Federated Learning in Dentistry: Chances and Challenges.

Building performant and robust artificial intelligence (AI)-based applications for dentistry requires large and high-quality data sets, which usually reside in distributed data silos from multiple sources (e.g., different clinical institutes). Collaborative efforts are limited as privacy constraints forbid direct sharing across the borders of these data silos. Federated learning is a scalable and privacy-preserving framework for collaborative training of AI models without data sharing, where instead the knowledge is exchanged in form of wisdom learned from the data. This article aims at introducing the established concept of federated learning together with chances and challenges to foster collaboration on AI-based applications within the dental research community.

Local cerebral glucose utilization in fetal guinea pigs at 0.75 gestation.

OBJECTIVE: Using the 2-deoxyglucose method, measurements of local cerebral glucose utilization in large fetal animals are very difficult and expensive. To circumvent these problems we recently modified the 2-deoxyglucose method for use in the fetal guinea pig in utero (Berger et al., J Neurochem 1994; 63: 271-279). The present study was designed to measure the rates of local cerebral glucose utilization in fetal guinea pigs at 0.75 of gestation. STUDY DESIGN: After intravenous injection of 14C 2-deoxyglucose into the dams, local cerebral glucose utilization of the fetuses was measured from the time integral of the tracer in the maternal plasma and the autoradiographically determined concentration of the tracer in various parts of the fetal brain. RESULTS: Fetal cerebral glucose utilization was low as compared to adult animals and varied in different brain structures from 19 +/- 4 to 29 +/- 7 mumol/100 g/min. CONCLUSION: This study demonstrates the feasibility to measure local cerebral glucose utilization in undisturbed fetal guinea pigs in utero. We conclude that the low rate of cerebral glucose utilization and its small overall variability may reflect the neurological immaturity of the fetal brain.

Activated astrocytes, but not pyramidal cells, increase glucose utilization in rat hippocampal CA1 subfield after ischemia.

The local cerebral glucose utilization (CMRglc) in the damaged rat hippocampal CA1 subfield increases 7 days after 10 min of cerebral ischemia. We have used the N-methyl-D-aspartate antagonist (NMDA antagonist) ketamine in rats 7 days after sham operation or cerebral ischemia to determine whether the elevated postischemic CMRglc of the CA1 subfield is due to long-lasting hyperexcitation of surviving or injured neurons, or, alternatively, to the metabolism of other cell types. The autoradiographic data were interpreted with the aid of histochemical analysis of the postischemic hippocampal cell changes. Anesthetic doses of ketamine significantly reduced the CMRglc in the CA1 strata oriens, pyramidale and radiatum of sham-operated rats, while the postischemic increases in CMRglc in these hippocampal CA1 strata were not affected by ketamine. In addition, there were ketamine-induced increases in the CMRglc of the CA1 stratum lacunosum moleculare of both sham-operated and postischemic rats. The immunoreactivity of the microtubule-associated protein 2 (MAP2), a postsynaptic protein marker, was decreased markedly in the CA1 subfield in postischemic rats, while the presynaptic protein marker, synaptophysin, remained the same in sham-operated and postischemic rats. The glial fibrillary acidic protein (GFAP) immunoreactivity of astrocytes raised markedly in the ischemically damaged CA1 subfield. Although it could be demonstrated that presynaptic terminals remain intact in the postischemic damaged CA1 subfield, the lacking ketamine effect on CA1 pyramidal neurons indicated that the increase in CMRglc in this brain area is not due to postsynaptic neural hyperexcitation, but probably has to be attributed to astrocytes activated by neuronal damage.

Vinpocetine increases the neuroprotective effect of adenosine in vitro.

Cultured neurons from chick embryo cerebral hemispheres were used to study the neuroprotective activity of vinpocetine and adenosine against hypoxic damage. Cytotoxic hypoxia was induced by adding 1 mM sodium cyanide to the nutrient medium for 30 min. The drugs were present in the nutrient medium 30 min before, and up to 1 day after hypoxia. To characterize the neuroprotective drug effects, the protein content per culture flask and the neuronal viability were determined 3 days after this hypoxic period. Adenosine, but not vinpocetine, was able to protect neurons against hypoxic damage, when added as single drug to the cultures. However, vinpocetine significantly enhanced the neuroprotective effect of adenosine. The results suggest that the neuroprotective effect of vinpocetine is mediated by adenosine.

Protective effect of vinpocetine against brain damage caused by ischemia.

