Time-based self-supervised learning for Wireless Capsule Endoscopy.

State-of-the-art machine learning models, and especially deep learning ones, are significantly data-hungry; they require vast amounts of manually labeled samples to function correctly. However, in most medical imaging fields, obtaining said data can be challenging. Not only the volume of data is a problem, but also the imbalances within its classes; it is common to have many more images of healthy patients than of those with pathology. Computer-aided diagnostic systems suffer from these issues, usually over-designing their models to perform accurately. This work proposes using self-supervised learning for wireless endoscopy videos by introducing a custom-tailored method that does not initially need labels or appropriate balance. We prove that using the inferred inherent structure learned by our method, extracted from the temporal axis, improves the detection rate on several domain-specific applications even under severe imbalance. State-of-the-art results are achieved in polyp detection, with 95.00 ± 2.09% Area Under the Curve, and 92.77 ± 1.20% accuracy in the CAD-CAP dataset.

Sequential Models for Endoluminal Image Classification.

Wireless Capsule Endoscopy (WCE) is a procedure to examine the human digestive system for potential mucosal polyps, tumours, or bleedings using an encapsulated camera. This work focuses on polyp detection within WCE videos through Machine Learning. When using Machine Learning in the medical field, scarce and unbalanced datasets often make it hard to receive a satisfying performance. We claim that using Sequential Models in order to take the temporal nature of the data into account improves the performance of previous approaches. Thus, we present a bidirectional Long Short-Term Memory Network (BLSTM), a sequential network that is particularly designed for temporal data. We find the BLSTM Network outperforms non-sequential architectures and other previous models, receiving a final Area under the Curve of 93.83%. Experiments show that our method of extracting spatial and temporal features yields better performance and could be a possible method to decrease the time needed by physicians to analyse the video material.

Morphological determinants of jumping performance in the Iberian green frog.

Predation is one of the main selective forces in nature, frequently selecting potential prey for developing escape strategies. Escape ability is typically influenced by several morphological parameters, such as morphology of the locomotor appendices, muscular capacity, body mass, or fluctuating asymmetry, and may differ between sexes and age classes. In this study, we tested the relationship among these variables and jumping performance in 712 Iberian green frogs Pelophylax perezi from an urban population. The results suggest that the main determinant of jumping capacity was body size (explaining 48% of variance). Larger frogs jumped farther, but jumping performance reached an asymptote for the largest frogs. Once controlled by structural body size, the heaviest frogs jumped shorter distances, suggesting a trade-off between fat storage and jumping performance. Relative hind limb length also determined a small but significant percentage of variance (2.4%) in jumping performance-that is, the longer the hind limbs, the greater the jumping capacity. Juveniles had relatively shorter and less muscular hind limbs than adults (for a given body size), and their jumping performance was poorer. In our study population, the hind limbs of the frogs were very symmetrical, and we found no effect of fluctuating asymmetry on jumping performance. Therefore, our study provides evidence that jumping performance in frogs is not only affected by body size, but also by body mass and hind limb length, and differ between age classes.

Habitat use in south-west European skinks (genus Chalcides).

BACKGROUND: Congeneric species of reptiles frequently exhibit partitioning in terms of their use of habitats or trophic resources in order to reduce competition. In this study, we investigated habitat use by two species of European skinks: Chalcides bedriagai and Chalcides striatus, based on 49 records from southern France, Spain, and Portugal. METHODS: We measured three levels of niche descriptors: macroscale (climate, topography, and substrate), mesoscale (plant associations), and microscale (vegetation cover and shelters). We assessed the associations between these environmental descriptors and the occurrence of the skinks. RESULTS: Our results showed that the two species occupied opposite extremes of the ecological gradient i.e., C. bedriagai in semi-arid environments and C. striatus in temperate-oceanic environments, but there was broad ecological overlap in transitional climates at all of the habitat scales examined. This overlap was demonstrated by the presence of syntopy in geographically distant sites with different environmental characteristics. DISCUSSION: The morphological differences between the two species, and possibly their different use of microhabitats, might favor this mesoscale overlap between congeneric species, which is relatively unusual in Mediterranean lizards.

Generic feature learning for wireless capsule endoscopy analysis.

The interpretation and analysis of wireless capsule endoscopy (WCE) recordings is a complex task which requires sophisticated computer aided decision (CAD) systems to help physicians with video screening and, finally, with the diagnosis. Most CAD systems used in capsule endoscopy share a common system design, but use very different image and video representations. As a result, each time a new clinical application of WCE appears, a new CAD system has to be designed from the scratch. This makes the design of new CAD systems very time consuming. Therefore, in this paper we introduce a system for small intestine motility characterization, based on Deep Convolutional Neural Networks, which circumvents the laborious step of designing specific features for individual motility events. Experimental results show the superiority of the learned features over alternative classifiers constructed using state-of-the-art handcrafted features. In particular, it reaches a mean classification accuracy of 96% for six intestinal motility events, outperforming the other classifiers by a large margin (a 14% relative performance increase).

Inventory and treatment of compost maturation emissions in a municipal solid waste treatment facility.

Emissions of volatile organic compounds (VOCs) from the compost maturation building in a municipal solid waste treatment facility were inventoried by solid phase microextraction and gas chromatography-mass spectrometry. A large diversity of chemical classes and compounds were found. The highest concentrations were found for n-butanol, methyl ethyl ketone and limonene (ppmv level). Also, a range of compounds exceeded their odor threshold evidencing that treatment was needed. Performance of a chemical scrubber followed by two parallel biofilters packed with an advanced packing material and treating an average airflow of 99,300 m(3) h(-1) was assessed in the treatment of the VOCs inventoried. Performance of the odor abatement system was evaluated in terms of removal efficiency by comparing inlet and outlet abundances. Outlet concentrations of selected VOCs permitted to identify critical odorants emitted to the atmosphere. In particular, limonene was found as the most critical VOC in the present study. Only six compounds from the odorant group were removed with efficiencies higher than 90%. Low removal efficiencies were found for most of the compounds present in the emission showing a significant relation with their chemical properties (functionality and solubility) and operational parameters (temperature, pH and inlet concentration). Interestingly, benzaldehyde and benzyl alcohol were found to be produced in the treatment system.