How can we quantify, explain, and apply the uncertainty of complex soil maps predicted with neural networks?

Artificial neural networks (ANNs) have proven to be a useful tool for complex questions that involve large amounts of data. Our use case of predicting soil maps with ANNs is in high demand by government agencies, construction companies, or farmers, given cost and time intensive field work. However, there are two main challenges when applying ANNs. In their most common form, deep learning algorithms do not provide interpretable predictive uncertainty. This means that properties of an ANN such as the certainty and plausibility of the predicted variables, rely on the interpretation by experts rather than being quantified by evaluation metrics validating the ANNs. Further, these algorithms have shown a high confidence in their predictions in areas geographically distant from the training area or areas sparsely covered by training data. To tackle these challenges, we use the Bayesian deep learning approach "last-layer Laplace approximation", which is specifically designed to quantify uncertainty into deep networks, in our explorative study on soil classification. It corrects the overconfident areas without reducing the accuracy of the predictions, giving us a more realistic uncertainty expression of the model's prediction. In our study area in southern Germany, we subdivide the soils into soil regions and as a test case we explicitly exclude two soil regions in the training area but include these regions in the prediction. Our results emphasize the need for uncertainty measurement to obtain more reliable and interpretable results of ANNs, especially for regions far away from the training area. Moreover, the knowledge gained from this research addresses the problem of overconfidence of ANNs and provides valuable information on the predictability of soil types and the identification of knowledge gaps. By analyzing regions where the model has limited data support and, consequently, high uncertainty, stakeholders can recognize the areas that require more data collection efforts.

NMDA Receptor-Mediated Signaling Pathways Enhance Radiation Resistance, Survival and Migration in Glioblastoma Cells-A Potential Target for Adjuvant Radiotherapy.

Glioblastoma is one of the most aggressive malignant brain tumors, with a survival time less than 15 months and characterized by a high radioresistance and the property of infiltrating the brain. Recent data indicate that the malignancy of glioblastomas depends on glutamatergic signaling via ionotropic glutamate receptors. In this study we revealed functional expression of Ca2+-permeable NMDARs in three glioblastoma cell lines. Therefore, we investigated the impact of this receptor on cell survival, migration and DNA double-strand break (DSB) repair in the presence of both, glutamate and NMDAR antagonists, and after clinically relevant doses of ionizing radiation. Our results indicate that treatment with NMDAR antagonists slowed the growth and migration of glutamate-releasing LN229 cells, suggesting that activation of NMDARs facilitate tumor expansion. Furthermore, we found that DSB-repair upon radiation was more effective in the presence of glutamate. In contrast, antagonizing the NMDAR or the Ca2+-dependent transcription factor CREB impaired DSB-repair similarly and resulted in a radiosensitizing effect in LN229 and U-87MG cells, indicating a common link between NMDAR signaling and CREB activity in glioblastoma. Since the FDA-approved NMDAR antagonists memantine and ifenprodil showed differential radiosensitizing effects, these compounds may constitute novel optimizations for therapeutic interventions in glioblastoma.

NMDA Receptor Signaling Mediates cFos Expression via Top2ß-Induced DSBs in Glioblastoma Cells.

The activation of Ca2+-permeable N-methyl-D-aspartic acid (NMDA) receptor channels (NMDARs) is crucial for the development and survival of neurons, but many cancers use NMDAR-mediated signaling as well, enhancing the growth and invasiveness of tumors. Thus, NMDAR-dependent pathways emerge as a promising target in cancer therapy. Here, we use the LN229 and U-87MG glioblastoma multiforme (GBM) cells and immunofluorescence staining of 53BP1 to analyze NMDAR-induced DNA double-strand breaks (DSBs), which represent an important step in the NMDAR signaling pathway in neurons by facilitating the expression of early response genes. Our results show that NMDAR activation leads to the induction of DSBs in a subpopulation of glioma cells. In a further analogy to neurons, our results demonstrate that the induction of DSBs in LN229 cells is dependent on the activity of topoisomerase IIß (Top2ß). Western blot analysis revealed that the inhibition of NMDARs, cAMP-responsive element binding transcription factor (CREB) and Top2ß decreased the expression of the proto-oncogene cFos. Knockdown of Top2ß with siRNAs resulted in a downregulation of cFos and increased the radiosensitivity of LN229 cells in clonogenic survival. We also observed impaired cFos expression upon NMDAR and Top2ß inhibition in a primary GBM cell line, suggesting that NMDAR signaling may be widely used by GBMs, demonstrating the potential of targeting NMDAR signaling proteins for GBM therapy.