Lymphatic spread determines treatment decisions in prostate cancer (PCa) patients. 68Ga-PSMA-PET/ ct can be performed, although cost remains high and availability is limited. therefore, computed tomography (ct) continues to be the most used modality for pca staging. We assessed if convolutional neural networks (CNNs) can be trained to determine 68Ga-PSMA-PET/CT-lymph node status from CT alone. In 549 patients with 68Ga-PSMA PET/CT imaging, 2616 lymph nodes were segmented. Using PET as a reference standard, three CNNs were trained. Training sets balanced for infiltration status, lymph node location and additionally, masked images, were used for training. cnns were evaluated using a separate test set and performance was compared to radiologists' assessments and random forest classifiers. Heatmaps maps were used to identify the performance determining image regions. The CNNs performed with an Area-Under-the-Curve of 0.95 (status balanced) and 0.86 (location balanced, masked), compared to an AUC of 0.81 of experienced radiologists. Interestingly, CNNs used anatomical surroundings to increase their performance, "learning" the infiltration probabilities of anatomical locations. in conclusion, cnns have the potential to build a well performing ct-based biomarker for lymph node metastases in PCa, with different types of class balancing strongly affecting cnn performance. Prostate cancer (PCa) is the most common malignant cancer in men worldwide, and the second most common cause of cancer related death in men 1. Patients with intermediate or high-risk PCa undergo regular staging examinations in order to determine if the tumor has spread beyond the prostate. As treatment success is highly dependent on the presence of systemic spread 2,3 , staging procedures with high sensitivity and specificity are necessary. Standard of care imaging for PCa staging typically includes contrast-enhanced computed tomography (CT) and Technetium-99m-methylene diphosphonate bone scans 4,5. Despite the continued recommendation of CT in staging, it has been shown that predicting lymph node infiltration (LNI) with CT scans is not very reliable 6,7 , with one study reporting a sensitivity and specificity of only 42% and 82% 8. This low performance is most likely due to the limited morphological criteria used to define a lymph node as positive for infiltration, with size being the most relevant 9. A threshold of 8-10 mm is often used despite the fact that 80% of lymph node metastases are less than 8 mm in the short axis 10. Further criteria, such as status of hilum fat, nodal shape, and enhancement characteristics are used to aid diagnosis, but it remains difficult to exclude LNI in large benign hyperplastic nodes or detect it in small nodes below the size threshold 11. In 2012 imaging agents binding to Prostate Specific Membrane Antigen (PSMA) were introduced, leading to the development of PSMA PET/CT 8. PSMA, an integral membrane glycoprotein expressed 100-1000 fold on membranes of PCa cells compared to prostate cells, has been shown to correlate ...
