Research


Discrete Stochastic Model for the Generation of Axonal Trees

Abstract: In this work we propose a 2D discrete stochastic model for the simulation of axonal biogenesis. The model is defined by a third order Markov Chain. The model considers two main processes: the growth process that models the elongation and shape of the neurites and the bifurcation process that models the generation of branches. The growth process depends, among other variables, on the external attraction field generated by a chemo attractant molecule secreted by the target area. We propose an estimation scheme of the involved parameters from real fluorescent confocal microscopy images of single neurons within intact adult Drosophila fly brains. Both normal neurons and neurons in which certain genes were inactivated have been considered (two mutations). In total, 53 images (18 normal, 21 type 1 mutant and 14 type 2 mutant) were used. The model parameters allow us to describe pathological characteristics of the mutated populations.


normal28     normal_synth     

mutant3     mutant_synth      

Real normal and mutant type 1 axonal trees (left top and bottom respectively) and synthetic trees (right, top and bottom) generated using the estimated parameters

Norm_normal_scale     Field_normal      

Norm (left) and direction (right) of the estimated attraction field for the normal population.



Axonal Tree Classification Using an Elastic Shape Analysis Based Distance

Abstract: The analysis of the morphological differences between normal and pathological neuronal structures is of paramount importance. Some methods for the comparison of axonal trees only take into account topological information (such as TED), while others also include geometrical information (such as Path2Path). In a previous work, we have presented a new method for comparing tree-like shapes based on the Elastic Shape Analysis Framework (ESA). In this paper, we extend this method by computing the mean shape of a population. Moreover, we propose to evaluate and compare these 3 approaches (TED, Path2Path and ESA) with a classification scheme based on feature computation and K-means. We evaluate these approaches on a database of 44 real 3D confocal microscopy images of two populations of neurons. Results show that the proposed method distinguishes better between the two populations.


normal1    normal2    normal3

mutant1    mutant2     mutant3

Examples of axonal trees from the normal (top) and mutant (bottom)populations (2D projections).

mean_shapes

Mean normal (left) and mutant (right) axonal trees (2D projections).



Tree-like Shapes Distance Using the Elastic Shape Analysis Framework


Abstract: The analysis and comparison of tree-like shapes is of great importance since many structures in nature can be described by them. In the field of biomedical imaging, trees have been used to describe structures such as neurons, blood vessels and lung airways.Since it is known that their morphology provides information on their functioning and allows the characterization of pathological states, it is of paramount importance to develop methods to analyze their shape and to quantify differences in structures.In this paper, we present a new method for comparing tree-like shapes that takes into account both topological and geometrical information. It is based on the Elastic Shape Analysis Framework, a framework originally designed for comparing shapes of 3D closed curves in Euclidean spaces. As a first application, we used our method for the comparison of axon morphology. The performance was tested on a group of 44 (20 normal and 24 mutant) 3D images, each containing one axonal tree. We have calculated inter and intra class distances between them and implemented a basic classification scheme.Results showed that the proposed method better distinguishes between the two populations than a pure topological metric. Furthermore, mean shapes can be obtained with this method.


normal1        normal2        geodesic

Original images (left, middle) and transformation between the two trees(2D projections).



Axon Extraction from Fluorescent Confocal Microscopy Images


Abstract: The morphological analysis of axonal trees is an important problem in neuroscience. The first step for such an analysis is the extraction of the axon. Due to the high volume of generated image data and the tortuous nature of the axons,manual processing is not feasible. Therefore, it is necessary to develop techniques for the automatic extraction of the neuronal structures. In this paper we present a new approach for the automatic extraction of axons from fluorescent confocal microscopy images. It combines algorithms for filament enhancement, binarization,skeletonization and gap filling in a pipeline capable of extracting the axons. The performance of the proposed method was evaluated on real images. Results support the potential use of this technique in helping biologists perform automatic extraction of axons from fluorescent confocal microscopy images.


segmentation

Comparison between original image (left), our result (middle) and ground truth (right) for two images (maximum intensity projections).

normal_axon             reconstruction

Original maximum intensity projection image of a normal axon (left) and extracted axon in 3D (right).



Detection and Tracking of Axonal Tips from Bi-Photon Microscopy Images

Abstract:Live cell two-photon microscopy is an effective tool for the analysis of dynamical processes occurring in living samples. This technique,when combined with fuorescence, allows the detection of objects of interest in 3D space and time.  Due to the high volume of data,the automatic analysis of the images is desired. To this end, the Marked Point Process (MPP) detection framework was selected. MPP is a probabilistic framework which has been successfully applied to the detection of objects in different image processing applications. Its main advantages are that the number of objects to be detected can be unknown, and that geometric constraints on the objects can be easily modeled. As a first approximation, we proposed a 3D MPP model of spheres to extract the axonal extremities. We define a prior energy designed to penalizes (but not forbid) overlaps, and a data energy based on the Bhattacharya distance between the distributions. Once the energy function has been defined, the issue is to find the configuration U(X)that minimizes it. Due to the intricate nature of U(x), usual minimization algorithms cannot be applied. Therefore, we have chosen Multiple Births and Deaths (MBD). The method was evaluated on a set of real 3D+t images. Although results show there is still work to be than in this area, they suggest this approach could be appropriate for solving the extraction problem.

film_biphoton

Dynamic 3D+time image sequences of developing neurons.

tracking

Tracking result for four consecutive frames (left to right).



Integrated Software for the Detection of Epileptogenic Zones in Refractory Epilepsy


Abstract: We present an integrated software designed to help nuclear medicine physicians in the detection of epileptogenic zones (EZ) by means of ictal-interictal SPECT and MR images. This tool was designed to be flexible, friendly and efficient. A novel detection method was included(A-contrario) along with the classical detection method (Subtraction analysis). The software’s performance was evaluated with two separate sets of validation studies: visual interpretation of 12patient images by an experimented observer and objective analysis of virtual brain phantom experiments by proposed numerical observers. Our results support the potential use of the proposed software to help nuclear medicine physicians in the detection of EZ in clinical practice.


detection

Ictal-interictal SPECT and MR images (left) and analysis result (right).

acontrario

Results for one of the patients Subtraction analysis (right). The EZ in A-contrario detection  shows less false activations.