Abstract
Laser speckle based superficial and deep tissue blood flow imaging is gaining interest with the advent of high speed cameras. Multi-exposure speckle intensity images are often utilized for this purpose, owing to the better quantification of flow. However, any uncertainty in selecting the required exposure range apriori and the data acquisition time associated with multi-exposure intensity measurements limit the temporal resolution of these systems. To address these concerns, we propose a deep learning-based imputation using Generative Adversarial Imputation Network (GAIN) to generate additional temporal samples from coarsely acquired multi-exposure speckle data. The feasibility of the proposed method has been verified by using simulations where the trade-off between temporal resolution and the accuracy of flow measurement is minimized.
© 2023 SPIE
PDF ArticleMore Like This
Faruk Beslija, Lisa Kobayashi Frisk, Manish Verma, Chen-Hao P. Lin, Nishigandha Patil, Sumana Chetia, Jason W. Trobaugh, Joseph P. Culver, and Turgut Durduran
126280X European Conference on Biomedical Optics (ECBO) 2023
Soumyajit Sarkar, Murali K, and Hari M Varma
126281J European Conference on Biomedical Optics (ECBO) 2023
Soumyajit Sarkar, Mohammad Zaffar, and Hari M Varma
1262827 European Conference on Biomedical Optics (ECBO) 2023