Adaptive region-growing with maximum curvature strategy for tumor segmentation in¹⁸F-FDG PET

Abstract

Accurate tumor segmentation in PET is crucial in many oncology applications. We developed an adaptive region-growing (ARG) algorithm with a maximum curvature strategy (ARG_MC) for tumor segmentation in PET. The ARG_MC repeatedly applied a confidence connected region-growing (CCRG) algorithm with increasing relaxing factor f. The optimal relaxing factor (ORF) was then determined at the transition point on the f-volume curve, where the volume just grew from the tumor into the surrounding normal tissues. The ARG_MC along with five widely used algorithms were tested on a phantom with 6 spheres at different signal to background ratios and on two clinic datasets including 20 patients with esophageal cancer and 11 patients with non-Hodgkin lymphoma (NHL). The ARG_MC did not require any phantom calibration or any a priori knowledge of the tumor or PET scanner. The identified ORF varied with tumor types (mean ORF = 9.61, 3.78 and 2.55 respectively for the phantom, esophageal cancer, and NHL datasets), and varied from one tumor to another. For the phantom, the ARG_MC ranked the second in segmentation accuracy with an average Dice similarity index (DSI) of 0.86, only slightly worse than Daisne’s adaptive thresholding method (DSI=0.87), which required phantom calibration. For both the esophageal cancer dataset and the NHL dataset, the ARG_MC had the highest accuracy with an average DSI of 0.87 and 0.84, respectively. The ARG_MC was robust to parameter settings and region of interest selection, and it did not depend on scanners, imaging protocols, or tumor types. Furthermore, the ARG_MC made no assumption about the tumor size or tumor uptake distribution, making it suitable for segmenting tumors with heterogeneous FDG uptake. In conclusion, the ARG_MC was accurate, robust and easy to use, it provides a highly potential tool for PET tumor segmentation in clinic.

The first MICCAI challenge on PET tumor segmentation

“…ARG is an adaptive region-growing algorithm specially designed for tumor segmentation in PET (Tan et al, 2017). Particularly, the ARG repeatedly applies a confidence connected region-growing (CCRG) algorithm with an increasing relaxing factor f. A maximum curvature strategy is used to automatically identify the optimal value for f as the transition point on the f-volume curve, where the volume just grows from the tumor into the surrounding normal tissues.…”

A phantom study to assess the reproducibility, robustness and accuracy of PET image segmentation methods against statistical fluctuations

“…However, this validation procedure is not suitable across multiple settings. Experimental or simulated phantom studies can overcome these limitations as the phantom can either be scanned or simulated with different settings [24], and thus phantoms are now widely used to validate PET segmentation algorithms [25–28]. A recent study proposes a reconstruction frame-work for simultaneous estimation of the activity distribution, parametric images and segmentation [29].…”

A preliminary PET radiomics study of brain metastases using a fully automatic segmentation method

“…By taking advantage of the great sensitivity and specificity of 11C-MET radio-tracers in discriminating between healthy and tumour tissues, the system identifies the PET slice containing the maximum SUV (SUV max ) in the whole PET dataset avoiding any user intervention. Consequently, the SUV max voxel is used as a target seed for a region growing segmentation [18] to automatically identify a ROI containing the lesion. It is worth noting that the region growing algorithm is used only to obtain a rough estimate of the lesion boundary.…”

“…[ 14 ], where ROIs were manually delineated. For this reason, many PET-based automatic segmentation methods have been proposed [ 15 – 18 ] but no consensus has been reached on the optimal delineation method recommending that no one single method can be used for general BTV delineation [ 19 ]. The large variability in the shape and texture of lesions makes difficult to generalize PET segmentation methods.…”