To present a deep learning segmentation model that can automatically and robustly segment all major anatomic structures on body CT images. In this retrospective study, 1204 CT examinations (from 2012, 2016, and 2020) were used to segment 104 anatomic structures (27 organs, 59 bones, 10 muscles, and eight vessels) relevant for use cases such as organ volumetry, disease characterization, and surgical or radiation therapy planning. The CT images were randomly sampled from routine clinical studies and thus represent a real-world dataset (different ages, abnormalities, scanners, body parts, sequences, and sites). The authors trained an nnU-Net segmentation algorithm on this dataset and calculated Dice similarity coefficients to evaluate the model’s performance. The trained algorithm was applied to a second dataset of 4004 whole-body CT examinations to investigate age-dependent volume and attenuation changes. The proposed model showed a high Dice score (0.943) on the test set, which included a wide range of clinical data with major abnormalities. The model significantly outperformed another publicly available segmentation model on a separate dataset (Dice score, 0.932 vs 0.871; P < .001). The aging study demonstrated significant correlations between age and volume and mean attenuation for a variety of organ groups (eg, age and aortic volume [r_s = 0.64; P < .001]; age and mean attenuation of the autochthonous dorsal musculature [r_s = −0.74; P < .001]). The developed model enables robust and accurate segmentation of 104 anatomic structures. The annotated dataset ( https://doi.org/10.5281/zenodo.6802613 ) and toolkit ( https://www.github.com/wasserth/TotalSegmentator ) are publicly available. Keywords: CT, Segmentation, Neural Networks Supplemental material is available for this article. © RSNA, 2023 See also commentary by Sebro and Mongan in this issue.

To compare the performance of magnetic resonance (MR)/positron emission tomography (PET) imaging in the staging of lung cancer with that of PET/computed tomography (CT) as the reference standard and to compare the quantification accuracy of a new whole-body MR/PET system with corresponding PET/CT data sets. Institutional review board approval and informed consent were obtained. Ten patients in whom bronchial carcinoma was proven or clinically suspected underwent clinically indicated fluorine 18 fluorodeoxyglucose (FDG) PET/CT and, immediately thereafter, whole-body MR/PET imaging with a new hybrid whole-body system (3.0-T MR imager with integrated PET system). Attenuation correction of MR/PET images was segmentation based with fat-water separation. Tumor-to-liver ratios were calculated and compared between PET/CT and MR/PET imaging. Tumor staging on the basis of the PET/CT and MR/PET studies was performed by two readers. Spearman rank correlation was used for comparison of data. MR/PET imaging provided diagnostic image quality in all patients, with good tumor delineation. Most lesions (nine of 10) showed pronounced FDG uptake. One lesion was morphologically suspicious for malignancy at CT and MR imaging but showed no FDG uptake. MR/PET imaging had higher mean tumor-to-liver ratios than did PET/CT (4.4 ± 2.0 [standard deviation] for PET/CT vs 8.0 ± 3.9 for MR/PET imaging). Significant correlation regarding the tumor-to-liver ratio was found between both imaging units (ρ = 0.93; P < .001). Identical TNM scores based on MR/PET and PET/CT data were found in seven of 10 patients. Differences in T and/or N staging occurred mainly owing to modality-inherent differences in lesion size measurement. MR/PET imaging of the lung is feasible and provides diagnostic image quality in the assessment of pulmonary masses. Similar lesion characterization and tumor stage were found in comparing PET/CT and MR/PET images in most patients.

Purpose To present a method that automatically segments and quantifies abnormal CT patterns commonly present in coronavirus disease 2019 (COVID-19), namely ground glass opacities and consolidations...

“Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review ...

In the adult kidney, renin-producing cells are typically located in the walls of afferent arterioles at the transition into the glomerular capillary network. The mechanisms that are responsible for restricting renin expression to the juxtaglomerular position are largely unknown. This study showed that in mice that lack connexin 40 (Cx40), the predominant connexin of renin-producing cells, renin-positive cells are absent in the vessel walls and instead are found in cells of the extraglomerular mesangium, glomerular tuft, and periglomerular interstitium. Blocking macula densa transport function by acute administration of loop diuretics strongly enhances renin secretion in vivo and in isolated perfused kidneys of wild-type mice. This effect of loop diuretics is markedly attenuated in vivo and even blunted in vitro in Cx40-deficient mice. Even after prolonged stimulation of renin secretion by severe sodium depletion, renin expression is not seen in juxtaglomerular cells or in cells of more proximal parts of the arterial vessel wall as occurs normally. Instead, renin remains restricted to the extra-/periglomerular interstitium in Cx40-deficient mice. In contrast to the striking displacement of renin-expressing cells in the adult kidney, renin expression in the vessels of the developing kidney was found to be normal. This is the first evidence to indicate that cell-to-cell communication via gap junctions is essential for the correct juxtaglomerular positioning and recruitment of renin-producing cells. Moreover, these findings support the notion that gap junctions are relevant for the macula densa signaling to renin-producing cells.