Selective Internal Radiation Therapy (SIRT) is a common way to treat liver cancer that cannot be treated surgically. SIRT involves administration of Yttrium – 90 (<sup>90</sup>Y) microspheres via the hepatic artery after a diagnostic procedure using <sup>99m</sup>Technetium (Tc)-macroaggregated albumin (MAA) to detect extrahepatic shunting to the lung or the gastrointestinal tract. Accurate quantification of radionuclide administered to patients and radiation dose absorbed by different organs is of importance in SIRT. Accurate dosimetry for SIRT allows optimization of dose delivery to the target tumor and may allow for the ability to assess the efficacy of the treatment. In this study, we proposed a method that can efficiently estimate radiation absorbed dose from <sup>90</sup>Y bremsstrahlung SPECT/CT images of liver and the surrounding organs. Bremsstrahlung radiation from <sup>90</sup>Y was simulated using the Compton window of <sup>99m</sup>Tc (78keV at 57%). <sup>99m</sup>Tc images acquired at the photopeak energy window were used as a standard to examine the accuracy of dosimetry prediction by the simulated bremsstrahlung images. A Liqui-Phil abdominal phantom with liver, stomach and two tumor inserts was imaged using a Philips SPECT/CT scanner. The Dose Point Kernel convolution method was used to find the radiation absorbed dose at a voxel level for a three dimensional dose distribution. This method will allow for a complete estimate of the distribution of radiation absorbed dose by tumors, liver, stomach and other surrounding organs at the voxel level. The method provides a quantitative predictive method for SIRT treatment outcome and administered dose response for patients who undergo the treatment.