No abstract

Automatic liver tumor segmentation would have a big impact on liver therapy planning procedures and follow-up assessment, thanks to standardization and incorporation of full volumetric information. In this work, we develop a fully automatic method for liver tumor segmentation in CT images based on a 2D fully convolutional neural network with an object-based postprocessing step. We describe our experiments on the LiTS challenge training data set and evaluate segmentation and detection performance. Our proposed design cascading two models working on voxel- and object-level allowed for a significant reduction of false positive findings by 85% when compared with the raw neural network output. In comparison with the human performance, our approach achieves a similar segmentation quality for detected tumors (mean Dice 0.69 vs. 0.72), but is inferior in the detection performance (recall 63% vs. 92%). Finally, we describe how we participated in the LiTS challenge and achieved state-of-the-art performance.

PurposePET/CT with the PSMA ligand is a powerful new method for the early detection of nodal metastases in patients with biochemical relapse. The purpose of this retrospective investigation was to evaluate the volume and dimensions of nodes identified by Glu-urea-Lys-(Ahx)-[68Ga(HBED-CC)] (68Ga-PSMA-11) in the setting of recurrent prostate cancer.MethodsAll PET/CT images were acquired 60 ± 10 min after intravenous injection of 68Ga-PSMA-11 (mean dose 176 MBq). In 21 patients with recurrent prostate cancer and rising PSA, 49 PSMA-positive lymph nodes were identified. Using semiautomated lymph node segmentation software, node volume and short-axis and long-axis dimensions were measured and compared with the maximum standardized uptake values (SUVmax). Round nodes greater than or equal to 8 mm were considered positive by morphological criteria alone. The percentage of nodes identified by elevated SUVmax but not by conventional morphological criteria was determined.ResultsThe mean volume of 68Ga-PSMA-11-positive nodes was 0.5 ml (range 0.2 – 2.3 ml), and the mean short-axis diameter was 5.8 mm (range 2.4 – 13.3 mm). In 7 patients (33.3 %) with 31 PSMA-positive nodes only 11 (36 %) were morphologically positive based on diameters >8 mm on CT. In the remaining 14 patients (66.7 %), 18 (37 %) of PSMA positive lymph nodes had short-axis diameters <8 mm with a mean short-axis diameter of 5.0 mm (range 2.4 – 7.9 mm). Thus, in this population, 68Ga-PSMA-11 PET/CT detected nodal recurrence in two-thirds of patients who would have been missed using conventional morphological criteria.Conclusion68Ga-PSMA-11 PET/CT is more sensitive than CT based 3D volumetric lymph node evaluation in determining the node status of patients with recurrent prostate cancer, and is a promising method of restaging prostate cancers in this setting.

No abstract

No abstract

In patients with lung cancer (LC), malignant melanoma (MM), gastroenteropancreatic neuroendocrine tumors (GEP NETs), and prostate cancer (PCA), lymph node (LN) staging is often performed by 18 F-FDG PET/CT (LC and MM), 68 Ga-DOTATOC PET/CT (GEP NET), and 68 Ga-labeled prostate-specific membrane antigen PET/CT (PCA) but is sometimes not accurate because of indeterminate PET findings. To better evaluate malignant LN infiltration, additional surrogate parameters, especially in cases with indeterminate PET findings, would be helpful. The purpose of this study was to evaluate whether SUV max in the PET examination might correlate with semiautomated density measurements of LNs in the CT component of the PET/CT examination. Methods: After approval by the institutional review board, 1,022 LNs in the PET/CT examinations of 148 patients were retrospectively analyzed (LC: 327 LNs of 40 patients; MM: 224 LNs of 33 patients; GEP NET: 217 LNs of 35 patients; and PCA: 254 LNs of 40 patients). PET/CT was performed before surgery, biopsy, chemotherapy, or internal or external radiation therapy, according to the clinical schedule; patients with prior chemotherapy or radiation therapy were excluded. SUV max analyses were based on uptake 60 min after tracer injection, and volumetric CT histogram analyses were based on the unenhanced CT images of the PET/CT scan. Results: PET findings were considered positive or negative on the basis of SUV max in the LN compared with that in the blood pool; histologic confirmation was not available. Of the 1,022 LNs, 331 were PET-positive (3 times the SUV max of the blood pool), 86 were PET-indeterminate (1-3 times the SUV max of the blood pool), and 605 were PET-negative (less than the SUV max of the blood pool). PET-positive LNs had significantly higher CT densities than PET-negative LNs, irrespective of the type of cancer. Conclusion: CT density measurements of LNs in patients with LC, MM, GEP NET, and PCA correlated with 18 F-FDG uptake, 68 Ga-DOTATOC uptake, and 68 Ga-PSMA uptake, respectively, and might therefore serve as an additional surrogate parameter for differentiating between malignant and benign LNs. The use of a 7.5-Hounsfield unit CT density threshold to differentiate between malignant and benign LN infiltration and 20 Hounsfield units to exclude benign LN processes might be possible in clinical routine and would be especially helpful for PET-indeterminate LNs. Oneoft he most widespread and reliable noninvasive staging tools for cancer is PET/CT-namely, 18 F-FDG PET/CT for lung cancer (LC (1,2)) and malignant melanoma (MM (1,3)), 68 Ga-labeled prostate-specific membrane antigen (PSMA) PET/ CT for prostate cancer (PCA (4)), and 68 Ga-DOTATOC PET/ CT for gastroenteropancreatic neuroendocrine tumors (GEP NETs (5)). An exact evaluation of lymph node (LN) status before therapy is crucial to therapy planning. False-positive PET findings are not uncommon with 18 F-FDG PET/CT, because the scan can mistakenly pick up on inflammation due to infectious etiologies, such as in the lung or the h...

No abstract

No abstract