Nicolas Boutry

Euler well-composedness

Abstract

In this paper, we define a new flavour of well-composedness, called Euler well-composedness, in the general setting of regular cell complexes: A regular cell complex is Euler well-composed if the Euler characteristic of the link of each boundary vertex is $1$. A cell decomposition of a picture $I$ is a pair of regular cell complexes $\big(K(I),K(\bar{I})\big)$ such that $K(I)$ (resp. $K(\bar{I})$) is a topological and geometrical model representing $I$ (resp. its complementary, $\bar{I}$). Then, a cell decomposition of a picture $I$ is self-dual Euler well-composed if both $K(I)$ and $K(\bar{I})$ are Euler well-composed. We prove in this paper that, first, self-dual Euler well-composedness is equivalent to digital well-composedness in dimension 2 and 3, and second, in dimension 4, self-dual Euler well-composedness implies digital well-composedness, though the converse is not true.

Continue reading

A new minimum barrier distance for multivariate images with applications to salient object detection, shortest path finding, and segmentation

Abstract

Distance transforms and the saliency maps they induce are widely used in image processing, computer vision, and pattern recognition. One of the most commonly used distance transform is the geodesic one. Unfortunately, this distance does not always achieve satisfying results on noisy or blurred images. Recently, a new (pseudo-)distance, called the minimum barrier distance (MBD), more robust to pixel variations, has been introduced. Some years after, Géraud et al. have proposed a good and fast-to compute approximation of this distance: the Dahu pseudo-distance. Since this distance was initially developped for grayscale images, we propose here an extension of this transform to multivariate images; we call it vectorial Dahu pseudo-distance. An efficient way to compute it is provided in this paper. Besides, we provide benchmarks demonstrating how much the vectorial Dahu pseudo-distance is more robust and competitive compared to other MB-based distances, which shows how much this distance is promising for salient object detection, shortest path finding, and object segmentation.

Continue reading

Using separated inputs for multimodal brain tumor segmentation with 3D U-Net-like architectures

Abstract

The work presented in this paper addresses the MICCAI BraTS 2019 challenge devoted to brain tumor segmentation using mag- netic resonance images. For each task of the challenge, we proposed and submitted for evaluation an original method. For the tumor segmentation task (Task 1), our convolutional neural network is based on a variant of the U-Net architecture of Ronneberger et al. with two modifications: first, we separate the four convolution parts to decorrelate the weights corresponding to each modality, and second, we provide volumes of size 240 * 240 * 3 as inputs in these convolution parts. This way, we profit of the 3D aspect of the input signal, and we do not use the same weights for separate inputs. For the overall survival task (Task 2), we compute explainable features and use a kernel PCA embedding followed by a Random Forest classifier to build a predictor with very few training samples. For the uncertainty estimation task (Task 3), we introduce and compare lightweight methods based on simple principles which can be applied to any segmentation approach. The overall performance of each of our contribution is honorable given the low computational requirements they have both for training and testing.

Continue reading

Experimental digital gabor hologram rendering of C. Elegans worms by a model-trained convolutional neural network (conference presentation)

Abstract

Continue reading

A two-stage temporal-like fully convolutional network framework for left ventricle segmentation and quantification on MR images

Abstract

Automatic segmentation of the left ventricle (LV) of a living human heart in a magnetic resonance (MR) image (2D+t) allows to measure some clinical significant indices like the regional wall thicknesses (RWT), cavity dimensions, cavity and myocardium areas, and cardiac phase. Here, we propose a novel framework made of a sequence of two fully convolutional networks (FCN). The first is a modified temporal-like VGG16 (the “localization network”) and is used to localize roughly the LV (filled-in) epicardium position in each MR volume. The second FCN is a modified temporal-like VGG16 too, but devoted to segment the LV myocardium and cavity (the “segmentation network”). We evaluate the proposed method with 5-fold-cross-validation on the MICCAI 2019 LV Full Quantification Challenge dataset. For the network used to localize the epicardium, we obtain an average dice index of 0.8953 on validation set. For the segmentation network, we obtain an average dice index of 0.8664 on validation set (there, data augmentation is used). The mean absolute error (MAE) of average cavity and myocardium areas, dimensions, RWT are 114.77 mm^2; 0.9220 mm; 0.9185 mm respectively. The computation time of the pipeline is less than 2 s for an entire 3D volume. The error rate of phase classification is 7.6364%, which indicates that the proposed approach has a promising performance to estimate all these parameters.

