CiNet Monthly Seminar

2019年1月17日（木）
16:00 ~ 17:00
会場 ： CiNet　1F　大会議室

“Patient-specific analysis of musculoskeletal anatomy and dynamics using medical images with deep learning”

奈良先端科学技術大学院大学
情報科学研究科　生体医用画像研究室
准教授
大竹義人

http://isw3.naist.jp/Contents/Research/ai-05-ja.html
https://www.jst.go.jp/kisoken/presto/project/204bigdata/14531241.html

担当 : 平島 雅也（内藤 グループ）

Abstract
This talk will discuss about the patient-specific modeling of musculoskeletal systems using medical images including CT, MRI and X-ray projection images. Understanding of musculoskeletal anatomy and dynamics is essential in diagnosis, treatment planning, and rehabilitation in orthopedic surgery as well as biomechanics modeling in sports science, ergonomic design, computer graphics and so on. We have been developing algorithms to automatically create computational model of individual muscle shapes and its internal muscle fiber arrangement from CT or MRI of the lower extremity using deep learning. In addition, the less invasive medical images such as X-ray projection allow a real-time acquisition which help us understand dynamics of joints and muscles. We combine the 2D projection images with a static 3D image using 2D-3D registration algorithm and analyze movement of bones in 3D with high time resolution. This talk will close with some thoughts on the broader challenges in musculoskeletal analysis.