Éditeur(s) : HAL CCSD
Résumé : International audience
In this paper we propose a hybrid approach using mutual information and multi-objective optimization for feature subset selection problem. The hybrid aspect is due to the sequence of a filter method and a wrapper method in order to take advantages of both. The filter method reduces the exploration space by keeping subsets having good internal properties and the wrapper method chooses among the remaining subsets with a classification performances criterion. In the filter step, the subsets are evaluated in a multi-objective way to ensure diversity within the subsets. The evaluation is based on the mutual information to estimate the dependency between features and classes and the redundancy between features within the same subset. We kept the non-dominated (Pareto optimal) subsets for the second step. In the wrapper step, the selection is made according to the stability of the subsets regarding classification performances during learning stage on a set of classifiers to avoid the specialization of the selected subsets for a given classifiers. The proposed hybrid approach is experimented on a variety of reference data sets and compared to the classical feature selection methods FSDD and mRMR. The resulting algorithm outperforms these algorithms and the computation complexity remains acceptable even if it increases with regards to these two fast selection methods.
Journal of Informatics and Mathematical Sciences

hal-01375982
https://hal.archives-ouvertes.fr/hal-01375982

Éditeur(s) : HAL CCSD
Résumé : International audience
Clear cell renal cell carcinoma (ccRCC) has a poor prognosis with a 50% risk of metastases. Little is known about the phenotypic and molecular profiles of metastases regarding their corresponding primary tumors. This study aimed to screen phenotypic and genotypic differences between metastases and their corresponding primary tumors. We selected four cases with available frozen material. The histological, immunohistochemical (VEGFA, CD31, SMA, Ki67, p53, PAR-3), FISH (VHL gene), next-generation sequencing (VHL and c-MET genes), multiplex ligation-dependent probe amplification, and array-(comparative genomic hybridization) CGH analyses were realized. Metastases were nodal, hepatic (synchronous), adrenal, and pulmonary (metachronous). High-grade tumor cells were significantly more frequent in metastases (p = 0.019). Metastases and high-grade zones of primary tumors shared similar characteristics compared to low-grade zones: a lower microscopic vascular density (43.5 vs 382.5 vessels/mm(2); p = 0.0027), a higher expression of VEGF (73 vs 10%, p = 0.045), Ki67 (37.6 vs 8.3%; p = 0.011), and p53 (54 vs 10.6%; p = 0.081), and a cytoplasmic and membranous PAR-3 staining. Metastases exhibited more chromosomal imbalances than primary tumors in total (18.75 ± 6.8; p = 0.044) with more genomic gains (13.5 ± 7; p = 0.013). The loss of chromosome 9 and gain of Xq were found in both primary tumors and metastases but gains of loci or chromosomes 2p, 3q, 5, 8q, 12, and 20 were only found in metastases. The VHL gene status was similar in each tumor couple. Although metastases and primary tumors share common histological features, this study highlights chromosomal differences specific to metastases which could be involved in ccRCC metastatic evolution.
Virchows Archiv

hal-01560281
https://hal-univ-rennes1.archives-ouvertes.fr/hal-01560281
DOI : 10.1007/s00428-017-2124-0
PUBMED : 28488172