Correction of oral contrast artifacts in CT-based attenuation correction of PET images using an automated segmentation algorithm

Oral contrast is usually administered in most X-ray computed tomography (CT) examinations of the abdomen and the pelvis as it allows more accurate identification of the bowel and facilitates the interpretation of abdominal and pelvic CT studies. However, the misclassification of contrast medium with high density bone in CT-based attenuation correction (CTAC) is known to generate artifacts in the attenuation map (mumap), resulting in overcorrection for attenuation of PET images. In this paper, we developed an automated segmentation algorithm for classification of regions containing oral contrast medium in order to correct for artifacts in CT attenuation-corrected PET images using the segmented contrast correction (SCC) technique. Our segmentation algorithm consists of two steps: (1) high CT number object segmentation using combined region- and boundary-based segmentation and (2) object classification to bone and contrast agent based on fuzzy classifier as knowledge-based nonlinear classifier. Thereafter, the CT numbers of pixels belonging to the region classified as contrast medium are substituted with their equivalent effective bone CT numbers based on the SCC algorithm. The generated CT images were down-sampled and followed by Gaussian smoothing to match the resolution of PET images. A bi-linear calibration curve was used to convert CT pixel values in HU to mumap at 511 keV. The visual assessment of segmented regions in clinical CT images performed by an experienced radiologist confirmed the accuracy of the segmentation algorithm for delineation of contrast enhanced regions. The mean attenuation coefficient of a small region in the generated mumaps before and after correction using the SCC algorithm was 0.151 and 0.098 cm^-1, respectively. Quantitative analysis of generated mumaps from a clinical dataset showed an overestimation of 19.7% of attenuation coefficients in the 3D regions classified as contrast agent. A clinical PET/CT study known to be problem- atic demonstrated the applicability of the technique. More importantly, correction of oral contrast artefacts improved the readability and interpretation of the PET scan and showed substantial decrease of the SUV (104.3%) after correction. In conclusion, we developed an automated segmentation algorithm for classification of irregular shapes of regions containing contrast medium usually found in clinical CT images for wider applicability of the SCC algorithm for correction of oral contrast artefacts in CTAC. The algorithm is being refined and further validated in clinical setting.