Spatial and temporal processing for functional imaging probes

Functional imaging probes can help surgeons to accurately locate residual tumors for a complete destruction of malignant tissues with minimal damage to healthy ones that can lead to better patient survival and recovery. In our previous work, we demonstrated that the combination of spatial and temporal processing could yield image frames with a fast update rate and good image quality for superior tumor detection performance. In this work, we further investigate more advanced spatial and temporal processing methods for functional imaging probes. Total variation (TV) based methods are used for spatial denoising and compared with Gaussian smoothing. For temporal processing, the key component of motion estimation is studied using both conventional energy-based and new TV-L1 norm based optical flow methods. Applied on Poisson noise corrupted projection images, TV based spatial denoising methods demonstrate superior performance over Gaussian smoothing, whereas the energy-based motion estimation method seems to work better than TV-L1 norm based method. More thorough investigations are needed to confirm these findings and to obtain the processing strategy for the optimal imaging performance of functional imaging probes.