET/CT studies are the standard of care for staging and following up of multiple cancers. Currently, assessing disease progression involves time consuming manual measurements of multiple index tumor lesions. Many have attempted building stand-alone AI models to automate this. However, these models often fail when encountering rare long tail cases in real clinical workflow. With advances in eye tracking hardware, we attempt to harness the fastest muscles in the human body in a Human Computer Interaction system for tumor quantification. We show that our baseline algorithm is on average 84% successful in utilizing insights gathered from expert gaze attention and features from PET images for measuring target hypermetabolic lesions. 

Mentors:
Dr. Guido Davidzon
Dr. Curtis Langlotz

A Novel Neural Network Architecture for Native Resolution AAA Segmentation

Abdominal aortic aneurysm (AAA) is common and asymptomatic throughout its disease course, but carries the risk of catastrophic rupture and hemorrhage. Highly accurate aortic segmentation is critical to avoid treatment errors near the treatment threshold as well as in the assessment of candidate pharmacological therapies. Previously published AAA segmentation models suffer from some combination of brittle preprocessing procedures, failure to account for global context, unacceptable downsampling and consequent loss of spatial resolution, lack of end-to-end training, or inadequate validation.

We propose a novel convolutional neural network (CNN) architecture which addresses all major limitations of the prior work. The proposed architecture involves three sequential modules responsible for coarse segmentation, transformation, and fine segmentation, respectively. 71 patients with small AAA enrolled in the LIMIT trial who have undergone baseline abdominopelvic CTA were included. Training and testing were performed using an 80:20 split.

The proposed model produces high-quality, native resolution, volumetric AAA segmentations. Ablation experiments demonstrating the effect of architectural choices and training techniques are performed.

We have presented a novel neural network architecture capable of providing high quality, native-resolution AAA segmentations. The segmentations produced will be valuable both for clinical practice and for assessment of candidate pharmacological therapies. Simple modifications to the proposed structure are likely to be valuable in other tasks where high-resolution segmentations of small structures are required.

Mentor: Dr. Dominik Fleischmann

Dissecting the Tumor Microenvironment with AI/ML Tools

New Ways of Thinking about  Breast Cancer Screening

Extracellular vesicles as non-invasive biomarkers for the management of prostate cancer

Extracellular vesicles (EVs) have emerged as promising biomarkers for cancer detection due to their ability to carry molecular cargo, including DNA, RNA, and proteins, reflecting their cell of origin. This presentation highlights the clinical relevance of tumor-derived EVs in the management of prostate cancer. We will discuss the characteristics of tumor-derived EVs and their potential as diagnostic and prognostic biomarkers in prostate cancer. Furthermore, we will explore how integrating EV analysis with imaging techniques can enhance prostate cancer diagnosis and risk stratification. Additionally, we will address the technical challenges associated with EV purification and profiling. This discussion aims to underscore the significance of EVs as a non-invasive tool in the comprehensive management of prostate cancer.

Session Learning Objectives:
Upon conclusion of this activity, participants will be able to:

• To describe basic principles of extracellular vesicles as non-invasive biomarkers  
• To summarize current state of extracellular vesicle-based liquid biopsies in prostate cancer
• To identify opportunities for integrating liquid biopsy with imaging in oncology

Of Skeletons, Scholars, and Sleuths: A Look Back, A Look Around, and A Quick Look Ahead

The history of musculoskeletal imaging at UCSD will be reviewed with emphasis on the importance of imaging-pathologic correlation. Some predictions for the future will also be introduced.

Learning objective
To illustrate the importance of imaging-pathologic correlation, especially with regard to degenerative and inflammatory disorders of the vertebral column

TBD

Abstract coming soon.