- This event has passed.
Jaeil Kim – Real-time decision support by light-weighted AI models trained with large-scale data for breast cancer diagnosis using ultrasound (AI PHI Affinity Group)
May 5, 2023 @ 9:00 am - 10:00 amFree
Real-time decision support by light-weighted AI models trained with large-scale data for breast cancer diagnosis using ultrasound
Breast cancer is the most common cancer for women worldwide. Early diagnosis of breast cancer generally increases the chance of successful treatment and effectively reduces mortality.
Ultrasound diagnosis is used for distinguishing benign tumors from malignant tumors, and has the advantages of being non-invasive, free of radiation risk, easy to use, and enabling real-time monitoring. Also, unlike mammography, tumors can be detected using ultrasound in dense breasts. Despite these advantages, ultrasound imaging still has several limitations. Healthcare providers should interpret more than 30,000 frame images per patient to detect suspicious lesions, to determine their degree of malignancy, and to make a decision on biopsy and surgical removal. Since the diagnosis process is subjective and requires a high level of expertise, there is a large difference in diagnostic performance between experts and non-experts. In addition, high false positive and false negative rates cause unnecessary biopsy and surgery.
In this presentation, we will introduce a real-time AI solution for breast ultrasound, CadAI-B, which supports real-time lesion detection, lesion-wise analysis, and differential diagnosis. Using a large-scale data over 500,000 breast US images, we developed AI-based decision support models for breast cancer. In order to achieve the real-time AI processing time, we utilized neural network compression and optimization methods which can reduce inference time and memory usage while maintaining diagnostic accuracy. In particular, our light-weighted AI is suitable for ultrasound imaging, highly mobile and easy to access, because it does not require a server or cloud computing.
Presented by Jaeil Kim
Prof. Kim (Ph.D. KAIST `15) is an associate professor of School of Computer Science and Engineering at KNU, Daegu, South Korea. He served as senior researcher of Samsung Electronics (2014-2016), and he was a postdoctoral fellow at IDEA lab, University of North Carolina at Chapel Hill (2016-2018). His research interests focus on computer-aided diagnosis, prognosis prediction, and shape analysis. He has (co-)authored over 70 peer-reviewed papers in international conferences and journals. In 2021, he founded BeamWorks, Inc., a medical artificial intelligence startup, with Prof. Won Hwa Kim, a radiologist and professor of School of Medicine, KNU, under the mission to develop artificial intelligence technologies sharing medical values with anyone anywhere in the world.
Artificial Intelligence in Cancer Research – AI PHI Affinity Group
(First Friday of each month)
This group was formed to discuss the current trends and applications of artificial intelligence in cancer research and clinical practice. The group brings together AI researchers in a variety of fields (computer science, engineering, nutrition, epidemiology radiology, etc) with clinicians and advocates. Students, trainees and faculty with any or no background in AI are encouraged to attend. The goal is to foster collaborative interactions to solve problems in cancer that were thought to be unsolvable a decade ago before the broad use of deep learning and AI in medicine.