Paper 13305-3
Neural network-based high-speed amplitude spectral dynamic optical coherence tomography
27 January 2025 • 10:45 AM - 11:00 AM PST | Moscone South, Room 203 (Level 2)
Abstract
Dynamic optical coherence tomography (DOCT) methods have been developed for tissue dynamics imaging. The amplitude-spectral (AS-) DOCT method, which evaluates intracellular motility utilizing the time-order information of many OCT frames for frequency analysis, suffers from slow imaging speed. To address this, a neural network (NN), which can extract time-order information and generate AS-DOCT images from only 16 OCT frames, is proposed. In validations using breast and colon cancer spheroids and zebrafish, the generated AS-OCT images showed high structural similarity to conventional AS-OCT images. The NN-based method significantly reduces the number of required OCT frames (96.9% reduction), enabling volumetric imaging in 26 s.