MICCAI 2023 Tutorial
Scope
Algorithm development in the medical field has progressed as a result of innovative and powerful technical contributions to MIC and CAI. However, the deployment of such methodologies to ultimately improve healthcare requires additional practical considerations, including factors that arise when humans and machines interact to jointly reach decisions. This tutorial presents an overview of a more human-centered approach to building solutions in MIC and CAI, and highlights opportunities for research at this frontier. The tutorial covers an introductory perspective to human-centered design, human-in-the-loop experimentation, and study design. Bringing together researchers with technical- and human factors-first expertise encourages discussions on the challenges and opportunities to build solutions that satisfy user needs.
Keynote Speakers
Professor in the Data Science department at EURECOM, France. Dr. Zuluaga received her Ph.D. in Signal and Image Processing from Université Claude Bernard Lyon & Uniandes. Her current research focuses on the development of machine learning techniques that can be safely deployed in high-risk domains, such as healthcare, by addressing data complexity, low tolerance to errors, and poor reproducibility.
Professor in the Computer Science department at Technische Universität München, Germany, and Johns Hopkins University, USA. Dr. Navab received his Ph.D. at INRIA, University of Paris XI. His research aims to improve the quality of medical intervention and bridge the gap between medicine and computer science, and his interests span medical augmented reality, computer-assisted surgery, medical robotics, computer vision, and machine learning.
Dr. Michael Rosen is an associate professor of anesthesiology and critical care medicine at the Johns Hopkins University School of Medicine. He is a human factors psychologist with research interests in the areas of teamwork and patient safety as well as simulation-based training, performance measurement, naturalistic decision making, and quality and safety improvement.
Dr. Ayse P. Gurses is an associate professor of anesthesiology and critical care medicine at the Johns Hopkins University School of Medicine. She holds a joint appointment in Johns Hopkins’ Bloomberg School of Public Health’s Department of Health Policy and Management. Her areas of expertise include human factors engineering, patient safety, healthcare technology design, and implementation and usability evaluation.
Tutorial Aims
The MICCAI community has demonstrated huge capabilities in developing computational solutions to challenging problems in MIC, CAI, and beyond - most recently by developing innovative and powerful machine learning and artificial intelligence-based algorithms. While the technological progress is evident, questions around the human-in-the-loop deployment or other human factors pertinent to the use of this technology are not yet widely addressed. With the maturity of these approaches and interest in clearly human-facing technology, such as explainable AI, the challenges and possible solutions to developing for and with humans must play a more prominent role in the MICCAI innovation cycle.
This tutorial aims to:
Provide an overview of a more human-centered approach to innovation in MIC and CAI
Highlighting key challenges in adopting design methodology in this space
Summarize important methodology that can help in building solutions that satisfy user needs
Program
October 12, 2023 (Pacific Daylight Time)
08:00 - 08:05: Opening: Humans-in-the-loop - why should we care? (Mathias Unberath) [Slides]
08:05 - 08:45: The need for more human-centered design in MIC and CAI (Catalina Gomez & Sue Min Cho) [Slides]
08:45 - 09:00: A primer on study design - Part I (Sue Min Cho & Catalina Gomez) [Slides]
09:00 - 09:30: A socio-technical systems approach to technology design and implementation (Ayse Gurses, invited keynote) [Slides]
09.30 - 10:00: Closing the loop for the Human-in-the-Loop (Maria A. Zuluaga, invited keynote) [Slides]
10:00 - 10:30: Coffee break
10:30 - 11:00: A primer on study design - Part II (Sue Min Cho & Catalina Gomez) [Slides]
11:00 - 11:30: "Human Factors Principles and Practices for Teaming with Technology" (Michael Rosen, invited keynote) [Slides]
11:30 - 12:00: From IA to AI - The Crucial Role of Human-Centric Design in Fostering Trust and Acceptance (Nassir Navab, invited keynote) [Slides]
12:00 - 12:05: Closing remarks
Abstracts
Maria A. Zuluaga: Closing the loop for the Human-in-the-Loop
In machine learning, the term Human-in-the-Loop refers to a set of techniques in which an expert assists a learning-based system in making highly accurate predictions. In the literature, most of the work focuses on the first part of the interaction between the expert and the system, i.e. how the expert provides inputs to the system. Less attention is placed on the second part of the interaction, where the system provides feedback to the expert, thus closing the so-called loop. Although less studied, the feedback loop is as important as the human input, especially when systems are to be deployed in critical applications, such as those in healthcare. In this talk, we will present a general framework to formalize the feedback loop. We will define the elements that constitute it, discuss different techniques used to implement it and show some examples illustrating different operating scenarios. We will conclude with a brief discussion on the open challenges.
Organizing Team
Catalina Gomez, MSE
Sue Min Cho, MSE
Mathias Unberath, PhD
Questions?
Contact unberath@jhu.edu for more information.