Meeting Banner
Abstract #4326

Hand Gesture Control for Interventional MRI

Florian Maier1, Renjie He1, Alexander Brunner2, Kamran Ahrar3, John D. Hazle1, R. Jason Stafford1

1Department of Imaging Physics, The University of Texas M. D. Anderson Cancer Center, Houston, TX, United States; 2Department of Medical Physics in Radiology, German Cancer Research Center (DKFZ), Heidelberg, Germany; 3Department of Interventional Radiology, The University of Texas M. D. Anderson Cancer Center, Houston, TX, United States

During MRI-guided interventions, e.g. percutaneous needle biopsies, interventionalists often need to change slice positions and sequence parameters to hit the planned target position accurately and safely. In this work, we propose a new MR-based hand gesture recognition method to allow for convenient parameter changes by the interventionalist during the procedure. A pulse sequence extension was developed that acquires an additional thick slice at a fixed position. The hand shapes are classified based on their first 16 Fourier descriptors. The results show that the method allows for robust parameter modification in real time.

Abstract PDF Presentation Video

Keywords

accurately acquisition acquisitions actual adaptability aligning allows applying approaches artifacts assess automated automatically available binary biopsies bore brown calculating calculation cancer centroid classification classified close closed closing coefficients coil complex connected control convenient coordinates created descriptor descriptors detect detection developed device diagram directly discarded discrete display distance distinguishable domain edge extended extension facilitate field fingers fist fixed flow formed frequency full function future gaps gesture gestures gray green guided hand holes immediately interrupting interventions investigation john kernel magnitude maier maintain make matrix measured morphological motion need needle normalized novel object objects often oneself operator orientation outer parallel patient percutaneous physics pixel pixels planned position positions post prevented procedure processing proposed pulse radiology radius real recognition reconstruction rejected resolution robust robustly room rotation scaling scanner scenarios selected sent shape shapes slice slices sorted spread steps straight successful system target template temporal thick threshold tracking transform update utilizes various view volume warranted wide