Improved automated tumor segmentation in whole-body 3D scans using multi-directional 2D projection-based priors.

Early cancer detection, guided by whole-body imaging, is important for the overall survival and well-being of the patients. While various computer-assisted systems have been developed to expedite and enhance cancer diagnostics and longitudinal monitoring, the detection and segmentation of tumors, especially from whole-body scans, remain challenging. To address this, we propose a novel end-to-end automated framework that first generates a tumor probability distribution map (TPDM), incorporating prior information about the tumor characteristics (e.g. size, shape, location). Subsequently, the TPDM is integrated with a state-of-the-art 3D segmentation network along with the original PET/CT or PET/MR images. This aims to produce more meaningful tumor segmentation masks compared to using the baseline 3D segmentation network alone. The proposed method was evaluated on three independent cohorts (autoPET, CAR-T, cHL) of images containing different cancer forms, obtained with different imaging modalities, and acquisition parameters and lesions annotated by different experts. The evaluation demonstrated the superiority of our proposed method over the baseline model by significant margins in terms of Dice coefficient, and lesion-wise sensitivity and precision. Many of the extremely small tumor lesions (i.e. the most difficult to segment) were missed by the baseline model but detected by the proposed model without additional false positives, resulting in clinically more relevant assessments. On average, an improvement of 0.0251 (autoPET), 0.144 (CAR-T), and 0.0528 (cHL) in overall Dice was observed. In conclusion, the proposed TPDM-based approach can be integrated with any state-of-the-art 3D UNET with potentially more accurate and robust segmentation results.

Spatial mapping of tumor heterogeneity in whole-body PET-CT: a feasibility study.

BACKGROUND: Tumor heterogeneity is recognized as a predictor of treatment response and patient outcome. Quantification of tumor heterogeneity across all scales may therefore provide critical insight that ultimately improves cancer management. METHODS: An image registration-based framework for the study of tumor heterogeneity in whole-body images was evaluated on a dataset of 490 FDG-PET-CT images of lung cancer, lymphoma, and melanoma patients. Voxel-, lesion- and subject-level features were extracted from the subjects' segmented lesion masks and mapped to female and male template spaces for voxel-wise analysis. Resulting lesion feature maps of the three subsets of cancer patients were studied visually and quantitatively. Lesion volumes and lesion distances in subject spaces were compared with resulting properties in template space. The strength of the association between subject and template space for these properties was evaluated with Pearson's correlation coefficient. RESULTS: Spatial heterogeneity in terms of lesion frequency distribution in the body, metabolic activity, and lesion volume was seen between the three subsets of cancer patients. Lesion feature maps showed anatomical locations with low versus high mean feature value among lesions sampled in space and also highlighted sites with high variation between lesions in each cancer subset. Spatial properties of the lesion masks in subject space correlated strongly with the same properties measured in template space (lesion volume, R = 0.986, p < 0.001; total metabolic volume, R = 0.988, p < 0.001; maximum within-patient lesion distance, R = 0.997, p < 0.001). Lesion volume and total metabolic volume increased on average from subject to template space (lesion volume, 3.1 ± 52 ml; total metabolic volume, 53.9 ± 229 ml). Pair-wise lesion distance decreased on average by 0.1 ± 1.6 cm and maximum within-patient lesion distance increased on average by 0.5 ± 2.1 cm from subject to template space. CONCLUSIONS: Spatial tumor heterogeneity between subsets of interest in cancer cohorts can successfully be explored in whole-body PET-CT images within the proposed framework. Whole-body studies are, however, especially prone to suffer from regional variation in lesion frequency, and thus statistical power, due to the non-uniform distribution of lesions across a large field of view.

An image registration method for voxel-wise analysis of whole-body oncological PET-CT.

Whole-body positron emission tomography-computed tomography (PET-CT) imaging in oncology provides comprehensive information of each patient's disease status. However, image interpretation of volumetric data is a complex and time-consuming task. In this work, an image registration method targeted towards computer-aided voxel-wise analysis of whole-body PET-CT data was developed. The method used both CT images and tissue segmentation masks in parallel to spatially align images step-by-step. To evaluate its performance, a set of baseline PET-CT images of 131 classical Hodgkin lymphoma (cHL) patients and longitudinal image series of 135 head and neck cancer (HNC) patients were registered between and within subjects according to the proposed method. Results showed that major organs and anatomical structures generally were registered correctly. Whole-body inverse consistency vector and intensity magnitude errors were on average less than 5 mm and 45 Hounsfield units respectively in both registration tasks. Image registration was feasible in time and the nearly automatic pipeline enabled efficient image processing. Metabolic tumor volumes of the cHL patients and registration-derived therapy-related tissue volume change of the HNC patients mapped to template spaces confirmed proof-of-concept. In conclusion, the method established a robust point-correspondence and enabled quantitative visualization of group-wise image features on voxel level.

Continuous human uterine NK cell differentiation in response to endometrial regeneration and pregnancy.

Immune cell differentiation is critical for adequate tissue-specific immune responses to occur. Here, we studied differentiation of human uterine natural killer cells (uNK cells). These cells reside in a tissue undergoing constant regeneration and represent the major leukocyte population at the maternal-fetal interface. However, their physiological response during the menstrual cycle and in pregnancy remains elusive. By surface proteome and transcriptome analysis as well as using humanized mice, we identify a differentiation pathway of uNK cells in vitro and in vivo with sequential acquisition of killer cell immunoglobulin-like receptors and CD39. uNK cell differentiation occurred continuously in response to the endometrial regeneration and was driven by interleukin-15. Differentiated uNK cells displayed reduced proliferative capacity and immunomodulatory function including enhanced angiogenic capacity. By studying human uterus transplantation and monozygotic twins, we found that the uNK cell niche could be replenished from circulation and that it was under genetic control. Together, our study uncovers a continuous differentiation pathway of human NK cells in the uterus that is coupled to profound functional changes in response to local tissue regeneration and pregnancy.

Reduced Vglut2/Slc17a6 Gene Expression Levels throughout the Mouse Subthalamic Nucleus Cause Cell Loss and Structural Disorganization Followed by Increased Motor Activity and Decreased Sugar Consumption.

The subthalamic nucleus (STN) plays a central role in motor, cognitive, and affective behavior. Deep brain stimulation (DBS) of the STN is the most common surgical intervention for advanced Parkinson's disease (PD), and STN has lately gained attention as target for DBS in neuropsychiatric disorders, including obsessive compulsive disorder, eating disorders, and addiction. Animal studies using STN-DBS, lesioning, or inactivation of STN neurons have been used extensively alongside clinical studies to unravel the structural organization, circuitry, and function of the STN. Recent studies in rodent STN models have exposed different roles for STN neurons in reward-related functions. We have previously shown that the majority of STN neurons express the vesicular glutamate transporter 2 gene (Vglut2/Slc17a6) and that reduction of Vglut2 mRNA levels within the STN of mice [conditional knockout (cKO)] causes reduced postsynaptic activity and behavioral hyperlocomotion. The cKO mice showed less interest in fatty rewards, which motivated analysis of reward-response. The current results demonstrate decreased sugar consumption and strong rearing behavior, whereas biochemical analyses show altered dopaminergic and peptidergic activity in the striatum. The behavioral alterations were in fact correlated with opposite effects in the dorsal versus the ventral striatum. Significant cell loss and disorganization of the STN structure was identified, which likely accounts for the observed alterations. Rare genetic variants of the human VGLUT2 gene exist, and this study shows that reduced Vglut2/Slc17a6 gene expression levels exclusively within the STN of mice is sufficient to cause strong modifications in both the STN and the mesostriatal dopamine system.