Objectives: Body tissue composition is a long-known biomarker with high diagnostic and prognostic value not only in cardiovascular, oncological, and orthopedic diseases but also in rehabilitation medicine or drug dosage. In this study, the aim was to develop a fully automated, reproducible, and quantitative 3D volumetry of body tissue composition from standard CT examinations of the abdomen in order to be able to offer such valuable biomarkers as part of routine clinical imaging. Methods: Therefore, an in-house dataset of 40 CTs for training and 10 CTs for testing were fully annotated on every fifth axial slice with five different semantic body regions: abdominal cavity, bones, muscle, subcutaneous tissue, and thoracic cavity. Multi-resolution U-Net 3D neural networks were employed for segmenting these body regions, followed by subclassifying adipose tissue and muscle using known Hounsfield unit limits. Results: The Sørensen Dice scores averaged over all semantic regions was 0.9553 and the intra-class correlation coefficients for subclassified tissues were above 0.99. Conclusions: Our results show that fully automated body composition analysis on routine CT imaging can provide stable biomarkers across the whole abdomen and not just on L3 slices, which is historically the reference location for analyzing body composition in the clinical routine. Key Points: • Our study enables fully automated body composition analysis on routine abdomen CT scans. • The best segmentation models for semantic body region segmentation achieved an averaged Sørensen Dice score of 0.9553. • Subclassified tissue volumes achieved intra-class correlation coefficients over 0.99.

Despite the great media attention for artificial intelligence (AI), for many health care professionals the term and the functioning of AI remain a “black box,” leading to exaggerated expectations on the one hand and unfounded fears on the other. In this review, we provide a conceptual classification and a brief summary of the technical fundamentals of AI. Possible applications are discussed on the basis of a typical work flow in medical imaging, grouped by planning, scanning, interpretation, and reporting. The main limitations of current AI techniques, such as issues with interpretability or the need for large amounts of annotated data, are briefly addressed. Finally, we highlight the possible impact of AI on the nuclear medicine profession, the associated challenges and, last but not least, the opportunities.

The objective of the present study was to evaluate whether integrated ¹⁸F-FDG-PET/MRI can improve the diagnostic workup in patients with cardiac masses. Methods: Twenty patients were prospectively assessed using integrated cardiac ¹⁸F-FDG-PET/MRI: 16 patients with cardiac masses of unknown identity and four patients with cardiac sarcoma after surgical therapy. All scans were performed on an integrated 3-T PET/MR device (mMR Biograph, Siemens Healthcare, Erlangen, Germany). The MR protocol consisted of HASTE, cine and T2-weighted images as well as T1-weighted images before and after injection of gadobutrol. PET data acquisition was performed simultaneously to the MR scan after injection of 199±58 MBq ¹⁸F-FDG. Patients were prepared with a high-fat, low-carbohydrate diet in a period of 24 hours before the examination and 50 IU/kg of unfractionated heparin were administered intravenously 15 minutes prior the ¹⁸F-FDG injection. Results: Cardiac masses were diagnosed as follows: 3 x metastases, 1 x direct tumor infiltration via pulmonary vein, 2 x local relapse of primary sarcoma after surgery, 1 Burkitt’s lymphoma, 2 x scar/patch tissue after surgery of primary sarcoma, 4 x myxoma, 1 x fibroelastoma, 3 x caseous calcification of mitral annulus (CCMA), 3 x thrombus. SUV_max in malignant lesions was significantly higher than in non-malignant cases (13.2±6.2 vs. 2.3±1.2, P = 0.0004). Using a threshold of ≥ 5.2, SUV_max was found to yield 100% sensitivity and 92% specificity for the differentiation between malignant and non-malignant cases. T2-weighted hyperintensity and contrast-enhancement both yielded 100% sensitivity, but a weak specificity of 54% and 46%, respectively. Morphologic tumor features as assessed by cine MRI yielded 86% sensitivity and 92% specificity. Consent reading using all available MR features yielded 100% sensitivity and 92% specificity. A Boolean ‘AND’ combination of SUV_max ≥ 5.2 with consent MRI reading improved sensitivity and specificity to 100%. Conclusion: In selected patients, ¹⁸F-FDG-PET/MRI can improve the noninvasive diagnosis and follow-up of cardiac masses.

Diagnostic and Interventional Radiology

Therapeutic decisions in non-small cell lung cancer (NSCLC) patients depend on the tumor stage. PET/CT with (18)F-FDG is widely accepted as the diagnostic standard of care. The purpose of this study was to compare a dedicated pulmonary (18)F-FDG PET/MR imaging protocol with (18)F-FDG PET/CT for primary and locoregional lymph node staging in NSCLC patients using histopathology as the reference. Twenty-two patients (12 men, 10 women; mean age ± SD, 65.1 ± 9.1 y) with histopathologically confirmed NSCLC underwent (18)F-FDG PET/CT, followed by (18)F-FDG PET/MR imaging, including a dedicated pulmonary MR imaging protocol. T and N staging according to the seventh edition of the American Joint Committee on Cancer staging manual was performed by 2 readers in separate sessions for (18)F-FDG PET/CT and PET/MR imaging, respectively. Results from histopathology were used as the standard of reference. The mean and maximum standardized uptake value (SUV(mean) and SUV(max), respectively) and maximum diameter of the primary tumor was measured and compared in (18)F-FDG PET/CT and PET/MR imaging. PET/MR imaging and (18)F-FDG PET/CT agreed on T stages in 16 of 16 of patients (100%). All patients were correctly staged by (18)F-FDG PET/CT and PET/MR (100%), compared with histopathology. There was no statistically significant difference between (18)F-FDG PET/CT and (18)F-FDG PET/MR imaging for lymph node metastases detection (P = 0.48). For definition of thoracic N stages, PET/MR imaging and (18)F-FDG PET/CT were concordant in 20 of 22 patients (91%). PET/MR imaging determined the N stage correctly in 20 of 22 patients (91%). (18)F-FDG PET/CT determined the N stage correctly in 18 of 22 patients (82%). The mean differences for SUV(mean) and SUV(max) of NSCLC in (18)F-FDG PET/MR imaging and (18)F-FDG PET/CT were 0.21 and -5.06. These differences were not statistically significant (P > 0.05). The SUV(mean) and SUV(max) measurements derived from (18)F-FDG PET/CT and (18)F-FDG PET/MR imaging exhibited a high correlation (R = 0.74 and 0.86, respectively; P < 0.0001). Size measurements showed an excellent correlation between (18)F-FDG PET/MR imaging and (18)F-FDG PET/CT (R = 0.99; P < 0.0001). The lower and upper limits of agreement between (18)F-FDG PET/CT and (18)F-FDG PET/MR imaging using Bland-Altman analysis were -2.34 to 3.89 for SUV(mean), -7.42 to 4.40 for SUV(max), and -0.59 to 0.83 for the tumor size, respectively. (18)F-FDG PET/MR imaging using a dedicated pulmonary MR imaging protocol, compared with (18)F-FDG PET/CT, does not provide advantages in thoracic staging in NSCLC patients.