Liliane Dusserre

Anonymous statistical methods versus cryptographic methods in epidemiology

Sensitive data are most often indirectly identifiable and so need to be rendered anonymous in order to ensure privacy. Statistical methods to provide anonymity require data perturbation and so generate data processing difficulties. Encryption methods, while preserving confidentiality, do not require data modification.

High cost factors for leukaemia and lymphoma patients: a new analysis of costs within these diagnosis related groups.

STUDY OBJECTIVE: To determine high cost factors to help managers and clinicians to analyse the reasons of adverse costs and provide indications for financial negotiation. DESIGN: To locate high cost or long stay patients, the analysis was designed on the basis of a mixture of Weibull distributions. In this new model, the proportion of high cost patients was expressed according to the multinomial logistic regression, permitting the determination of high cost factors. SETTING: The 1993 French reference database, constituted in the framework of the national study of DRG costs, conducted by the French Ministry of Health. The database of discharge abstracts recorded in 1993 in the Dijon public teaching hospital. PARTICIPANTS: The analyses were based on 1352 abstracts from the French reference database and 368 from the Dijon database concerning patients, aged 18 and over, suffering from leukaemia and lymphoma. MAIN RESULTS: High cost and long stay factors were the same: number of stays, death, transfer, acute leukaemia, neutropenia, septicaemia, high dose aplastic chemotherapy, central venous catheterisation, parenteral nutrition, protected or laminar airflow room, blood transfusion, and intravenous antibiotherapy. CONCLUSIONS: Taking into account high cost predictive factors, as shown in the case of leukaemia and lymphoma patients, would help to reduce the adverse effects of a prospective payment system.

The Fourier adaptive smoothness constraint for computing optical flow on sequences of angiographic images

This paper deals with a new constraint for determining optical flow fields. This constraint is called the Fourier adaptive smoothness. It can be seen as an improvement of some existing differential techniques. It includes the magnitude image of an extended temporal Fourier transform in the smoothness constraint of the equation used for calculating the flow field. In fact a temporal window of a few frames is displaced along the sequence in order to calculate the one-dimensional corresponding Fourier transform. The resulting local magnitude image allows to constrain the smoothness of the displacement field. The authors compare on angiographic images this approach with two other well known optical flow methods wing differential techniques. A criterion is defined to allow a quantitative analysis of the results. Qualitative remarks are also given. The authors show that the constraint introduced in this paper brings better results in most cases than the two others in addition to its processing speed.

An adaptive smoothness constraint for computing optical flow on sequences of cardiac images

Different algorithms are available for processing an optical flow field between two successive frames. In most situations, two constraints are necessary to estimate such a displacement vector field. The first constraint is based on the gray level variations. The second one is usually based on the smoothness of the flow field. Unfortunately, the results of these constraints are not really adapted to the movements of the ventricle. Here, the authors propose a new method specifically adapted to sequences of cardiac images, for processing the optical flow. They calculate the amplitude image of a sequence of images by the Fourier transform. Unlike methods which identically smooth the optical flow in each direction, the authors developed a Fourier adaptive smoothness method which restricts the variation of the flow field in the directions with small variations of the Fourier amplitude. Moreover, the authors have generalized this method to sequences of non-cyclic images.

A computer model based on superquadrics for the analysis of movement disabilities

The use of motion analysis equipment to study movement disabilities is a major step toward improving treatments of disabled persons. We propose such an approach allowing the analysis of human motion in 3D space without using the tracking of markers. The system we develop is composed of three cameras which acquire synchronized sequences of a human in motion. Then the treatment of our approach is divided into three steps: first, a set of 2D points belonging to the boundaries of the human body in motion are extracted from the three images at each acquisition time. The projection of a human body model is matched with these data. Then, these matchings allow to obtain by reconstruction the configuration of the model in the 3D space at each acquisition time. Finally, some motion parameters can be extracted. The aim of this paper is to describe the human body model that we use and the procedure of matching of its projection. We use an articulated 3D model which consists of a set of non deformable superquadrics. We show that the perspective projection of a class of them can be approximated by a 2D tapered superquadric. Consequently, we propose a fuzzy clustering algorithm adapted to the 2D recognition of the superquadrics.

A method for automatically detecting the systole and diastole phases in sequences of angiographic images

This paper deals with a robust technique for automatically detecting the systole and the diastole phases in sequences of angiographic images. First of all, an optical flow method is used to process a field of displacement vectors between a set of couples of successive images. Then, the calculation of an histogram based on the directions of the vectors of each of the whole optical flow images yields an index, that the authors call the polarization index. It indicates the main motion of the ventricle walls detected between two corresponding initial images: a dilation or a contraction movement. However, some undetermined cases appear: the study of the histogram is not enough for determining the main motion of the ventricle. An interpolation of the values of the polarization index along the sequence is necessary; it gives an accurate measure of the systole and the diastole moments along the sequence. This robust and relatively simple to implement method may be used in lot of applications, and it could be simply integrated as a link of the chain of processes in a system of digital angiography.