Attenuation Correction in PET

Accurate attenuation correction (AC) is essential to emission computed tomography such as PET and SPECT, for both quantitative and qualitative interpretation of the results (Chapter 2 of Handbook of Biomedical Image Analysis: Segmentation, Volume I). In PET, for instance, AC factors are most often determined by calculating the pixel-wise ratio of a blank scan acquired before positioning the patient in the gantry of the scanner, and a transmission scan performed with the patient in the gantry. The major drawback of this approach is that statistical noise in the transmission data would propagate to the emission data [119, 120]. Depending on several factors such as body size and composition, transmission scans of 15-30 min are often performed to minimize the propagation of noise to the emission data through AC, at the price of reducing the patient throughput and increasing the errors due to patient motion, causing misalignment between transmission and emission data. Segmented AC methods, which employ image segmentation to partition the transmission images into regions of homogeneous density such as soft tissue, lung, and solid bones whose AC factors are known a priori and can be assigned, are particularly useful in cases where propagation of noise in transmission measurements during AC becomes a significant effect. A number of approaches based on the framework of pixel classification techniques and region-based segmentation approaches have been proposed and examined for segmented AC in PET. For example, Huang et al. [121] proposed a method where the operator manually defines the outlines of the body and the lung on the attenuation images. Known attenuation coefficients are then assigned to these regions and noiseless AC factors are then obtained by forward projecting the attenuation images with assigned attenuation coefficients. This approach has been further extended by a number of investigators by automating the determination of lung and the body regions using image segmentation techniques. For instance, Xu et al. used local thresholding approach to segment attenuation images into air, lung, and soft tissue [122]. Meikle et al. [123] used histogram fitting techniques to assign the attenuation values based on an assumed probability distribution for the lung and soft tissue components. Pa-penfuss et al. [124] used expectation-maximization clustering technique in conjunction with thresholding to produce fuzzy segmentation of attenuation images. Likewise, Bettinardi et al. [125] proposed an adaptive segmentation technique, also based on fuzzy clustering. This method can automatically determine the number of tissue classes in the attenuation images. The method can generally be applied to any region of the body. At least one of the aforementioned methods is currently in routine use by many PET centers worldwide.

0 0

Post a comment