Determination of the optimum filter for 99mTc SPECT breast imaging using a wire mesh collimator

Background: The development of wire mesh collimator (WMC), has improved the performance of Single Photon Emission Computed tomography (SPECT) because of its higher sensitivity whilst maintaining the same resolution. Consequently, the WMC allows better detection of early–stage cancer. The purpose of this study was to find an optimal filter for image reconstruction for breast SPECT imaging. Methods: Half–ellipsoidal breast phantom in the prone position was simulated with dual–head SPECT camera by Monte Carlo N–Particle Transport Code, version 5 (MCNP5). Six different filters were compared with 17 different cutoff frequencies (Fc), ranging from 0.2 Nyquist frequency (Nq) to 1.0 Nq, with step 0.05. For Butterworth filter, order was from 3 to 12 with step 1. A total of 255 central slices with different parameters were reconstructed by filtered back projection (FBP) to compare the image performance in terms of contrast, noise level and tumour size. Results: The values of tumour size, contrast and noise level were greatly influenced by different filter types and the value of Fc. Ramp and Butterworth produced the best value of tumour size and contrast, whilst Parzen, Hann and Hamming filters gave smoother images. Overall, results showed that Butterworth filter with the highest mean score. Conclusion: Butterworth filter proved to be the provided the best image quality at a higher sensitivity and was found to be the optimum filter for quantitative analysis.