A new approach to assess the degree of contamination and determine sources and risks related to PTEs in an urban environment: the case study of Santiago (Chile)

In 2017, a geochemical survey was carried out across the Commune of Santiago, a local administrative unit located at the center of the namesake capital city of Chile, and the concentration of a number of major and trace elements (53 in total) was determined on 121 topsoil samples. Multifractal IDW (MIDW) interpolation method was applied to raw data to generate geochemical baseline maps of 15 potential toxic elements (PTEs); the concentration–area (C-A) plot was applied to MIDW grids to highlight the fractal distribution of geochemical data. Data of PTEs were elaborated to statistically determine local geochemical baselines and to assess the spatial variation of the degree of soil contamination by means of a new method taking into account both the severity of contamination and its complexity. Afterwards, to discriminate the sources of PTEs in soils, a robust Principal Component Analysis (PCA) was applied to data expressed in isometric log-ratio (ilr) coordinates. Based on PCA results, a Sequential Binary Partition (SBP) was also defined and balances were determined to generate contrasts among those elements considered as proxies of specific contamination sources (Urban traffic, productive settlements, etc.). A risk assessment was finally completed to potentially relate contamination sources to their potential effect on public health in the long term. A probabilistic approach, based on Monte Carlo method, was deemed more appropriate to include uncertainty due to spatial variation of geochemical data across the study area. Results showed how the integrated use of multivariate statistics and compositional data analysis gave the authors the chance to both discriminate between main contamination processes characterizing the soil of Santiago and to observe the existence of secondary phenomena that are normally difficult to constrain. Furthermore, it was demonstrated how a probabilistic approach in risk assessment could offer a more reliable view of the complexity of the process considering uncertainty as an integral part of the results.

Majority quantifiers based on nouns

Strings of the type the largest/LARGER PART or (THE) LARGE PART, together with MAJORITY nouns, are the most widespread means of expressing majority judgments. We take this to constitute evidence in favor of a compositional analysis, which builds the majority reading by combining the superlative form of the adjective LARGE (or the comparative or positive forms, in some languages) with the functional noun PART, which introduces an unspecified binary partition. We propose a possible extension of this analysis to abstract nouns of the MAJORITY type. We also discuss a peculiar type of relative superlative reading allowed by MAJORITY nouns (in addition to their majority reading), which is identical to the one observed by Kotek et al. (2011) for most of in English. We finally offer three case studies for the majority quantifiers found in Hindi, Latin, and Syrian Arabic.

Segment-Based Clustering of Hyperspectral Images Using Tree-Based Data Partitioning Structures

Hyperspectral image classification has been increasingly used in the field of remote sensing. In this study, a new clustering framework for large-scale hyperspectral image (HSI) classification is proposed. The proposed four-step classification scheme explores how to effectively use the global spectral information and local spatial structure of hyperspectral data for HSI classification. Initially, multidimensional Watershed is used for pre-segmentation. Region-based hierarchical hyperspectral image segmentation is based on the construction of Binary partition trees (BPT). Each segmented region is modeled while using first-order parametric modelling, which is then followed by a region merging stage using HSI regional spectral properties in order to obtain a BPT representation. The tree is then pruned to obtain a more compact representation. In addition, principal component analysis (PCA) is utilized for HSI feature extraction, so that the extracted features are further incorporated into the BPT. Finally, an efficient variant of k-means clustering algorithm, called filtering algorithm, is deployed on the created BPT structure, producing the final cluster map. The proposed method is tested over eight publicly available hyperspectral scenes with ground truth data and it is further compared with other clustering frameworks. The extensive experimental analysis demonstrates the efficacy of the proposed method.

A tree-based dictionary learning framework

We propose a new outline for adaptive dictionary learning methods for sparse encoding based on a hierarchical clustering of the training data. Through recursive application of a clustering method, the data is organized into a binary partition tree representing a multiscale structure. The dictionary atoms are defined adaptively based on the data clusters in the partition tree. This approach can be interpreted as a generalization of a discrete Haar wavelet transform. Furthermore, any prior knowledge on the wanted structure of the dictionary elements can be simply incorporated. The computational complexity of our proposed algorithm depends on the employed clustering method and on the chosen similarity measure between data points. Thanks to the multiscale properties of the partition tree, our dictionary is structured: when using Orthogonal Matching Pursuit to reconstruct patches from a natural image, dictionary atoms corresponding to nodes being closer to the root node in the tree have a tendency to be used with greater coefficients.

Establishment of an integrated model for predicting survival and guiding treatment in local recurrence nasopharyngeal carcinoma

Objective:In this study, we aimed to establish an integrated prognostic model for local recurrence nasopharyngeal carcinoma (lrNPC) patients, and evaluate the benefit of re-radiotherapy (RT) in patients with different risk levels.Materials and methods:In total, 271 patients with lrNPC were retrospectively reviewed in this study. Overall survival (OS) was the primary endpoint. Multivariate analysis was performed to select the significant prognostic factors (P<0.05). A prognostic model for OS was derived by recursive partitioning analysis (RPA) combining independent predictors using the algorithm of optimized binary partition.Results:Three independent prognostic factors (age, relapsed T [rT] stage, and Epstein-Barr virus [EBV] DNA) were identified from multivariable analysis. Five prognostic groups were derived from an RPA model that combined rT stage and EBV DNA. After further pair-wise comparisons of survival outcome in each group, three risk groups were generated. We investigated the role of re-RT in different risk groups, and found that re-RT could benefit patients in the low (P<0.001) and intermediate-risk subgroups (P=0.017), while no association between re-RT and survival benefit was found in the high-risk subgroup (P=0.328).Conclusion:Age, rT stage and EBV DNA were identified as independent predictors for lrNPC. We established an integrated RPA-based prognostic model for OS incorporating rT stage and EBV DNA, which could guide individual treatment for lrNPC.