Feature Selection based on Machine Learning in MRIs for Hippocampal Segmentation

Neurodegenerative diseases are frequently associated with structural changes in the brain. Magnetic Resonance Imaging (MRI) scans can show these variations and therefore be used as a supportive feature for a number of neurodegenerative diseases. The hippocampus has been known to be a biomarker for Alzheimer disease and other neurological and psychiatric diseases. However, it requires accurate, robust and reproducible delineation of hippocampal structures. Fully automatic methods are usually the voxel based approach, for each voxel a number of local features were calculated. In this paper we compared four different techniques for feature selection from a set of 315 features extracted for each voxel: (i) filter method based on the Kolmogorov-Smirnov test; two wrapper methods, respectively, (ii) Sequential Forward Selection and (iii) Sequential Backward Elimination; and (iv) embedded method based on the Random Forest Classifier on a set of 10 T1-weighted brain MRIs and tested on an independent set of 25 subjects. The resulting segmentations were compared with manual reference labelling. By using only 23 features for each voxel (sequential backward elimination) we obtained comparable state of-the-art performances with respect to the standard tool FreeSurfer.

Data Driven Discovery in Astrophysics

We review some aspects of the current state of data-intensive astronomy, its methods, and some outstanding data analysis challenges. Astronomy is at the forefront of big data science, with exponentially growing data volumes and data rates, and an ever-increasing complexity, now entering the Petascale regime. Telescopes and observatories from both ground and space, covering a full range of wavelengths, feed the data via processing pipelines into dedicated archives, where they can be accessed for scientific analysis. Most of the large archives are connected through the Virtual Observatory framework, that provides interoperability standards and services, and effectively constitutes a global data grid of astronomy. Making discoveries in this overabundance of data requires applications of novel, machine learning tools. We describe some of the recent examples of such applications.

The detection of globular clusters in galaxies as a data mining problem

We present an application of self-adaptive supervised learning classifiers derived from the Machine Learning paradigm, to the identification of candidate Globular Clusters in deep, wide-field, single band HST images. Several methods provided by the DAME (Data Mining & Exploration) web application, were tested and compared on the NGC1399 HST data described in Paolillo 2011. The best results were obtained using a Multi Layer Perceptron with Quasi Newton learning rule which achieved a classification accuracy of 98.3%, with a completeness of 97.8% and 1.6% of contamination. An extensive set of experiments revealed that the use of accurate structural parameters (effective radius, central surface brightness) does improve the final result, but only by 5%. It is also shown that the method is capable to retrieve also extreme sources (for instance, very extended objects) which are missed by more traditional approaches.