Lung Pattern Classification for Interstitial Lung Diseases Using a Deep Convolutional Neural Network

Automated tissue characterization is one of the most crucial components of a computer aided diagnosis (CAD) system for interstitial lung diseases (ILDs). Although much research has been conducted in this field, the problem remains challenging. Deep learning techniques have recently achieved impressive results in a variety of computer vision problems, raising expectations that they might be applied in other domains, such as medical image analysis. In this paper, we propose and evaluate a convolutional neural network (CNN), designed for the classification of ILD patterns. The proposed network consists of 5 convolutional layers with 2×2 kernels and LeakyReLU activations, followed by average pooling with size equal to the size of the final feature maps and three dense layers. The last dense layer has 7 outputs, equivalent to the classes considered: healthy, ground glass opacity (GGO), micronodules, consolidation, reticulation, honeycombing and a combination of GGO/reticulation. To train and evaluate the CNN, we used a dataset of 14696 image patches, derived by 120 CT scans from different scanners and hospitals. To the best of our knowledge, this is the first deep CNN designed for the specific problem. A comparative analysis proved the effectiveness of the proposed CNN against previous methods in a challenging dataset. The classification performance (~85.5%) demonstrated the potential of CNNs in analyzing lung patterns. Future work includes, extending the CNN to three-dimensional data provided by CT volume scans and integrating the proposed method into a CAD system that aims to provide differential diagnosis for ILDs as a supportive tool for radiologists.

National competence center for wildlife diseases in Switzerland: Mandate, development and current strategies.

The need for wildlife health surveillance has become increasingly recognized. However, comprehensive programs which cover a wide spectrum of species, pathogens and geographic areas are still lacking in most European countries and practical examples of systems in place remain scarce. This article provides an overview of the organization of wildlife health surveillance in Switzerland, with a focus on the development, current strategies and the activities of the national program carried out by the Centre for Fish and Wildlife Health (FIWI), University of Bern. This documentation may stimulate on-going discussions on the design and development of national wildlife health surveillance programs in other countries. Investigations into wildlife health in Switzerland date back to the 1950s. The FIWI acts as a national competence center for wildlife diseases on mandate of the Swiss federal authorities. The mandate includes four main activities: disease diagnostics, research, consulting and teaching. In line with this, the FIWI has made continuous efforts to strengthen a national network of field partners and implemented strategies to facilitate long-term and metastudies.

Use of a modified Delphi panel to identify and weight criteria for prioritization of zoonotic diseases in Switzerland

Zoonotic diseases have a significant impact on public health globally. To prevent or reduce future zoonotic outbreaks, there is a constant need to invest in research and surveillance programs while updating risk management strategies. However, given the limited resources available, disease prioritization based on the need for their control and surveillance is important. This study was performed to identify and weight disease criteria for the prioritization of zoonotic diseases in Switzerland using a semi-quantitative research method based on expert opinion. Twenty-eight criteria relevant for disease control and surveillance, classified under five domains, were selected following a thorough literature review, and these were evaluated and weighted by seven experts from the Swiss Federal Veterinary Office using a modified Delphi panel. The median scores assigned to each criterion were then used to rank 16 notifiable and/or emerging zoonoses in Switzerland. The experts weighted the majority of the criteria similarly, and the top three criteria were Severity of disease in humans, incidence and prevalence of the disease in humans and treatment in humans. Based on these weightings, the three highest ranked diseases were Avian Influenza, Bovine Spongiform Encephalitis, and Bovine Tuberculosis. Overall, this study provided a preliminary list of criteria relevant for disease prioritization in Switzerland. These were further evaluated in a companion study which involved a quantitative prioritization method and multiple stakeholders.

DNA-based analysis of protein variants reveals different genetic variability of the paralogous equine ß-lactoglobulin genes LGB1 and LGB2

The genetic variability of milk protein genes may influence the nutritive value or processing and functional properties of the milk. While numerous protein variants are known in ruminants, knowledge about milk protein variability in horses is still limited. Mare's milk is, however, produced for human consumption in many countries. Beta-lactoglobulin belonging to the protein family of lipocalins, which are known as common food- and airborne allergens, is a major whey protein. It is absent from human milk and thus a key agent in provoking cow's milk protein allergy. Mare's milk is, however, usually better tolerated by most affected people. Several functions of β-lactoglobulin have been discussed, but its ultimate physiological role remains unclear. In the current study, the open reading frames of the two equine β-lactoglobulin paralogues LGB1 and LGB2 were re-sequenced in 249 horses belonging to 14 different breeds in order to predict the existence of protein variants at the DNA-level. Thereby, only a single signal peptide variant of LGB1, but 10 different putative protein variants of LGB2 were identified. In horses, both genes are expressed and in such this is a striking previously unknown difference in genetic variability between the two genes. It can be assumed that LGB1 is the ancestral paralogue, which has an essential function causing a high selection pressure. As horses have very low milk fat content this unknown function might well be related to vitamin-uptake. Further studies are, however, needed, to elucidate the properties of the different gene products.

Certolizumab treatment during late pregnancy in patients with rheumatic diseases: Low drug levels in cord blood but possible risk for maternal infections. A case series of 13 patients

OBJECTIVE Due to reduction of immune-suppressive drugs, patients with rheumatic diseases can experience an increase in disease activity during pregnancy. In such cases, TNF-inhibitors may be prescribed. However, monoclonal antibodies with the Fc moiety are actively transported across the placenta, resulting in therapeutic drug levels in the newborn. As certolizumab (CZP) lacks the Fc moiety, it may bear a lower risk for the child. METHOD We report a case series of thirteen patients (5 with rheumatoid arthritis and 8 with spondyloarthritis) treated with CZP during late pregnancy to control disease activity. RESULT CZP measured in cord blood of eleven infants ranged between undetectable levels and 1μg/mL whereas the median CZP level of maternal plasma was 32.97μg/mL. Three women developed an infection during the third trimester, of whom one had a severe infection and one had an infection that resulted in a pre-term delivery. During the postpartum period, 6 patients remained on CZP while breastfeeding. CZP levels in the breast milk of two breastfeeding patients were undetectable. CONCLUSION The lack of the active transplacental transfer of CZP gives the possibility to treat inflammatory arthritis during late gestation without potential harm to the newborn. However, in pregnant women treated with TNF-inhibitors and prednisone, attention should be given to the increased susceptibility to infections, which might cause prematurity. CZP treatment can be continued while breastfeeding.