2021-08-12
▲ A Fully Convolutional Network (FCN) in medical imaging, delivered by Dr Hao Chen
It is an honour and a privilege to be invited to participate in this Symposium, and Imsight Dr Hao Chen delighted to be here this year. I wish to thank the Dept of Pathology from HKU Medicine for this very kind invitation.
It may be one of the most important meetings in the field of Pathology and it has been hosting over 10 years. There is no doubt that the role of digitalization and artificial intelligence has evolved over the years, providing a better resource planning for the sustainable development in the field of disease diagnosis and clinical or pharma research discovery.
Likewise, Dr. Chen Hao was invited to deliver how AI Deep Learning has assisted the development of Anatomical & Cellular Pathology from general computer scientists’ perspective.
Take the Challenges and Tech applied into daily use
“Pathologic imaging requires formidable size with billions of pixels,” said Hao, “Needless to mention the diversity and complexity of biological structure and texture.” However, thanks to the rapid progress of digital pathology, users are easier to manage images and embed computer-aided diagnosis technologies into daily usage.
The rise of Cloud Computing for better-standardized workflows in Cytohistology analysis
Imsight believes the recent rapid adoption of cloud is mainly due to the understanding of “ease of use and scalability” of the technology. Thanks to the expansion of understanding of the enormous benefits of cloud computing, the healthcare industry starts to conduct all kinds of solutions and tasks on the cloud and even migrate entire applications to the cloud.
Dr Hao Chen also pointed out how technologies and AI can streamline the diagnostic process via AI Cloud Platform. “Images scanned using a whole slide scanner will be stored in the AI Cloud platform where AI can assist to deliver efficient diagnosis by possibly unified with 3rd opinions in the process .”
▲ AI Cloud Platform for Histology Analysis
Utilizing FCN an architecture used mainly for semantic segmentation
A fully convolutional network, known as FCN in short, allows greater inputs to boost efficiency by avoiding the use of dense layers, in return for a faster training network. Deep learning holds promises for accurate and efficient Whole Slide Imaging (WSI) analysis such as tumor localization, qualification and classification. Hence, not only it can enhance work efficiency; reduce human errors, but also to maximize patients’ benefits by improving satisfaction rate in general.
Theme: Hong Kong Pathology Forum 2021
Date: 30 Jan, 2021
Agenda: Artificial Intelligence & Computational Pathology
Organizer: Dept. of Pathology, HKU Medicine
Chairperson: Prof Annie NY Cheung (QMH)
Speaker:
- Dr Rex Au-Yeung (QMH)
- Dr Chen Hao (Imsight Technology)
- Dr Kam-Cheong Lee (PMH)
- Dr King-Chung Lee (St.Paul’s Hospital)
▲ HK Pathology Forum 2021
