Graziano Frosini

One-dimensional and two-dimensional generalised discrete fourier transforms

One-dimensional and two-dimensional generalized discrete Fourier transforms (GFT) are introduced. If a one-dimensional vector A is fractured into a two-dimensional matrix B, a one-dimensional GFT on A and a two-dimensional GFT on B give the same result and require the same number of operations to be computed. The result holds also for the DFT, as it is a particular case of the GFT.

Performing automatic exams

Abstract In this paper we describe a tool to build software systems which replace the role of the examiner during a typical Italian academic exam in technical/scientific subjects. Such systems are designed so as to exploit the advantages of self-adapted testing (SAT) for reducing the effect of anxiety and of computerised adaptive testing (CAT) for increasing the assessment efficiency. A SAT-like pre-exam determines the starting difficulty level of the following CAT. The exam can thus be dynamically adapted to suit the ability of the student, i.e. by making it more difficult or easier as required. The examiner simply needs to associate the level of difficulty with a suitable number of initial queries. After posing these queries to a sample group of students and collecting statistics, the tool can automatically associate a level of difficulty with each subsequent query by submitting it to sample groups of students. In addition, the tool can automatically assign a score to the levels and to the queries. Finally, the systems collect statistics so as to measure the easiness and selectivity of each query and to evaluate the validity and reliability of an exam.

Procedures for computing the discrete Fourier transform on staggered blocks

In this work the problem of evaluating successively the discrete Fourier transform (DFT) on ordered sets of N elements staggered of M is considered. Three procedures for solving such a problem are given, of which two are recursive and one nonrecursive. The complexity of each procedure, in number of complex multiplications, is about (N/2) \log_{2} 4M .

A CAD System for Lung Nodule Detection based on an Anatomical Model and a Fuzzy Neural Network

An automated system was developed for lung nodule detection in lung low-dose computer tomography (CT) scans. The system exploits a thorax anatomical model in order to distinguish the 3D anatomical structures corresponding, in the order, to the chest wall, the trachea and the two lung lobes. Each anatomical structure is described in terms of characteristics like volume, X-ray attenuation, and position with respect to structures already recognised. Once the two pulmonary parenchymas have been isolated from the rest of the chest, nodules are looked for inside them. More precisely, the robust fuzzy c-means algorithm is applied to classify the lung area into two clusters: the former includes nodules and blood vessels, the latter consists of air. 3D regions of interest (ROIs) are then identified in the former cluster and features are extracted from the ROIs for fuzzy neural network-based recognition of nodules from vessels. Experiments on 20 clinical cases (including about 7,000 overall images) showed 100% and nearly 82% correct recognition rate for nodules (0.5-3 cm) and micro-nodules (3-5 mm), respectively, and an average of 1.4 false positives for image

A modified fuzzy C-means algorithm for feature selection

In this paper we propose a novel method for feature selection based on a modified fuzzy C-means algorithm with supervision (MFCMS). MFCMS adopts an appropriately modified version of the objective function used by the classic fuzzy C-means. We applied MFCMS to some real-world pattern classification benchmarks. To test the effectiveness of MFCMS as feature selector, we used the well-known k-nearest neighbor as learning algorithm. In our experiments we found that the classification performance using the set of features selected by MFCMS is better than that using all the original features. Furthermore, our approach proved to be less time consuming than other feature selection methods.

Automated Detection of Pulmonary Nodules in CT Scans

In this paper, we describe a computer-aided diagnosis (CAD) system for automated detection of pulmonary nodules in computed-tomography (CT) examinations. After a segmentation phase based on the robust fuzzy c-means (RFCM) algorithm proposed by Pham, the regions of interest (ROIs) undergo both 2D and 3D morphological analysis in order to distinguish between nodule and blood vessel sections. The system has been applied to eight CT scans, including four at high-dose radiation and four at low-dose radiation, with a total of about 2400 digital images. Three of the eight scans are pertinent to as many patients suffering from lung cancer. A total of 10 malignant nodules are present in these scans, ranging from 4 mm to 10 mm in diameter. We achieved no false negatives (i.e., true nodules that are not found by the algorithm) and an average of approximately two false-positives (i.e., non-nodules recognized as nodules) per CT image

Distributing and revoking access authorizations on abstract objects: a capability approach

In this paper a capability addressing environment is presented, based on the concept of extended capability. First of all it is shown that such an environment is well suited for implementing objects of abstract type. Then the problem of distributing and revoking access authorizations on abstract objects is considered and an efficient solution is presented. The revocation mechanism results in being selective, transitive and deferred.

The Serial Microprocessor Array (SMA): Microprogramming and application examples

The structure of the Processing Element (PE), which is the basic component of SMA1, is presented. The PE consists of a simple serial arithmetic unit, a local high speed data memory, serial input and output ports, serial communication channels with neighbouring PEs, and some local control logic. The PE array operates under the control of a microprogrammed Array Control Unit (ACU). The peculiarities of ACU microprogramming are discussed, and some typical microprograms are reported. After presentation of the SMA principal instructions, some application programs are described implementing common radar filtering algorithms.

Modelling a team of radiologists for lung nodule detection in CT scans

This paper describes a system for automatic detection of pulmonary nodules in lung CT (Computed Tomography) images. After modelling the activity of a single radiologist as two subsequent phases, namely, the regions of interest (ROIs) detection phase and the nodule detection phase, we built a system which emulates a team of radiologists. This is achieved by providing a further phase of collaboration and opinion exchange among the experts at the end of each of the previous phases. We also present experimental results, based on the ROC convex hull method, which show how the team of radiologists obtains better performance than the single best radiologist in both phases. In particular, we achieved a sensitivity of 92.48% against a specificity of about 83.54% in the nodule detection phase.

A new approach to combining outputs of multiple classifiers

This paper presents a novel method for multiple classifier fusion. The classifier combiner operates on the single classifier outputs, which consist of vectors of pairs (c, d), with c being a class name and d the confidence degree with which a pattern is recognized as belonging to class c. The main idea of the combiner is to exploit the knowledge of the statistical behavior of the single classifiers on the training set to re-calculate a global recognition confidence degree based on the a posteriori probability that the input pattern belongs to a given class conditioned by the specific responses of the classifiers. Applying the Bayess theorem we can also easily adapt our classifier combiner to a specific application. We compare our model with some popular techniques for classifier fusion on the Satimage and Phoneme data sets from. the database ELENA.. We show that our method is in most cases superior (or substantially equivalent) to the other techniques on both data sets.