Continue reading

Connected filters on generalized shape-spaces

Abstract

Classical hierarchical image representations and connected filters work on sets of connected components (CC). These approaches can be defective to describe the relations between disjoint objects or partitions on images. In practice, objects can be made of several connected components in images (due to occlusions for example), therefore it can be interesting to be able to take into account the relationship between these components to be able to detect the whole object. In Mathematical Morphology, second-generation connectivity (SGC) and tree-based shape-space study this relation between the connected components of an image. However, they have limitations. For this reason, we propose in this paper an extension of the usual shape-space paradigm into what we call a Generalized Shape-Space (GSS). This new paradigm allows to analyze any graph of connected components hierarchically and to filter them thanks to connected operators.

Continue reading

One more step towards well-composedness of cell complexes over $n$-D pictures

Abstract

An $n$-D pure regular cell complex $K$ is weakly well-composed (wWC) if, for each vertex $v$ of $K$, the set of $n$-cells incident to $v$ is face-connected. In previous work we proved that if an $n$-D picture $I$ is digitally well composed (DWC) then the cubical complex $Q(I)$ associated to $I$ is wWC. If $I$ is not DWC, we proposed a combinatorial algorithm to locally repair $Q(I)$ obtaining an $n$-D pure simplicial complex $P_S(I)$ homotopy equivalent to $Q(I)$ which is always wWC. In this paper we give a combinatorial procedure to compute a simplicial complex $P_S(\bar{I})$ which decomposes the complement space of $|P_S(I)|$ and prove that $P_S(\bar{I})$ is also wWC. This paper means one more step on the way to our ultimate goal: to prove that the $n$-D repaired complex is continuously well-composed (CWC), that is, the boundary of its continuous analog is an $(n-1)$-manifold.

Continue reading

An equivalence relation between morphological dynamics and persistent homology in 1D

Abstract

We state in this paper a strong relation existing between Mathematical Morphology and Discrete Morse Theory when we work with 1D Morse functions. Specifically, in Mathematical Morphology, a classic way to extract robust markers for segmentation purposes, is to use the dynamics. On the other hand, in Discrete Morse Theory, a well-known tool to simplify the Morse-Smale complexes representing the topological information of a Morse function is the persistence. We show that pairing by persistence is equivalent to pairing by dynamics. Furthermore, self-duality and injectivity of these pairings are proved.

Continue reading

How to make $n$-D plain maps Alexandrov-well-composed in a self-dual way

Abstract

In 2013, Najman and Géraud proved that by working on a well-composed discrete representation of a gray-level image, we can compute what is called its tree of shapes, a hierarchical representation of the shapes in this image. This way, we can proceed to morphological filtering and to image segmentation. However, the authors did not provide such a representation for the non-cubical case. We propose in this paper a way to compute a well-composed representation of any gray-level image defined on a discrete surface, which is a more general framework than the usual cubical grid. Furthermore, the proposed representation is self-dual in the sense that it treats bright and dark components in the image the same way. This paper can be seen as an extension to gray-level images of the works of Daragon et al. on discrete surfaces.

Continue reading

Weakly well-composed cell complexes over $n$D pictures

Abstract

In previous work we proposed a combinatorial algorithm to “locally repair” the cubical complex $Q(I)$ that is canonically associated with a given 3D picture I. The algorithm constructs a 3D polyhedral complex $P(I)$ which is homotopy equivalent to $Q(I)$ and whose boundary surface is a 2D manifold. A polyhedral complex satisfying these properties is called well-composed. In the present paper we extend these results to higher dimensions. We prove that for a given $n$-dimensional picture the obtained cell complex is well-composed in a weaker sense but is still homotopy equivalent to the initial cubical complex.

Continue reading