Flowchart of system operation

The flow chart of the software program showing the data flow between the CPU and graphics processing unit (GPU) processes and the timing diagram for scanning and image acquisition as presented in Fig. 2. LabVIEW 2017 software was used to display the cross-section (B-scan) for data acquisition. In the developed SD-OCT system, a frame grabber (PCIe-1433, National Instruments, USA) was used to transmit and receive line-scan camera signals. We also utilized Compute Unified Device Architecture (CUDA) through a 2816-core GPU using multithreading (GeForce GTX 980 Ti, NVIDIA, USA) for fast data processing. To remove background noise and nonlinearity in the raw signal, data processing, including background removal, k-linearization (wavenumber linearization), and fast Fourier transform, were applied to the CUDA sub-processor of the GPU. As the final step in GPU processing, log scaling was applied to the data, and the resulting data was sent back to the CPU thread to display the B-scan image in real-time. The frame rate was 50 frames per second; therefore, it took 20 ms to display a single B-scan image. In the present study, 1000 B-scan images were combined, and it took 20 s to acquire MAP (C-scan) images. All OCT B-scan images were reconstructed to generate cross-section, En-face, and MAP images.

Description of the flow diagram about the systemic operation. a Signal processing in a software part of the OCT b A timing diagram showing the scheduling of each hardware component