A New Conversion Method of Pixel Values to Hounsfield Units for Cone Beam CT Images
K TATEOKA*, T Abe, T Nakazawa, Y Saito, A Nakata, M Yano, K Sakata, Sapporo Medical University, Sapporo, HokkaidoSU-E-T-467 Sunday 3:00PM - 6:00PM Room: Exhibit Hall
Purpose: The purpose of this study is to develop a new correction method for converting pixel values to Hounsfield units (HUs) using the histograms of pixel values for the cone beam computed tomography (CBCT) to calculate the dose distributions on the CBCT images.
Methods:The simulation computed tomography (sim-CT) and CBCT images of an electron density phantom and 5 prostate cancer patients were acquired. The histograms of pixel values for each slice of the sim-CT and CBCT images of the electron density phantom and of the prostatic cancer patients were obtained with an in-house program. To perform correction of the pixel values for each slice of the CBCT images, the shapes of the histograms for each slice of the CBCT images were closed to the shapes of the histograms of the sim-CT images using linear scaling method with a minimum sum of squared differences for the regions of soft tissue and bone. To evaluate clinical significance, the dose distribution on the mCBCT images was compared with the original plan on the sim-CT images by using gamma analysis with a 3% / 2 mm criterion.
Results:The linear scaling of pixel values in the CBCT images was successfully applied to the histograms. After the linear scaling, the histograms of mCBCT resembled the histograms of sim-CT. Thus, it is possible to judge pixel values for CBCT images in HUs. However, because the artifacts peculiar to CBCT remained, differences between both HUs were observed The pass rate of gamma analysis (3% / 2 mm) were more than 95%.
Conclusion:The proposed novel method using histograms is a feasible method for converting pixel values in CBCT images to HUs; this may facilitate calculation of accurate doses using CBCT images.