A Pilot Study of Automatic Lung Tumor Segmentation from Positron Emission Tomography Images using Standard Uptake Values

Abstract

Positron Emission Tomography (PET) is a medical imaging procedure that shows the physiological function of an organ or tissue. The role of PET during the past decade has evolved rapidly in the detection of lung tumors but the research on quantitative evaluation of PET images is still in its infancy. PET commonly involves scanning the patient after administration of a radioactive analogue of glucose called fluoro deoxy-glucose (FDG). Tumor cells metabolise more glucose than most normal cells. In PET lung images the heart is often visible and because of its constant pumping of blood it requires more glucose and hence both the tumor and the heart appear brighter than the rest in the PET image. In this paper we present a novel segmentation scheme for detecting the tumor alone in lung PET images using standard uptake values (SUV) and connected component analysis. We perform the segmentation in two steps. In coarse segmentation, a non linear scaling of SUV values is performed and then a threshold is chosen adaptively to convert the gray image into the binary image. Fine segmentation is performed on the coarse segmented data in order to narrow down the region of interest using connected component labeling. To our knowledge no one has used connected component analysis for segmenting PET images. We compare our proposed scheme with several commonly used medical image segmentation techniques like threshold, sobel edge detector, laplacian of gaussian (LoG) edge detector, region growing and SUV based segmentation (applied only to PET as SUV is specific to PET). One of the problems in lung tumor detection is the presence of the heart in the image which accumulates activity and often gets recognized as a hot spot (a probable tumor). All the other segmentation schemes detected both the heart and the tumor as hot spots while our segmentation scheme detected the tumor alone as the hot spot. The preliminary study of the proposed scheme has yielded very promising results and will be studied for more lung tumor detection scenarios in future.

Automatic lung tumor segmentation with leaks removal in follow-up CT studies

“…We choose a parabolic surface to remove the leak since it is similar to the way a human delineator would complete the missing boundary segment. We empirically found that this is better than using the shape prior defined by the baseline tumor delineation, e.g., as described in [15].…”

Pulmonary Tumor Volume Detection from Positron Emission Tomography Images

“…This is due to the presence of heart as a hot spot in the lung region. A detailed comparison of lung tumor segmentation using standard schemes like threshold, region growing, SUV based segmentation was presented in our earlier work [9]. Although the proposed method was shown to work better than the earlier methods [9], it failed in segmenting tumor alone in cases when the activity of the heart is greater than tumor.…”

Multimodal Medical Image Registration and Fusion in 3D Conformal Radiotherapy Treatment Planning

“…According to Aparna(Aparna Kanakatte et al, 2007), the higher the SUV is, the more aggressive the tumor is. The SUV is also used to distinguish the malignant tumor and benign tumor.…”

“…For the calculation of the fusion of PET and CT images, the standard uptake value(SUV) is frequently used for fluorodeoxyglucose(FDG) PET image to evaluate its uptake value quantitatively(S-C. Huang, 2000) (Aparna Kanakatte et al, 2007). In general, if there exists a tumor it will appear brighter than healthy cells in a PET image.…”