Chi Wan Koo1, Jin Ye Yeo2
1Department of Radiology, Mayo Clinic, Rochester, MN, USA; 2JTD Editorial Office, AME Publishing Company
Correspondence to: Jin Ye Yeo. JTD Editorial Office, AME Publishing Company. Email: jtd@amepc.org
Expert Introduction
Dr. Koo (Figure 1) graduated magna cum laude in Biochemistry with Honors from New York University (NYU) and received her medical degree from New York University School of Medicine. She completed a fellowship in cardiothoracic imaging at the Hospital of University of Pennsylvania. Prior to joining Mayo Clinic, she was an Assistant Professor at NYU School of Medicine. She has served as the site leader for the Mayo Clinic Southwest Minnesota Radiology Convergence Cardiovascular Committee and has been a member of the Mayo Clinic Cancer Center Scientific Review Committee and Mayo Clinic Thoracic eTumor Board. She has collaborated with the Journal of Thoracic Disease (JTD) in multiple capacities over the years, having served as a guest editor for the focus issue entitled “Current Practice in Thoracic Neoplasm Diagnosis, Evaluation and Treatment" for which she was recognized as a Distinguished Guest Editor in 2020 and is currently on the JTD editorial board. Additionally, she is currently the AI section editor and serves on the editorial board for the Journal of Thoracic Imaging.
Dr. Koo’s research focuses on advanced CT and MR imaging of pulmonary malignancies and interstitial lung diseases. Her research has been funded by the Mayo Clinic Career Development Award, foundations such as Society of Thoracic Radiology and American Association for Women Radiologists as well as by industry including Siemens Medical Solutions USA, Inc.
Figure 1 Dr. Chi Wan Koo
Interview
JTD: What drove you to specialize in thoracic radiology?
Dr. Koo: I am deeply fascinated by thoracic radiology because it truly integrates complex anatomy and pathophysiology in diagnosing diseases. Moreover, thoracic radiologists often play an essential role in patient care as certain thoracic conditions are not apparent on physical exam and non-imaging testing, but readily visible on imaging exams. For instance, patients with combined pulmonary fibrosis and emphysema (CPFE) may present with highly disabling symptoms but normal spirometry. However, CPFE pathology is typically readily apparent on computed tomography exams. Aside from conventional imaging exams, quantitative image-based machine learning models and novel theranostics are emerging technologies that hold tremendous potential for thoracic disease diagnosis and management. These emerging technologies make it an even more exciting time to be in thoracic radiology.
JTD: Could you provide an overview of the recent advancements in imaging techniques for interstitial lung diseases and pulmonary malignancies? Were there any examples that impressed you?
Dr. Koo: We have come a long way since the “birth” of artificial intelligence (AI) more than 60 years ago. In just a few short years, AI applications in radiology have “exploded” with radiology being the leading subspecialty in United States Food and Drug Administration (FDA) approved AI algorithms. The number of FDA-cleared AI models for chest is only second to neuroradiology, with many more developing in the research pipeline. One notable recently FDA-cleared software is Imvaria’s Fibresolve, which is the first-ever FDA-authorized diagnostic tool for lung fibrosis with simultaneously adopted current procedural terminology (CPT) billing codes.
In addition, recent advancements in theranostics and imaging techniques (such as photon-counting computed tomography, low-field and hyperpolarized gas magnetic resonance imaging) are just a few novelties representative of the continued growth in the field.
JTD: Over the years, AI has become a hot topic and shows great potential in its application in the medical field. One of your publications last year (1) indicated that the incorporation of machine learning models into clinical radiology workflow can be envisioned and will be especially helpful in settings where expert thoracic radiology is scarce such as in a general radiology practice. How do you perceive the role and future of Artificial Intelligence in clinical radiology to be?
Dr. Koo: I think machine learning can play many important roles in clinical practice, ranging from improving scan techniques for interstitial lung diseases (ILDs), such as driving down the radiation dose for recurring scans through de-noising techniques, to assisting the general radiologist in interpreting the scans to flagging the cases that need the attention of specialists, including both subspecialized thoracic radiologists and pulmonologists.
JTD: Could you share any challenges or setbacks you have been through in your career?
Dr. Koo: I am very thankful that I have been blessed with a relatively smooth career path filled with many supportive friends, colleagues, and mentors. I think as most academicians might have experienced, it is always challenging to juggle research and teaching while providing full-time clinical service without protected time for scholarly activities. However, it is also precisely such a challenge that has taught me over the years to be efficient and flexible.
JTD: Could you share some ongoing projects you are involved in? How will these projects contribute to the future of thoracic radiology?
Dr. Koo: I am currently involved with several machine learning (ML) projects, ranging from a quantitative technique for ILD assessment to denoising for photon-counting computed tomography (CT). Specifically, we are investigating the effects of photon counting (PCD) CT on our quantitative machine learning model for ILD and we are examining our deep learning denoising method for improving image quality utilizing photon counting CT’s ultra-high-resolution mode. Although PCD-CT is currently an emerging technology, I certainly anticipate many current scanners will be replaced by this technology in the next few decades. Given the many improvements that PCD-CT has to offer, we would expect PCD-CT-based ML models to be at least as robust, if not better than conventional CT-based algorithms. However, it is known that scan parameters can affect the performance of CT quantification models, so further studies are needed to delineate the exact effects of PCD-CT on ML ILD evaluation and our study will fulfill such an unmet need. Our denoising method will improve the image quality of photon counting CT when ultra-high-resolution mode is used, thus enabling more accurate characterization of ILD.
JTD: Do you have any advice for doctors/researchers who would like to dive into the field of thoracic radiology?
Dr. Koo: With emerging technologies ranging from photon counting CT to low-field and hyperpolarized MRI, it is an exciting time for thoracic radiology. I would encourage all to be inquisitive and dive into thoracic radiology without reservation and explore all the possibilities that these emerging technologies have to offer—the world is your oyster!
JTD: How has your experience been as an Editorial Board Member of JTD over the past few years?
Dr. Koo: I have enjoyed my journey as a JTD editorial board member thoroughly, ranging from performing editorial work for the journal and collaborating with eminent authors on various articles to playing a leadership role in putting together the JTD special series “Contemporary practice in thoracic neoplasm diagnosis, evaluation and treatment”.
JTD: As the newly appointed Associate Editor-in-Chief of JTD, what are your expectations for JTD?
Dr. Koo: Using a well-respected journal such as New England Journal of Medicine as a model, I hope to assist JTD in becoming the go-to reference for the most current, clinically relevant, and highest quality research in the field of thoracic disease through identifying and publishing cutting edge, impactful research and reviews that will improve health outcomes.
Reference
- Suman G, Koo CW. Recent Advancements in Computed Tomography Assessment of Fibrotic Interstitial Lung Diseases. Journal of Thoracic Imaging. 2023;38(Suppl 1):S7-S18. doi: 10.1097/RTI.0000000000000705.