Whole‐body tumor segmentation from PET/CT images using a two‐stage cascaded neural network with camouflaged object detection mechanisms

Abstract

BackgroundWhole‐body Metabolic Tumor Volume (MTVwb) is an independent prognostic factor for overall survival in lung cancer patients. Automatic segmentation methods have been proposed for MTV calculation. Nevertheless, most of existing methods for patients with lung cancer only segment tumors in the thoracic region.PurposeIn this paper, we present a Two‐Stage cascaded neural network integrated with Camouflaged Object Detection mEchanisms (TS‐Code‐Net) for automatic segmenting tumors from whole‐body PET/CT images.MethodsFirstly, tumors are detected from the Maximum Intensity Projection (MIP) images of PET/CT scans, and tumors' approximate localizations along z‐axis are identified. Secondly, the segmentations are performed on PET/CT slices that contain tumors identified by the first step. Camouflaged object detection mechanisms are utilized to distinguish the tumors from their surrounding regions that have similar Standard Uptake Values (SUV) and texture appearance. Finally, the TS‐Code‐Net is trained by minimizing the total loss that incorporates the segmentation accuracy loss and the class imbalance loss.ResultsThe performance of the TS‐Code‐Net is tested on a whole‐body PET/CT image data‐set including 480 Non‐Small Cell Lung Cancer (NSCLC) patients with five‐fold cross‐validation using image segmentation metrics. Our method achieves 0.70, 0.76, and 0.70, for Dice, Sensitivity and Precision, respectively, which demonstrates the superiority of the TS‐Code‐Net over several existing methods related to metastatic lung cancer segmentation from whole‐body PET/CT images.ConclusionsThe proposed TS‐Code‐Net is effective for whole‐body tumor segmentation of PET/CT images. Codes for TS‐Code‐Net are available at: https://github.com/zyj19/TS‐Code‐Net.

How Does Target Lesion Selection Affect RECIST? A Computer Simulation Study

“…Volumetric measures have been described as a better discriminator of tumor size changes than RECIST's linear measurements, 31 related to the capacity of tumor volumetry better describing the size of irregular lesions 32,33 and to the reduced interobserver variability. [34][35][36] Although manual tumor segmentation is very labor-demanding, advances in artificial intelligence for automatic volumetric tumor segmentation of whole-body scans [37][38][39] might soon be integrated in clinical practice. Such a development could change the landscape of response assessment 2 by replacing target lesion selection with TTB estimation and rethinking the coarse threshold-based compartmentalization of response.…”

Development and Validation of a Predictive Model for Metastatic Melanoma Patients Treated with Pembrolizumab Based on Automated Analysis of Whole-Body [18F]FDG PET/CT Imaging and Clinical Features

“…It may also allow the exploration of a wider range of image-derived parameters during clinical research studies. Recently, a large and growing body of literature has investigated lesion segmentation for PET/CT [ 21 , 22 , 23 , 24 , 25 , 26 , 27 ]. However, to the best of our knowledge, the effect of adopting such an automated approach in the pipeline for melanoma prognosis prediction remains to be investigated.…”