Candidate: Eric-Khang Dao
Date: April 10, 2025
Time: 1:00 pm
Location: Zoom
Supervisor: Dr. Vincent Gaudet
All are welcome!
Abstract:
Position Emission Tomography (PET) is an essential imaging technique used in clinical settings for diagnosing conditions such as cancer and neurological disorders. However, its dependence on radiopharmaceuticals poses potential radiation exposure risks. Lowering the administered dose can help improve patient safety, but it also results in imagery with a reduced Signal-to-Noise (SNR), which can impact diagnostic accuracy. The trade-off between minimizing radiation exposure and maintaining image quality remains a key challenge in PET imaging. Recently, deep learning-based denoising techniques, such as Denoising Convolutional Neural Network (DnCNN), have been proven effective in restoring noisy images to standard quality. However, traditional implementations relying on Central Processing Units (CPU) and Graphics Processing Units (GPU) are often constrained by high power consumption and hardware overhead, limiting their feasibility in edge-compute applications.
To address these challenges, this thesis explores Field-Programmable Gate Array (FPGA)-based acceleration for PET image denoising. A dataset is constructed using PET scans from 10 Alzheimer’s disease patients from the Alzheimer’s Disease Neuroimaging Initiative (ADNI) database, with only 0.5% of the original radiotracer dose used. A software-based implementation is first developed using a proposed U-Net-like architecture. The model is then ported to an FPGA using OpenVINO and Intel’s FPGA AI Suite for hardware emulation. Experimental results show that the FPGA implementation offers a 77% improvement in the performance-to-watt ratio compared to the GPU-based solution, and a 2x reduction in latency compared to the CPU-based solution.