Noise and artifacts may reduce lesion delectability in dual-isotope (Tc-99m/In-111) single photon emission computed tomography (SPECT) studies. The effects of attenuation can lead to artifacts near high-activity structures, e.g., the liver, spleen, kidneys, or urinary bladder. Without effective attenuation correction, these artifacts may reduce diagnostic utility. This study compares three processing, methods: 1) zeroth-order Chang's attenuation correction (ZOCAC) with three-dimensional (3-D) Butterworth filtering; 2) quasi-optimal attenuation correction (QOAC) with 3-D Butterworth filtering; and 3) QOAC with a new adaptive filter. A physical phantom is used to determine effective attenuation coefficients to use with Tc-99m and with In-111. Computer simulation studies are used to optimize the parameters of the filters and to compare the three processing methods using bias versus variance plots. The results indicate that QOAC and the adaptive filter both improve the bias-variance tradeoff. Differences between the ZOCAC and the QOAC reconstructions are seen in clinical studies. QOAC noticeably reduces artifacts which have artificially low activity concentration near high-activity structures, e.g-, near the urinary bladder and great vessels in Tc-99m labeled red blood cell images. QOAC, combined with the new adaptive filter, holds promise for improving Tc-99m/In-111 SPECT reconstructions.