The effects of vinpocetine against hippocampal neuronal damage and on local cerebral blood flow (LCBF) were examined in a rat model of forebrain ischemia (10-min occlusion of the carotid arteries and hypotension). Histological evaluation of neuronal loss in the hippocampus was performed 7 days after ischemia. LCBF was measured before ischemia as well as after 2 min and 1 hr of recirculation. Vinpocetine (10 mg/kg) administered pre- or post-ischemically reduced the hippocampal neuronal necrosis, while pre-ischemic administration of 2 or 20 mg/kg vinpocetine was ineffective. Since vinpocetine increased the LCBF after 1 hr of recirculation, it cannot be excluded that blood flow improvements contribute to its neuroprotective activity. On the other hand, there is no clear evidence that an elevation of post-ischemic hypoperfusion could protect neurons against ischemic damage. It is, therefore, suggested that vinpocetine acts directly on brain cells.

Preischemic administration of flunarizine or phencyclidine reduces local cerebral glucose utilization in rat hippocampus seven days after ischemia.

The purpose of the present study was to investigate the influence of ischemia on postischemic metabolic activity of the brain. Furthermore, the effect of preischemic application of neuroprotective agents such as flunarizine or phencyclidine on postischemic local cerebral glucose utilization (LCGU) was examined. Forebrain ischemia in the rat was performed for 10 min with bilateral carotid clamping, administration of trimethaphan and blood withdrawal to obtain a mean arterial blood pressure of 40 mm Hg. LCGU was determined 7 days after ischemia by injecting 14C-deoxy-D-glucose in saline solution. A significant increase in LCGU in the CA1 subfield of the hippocampus was found 7 days after ischemia, whereas preischemic administration of flunarizine or phencyclidine inhibited this increase. Alterations in LCGU of other brain regions were insignificant.

Postischemic neuronal damage causes astroglial activation and increase in local cerebral glucose utilization of rat hippocampus.

The purpose of the present study was to determine the consequences of postischemic neuronal damage on CMRglc. Forebrain ischemia of 10 min duration was induced in male Wistar rats. The extent of neuronal damage and the numbers of immunocytochemically detected astrocytes in the hippocampal CA1 subfield as well as CMRglc were determined 2, 5, 7, and 14 days after ischemia. CBF was additionally measured 7 days postischemia. CMRglc was decreased in cortical and thalamic structures up to 5 days postischemia, and was normalized again on day 7 after ischemia. In the hippocampal areas, CMRglc was decreased only on day 2 after ischemia, was normalized after 5 days, and increased in the stratum oriens and pyramidale of the CA1 subfield from postischemic day 7 onward. Neuronal damage was clearly demonstrable 5 days after ischemia and further increased up to day 7. The number of GFAP-reactive astrocytes increased markedly at day 7 postischemia. It is assumed that the activation of astrocytes is induced by neuronal damage, and that the astroglial metabolism is responsible for the increase in CMRglc of the CA1 subfield 7 days after ischemia. The decrease in CBF of the CA1 subfield 7 days after ischemia could be caused by a reduced density of perfused capillaries.

Effects of vinpocetine on local cerebral blood flow and glucose utilization seven days after forebrain ischemia in the rat.

A marked neuronal cell loss has been determined in the hippocampal CA1-subfield of the rat 7 days after 10 min of ischemia. The purpose of the present study was to evaluate the consequences of ischemia-induced hippocampal neuronal damage on local cerebral glucose utilization (LCGU) and blood flow (LCBF). Forebrain ischemia of 10 min duration was induced in male Wistar rats. Seven days after ischemia LCGU was measured with the [14C]-2-deoxy-D-glucose method, and LCBF was determined with the [14C]-iodoantipyrine technique in sham-operated and in ischemic rats. Furthermore, postischemic LCGU and LCBF were determined in vinpocetine-treated ischemic rats. Vinpocetin (14-ethoxycarbonyl-(3 alpha, 16 alpha-ethyl)-14,15-eburnamine) has already proved to protect hippocampal neurons against ischemic damage. In comparison with sham-operated rats, LCGU increased and LCBF decreased significantly in the CA1-subfield 7 days after ischemia. Both effects were abolished by preischemic vinpocetine treatment supporting the presumption that this drug is protective against ischemic damage.

Vinpocetine prevents ischemic cell damage in rat hippocampus.

The effects of vinpocetine on hippocampal cell damage and local cerebral blood flow (LCBF) were measured in a rat model of forebrain ischemia (2-vessel occlusion and hypotension). Duration of ischemia was 10 min. LCBF was determined after 2 min of recirculation using the 14C-iodoantipyrine technique. Hippocampal cell loss was quantified histologically 7 days post-ischemia. Intraperitoneal application of vinpocetine (10 mg/kg) 15 min prior to ischemia significantly reduced neuronal cell loss in hippocampal CA 1 sector from 60% to 28%. The drug led to a marked increase in blood flow in cortical areas, whereas LCBF remained unchanged in hippocampus and all other structures measured. It is suggested that the protective effect of vinpocetine does not depend on increased postischemic blood flow.