• Wing disc pouches of fruit flies are a powerful genetic model for studying physiological intercellular calcium ($Ca^{2+}$) signals for dynamic analysis of cell signaling in organ development and disease studies. A key to analyzing spatial-temporal patterns of $Ca^{2+}$ signal waves is to accurately align the pouches across image sequences. However, pouches in different image frames may exhibit extensive intensity oscillations due to $Ca^{2+}$ signaling dynamics, and commonly used multimodal non-rigid registration methods may fail to achieve satisfactory results. In this paper, we develop a new two-phase non-rigid registration approach to register pouches in image sequences. First, we conduct segmentation of the region of interest. (i.e., pouches) using a deep neural network model. Second, we obtain an optimal transformation and align pouches across the image sequences. Evaluated using both synthetic data and real pouch data, our method considerably outperforms the state-of-the-art non-rigid registration methods.
  • Image segmentation is a fundamental problem in biomedical image analysis. Recent advances in deep learning have achieved promising results on many biomedical image segmentation benchmarks. However, due to large variations in biomedical images (different modalities, image settings, objects, noise, etc), to utilize deep learning on a new application, it usually needs a new set of training data. This can incur a great deal of annotation effort and cost, because only biomedical experts can annotate effectively, and often there are too many instances in images (e.g., cells) to annotate. In this paper, we aim to address the following question: With limited effort (e.g., time) for annotation, what instances should be annotated in order to attain the best performance? We present a deep active learning framework that combines fully convolutional network (FCN) and active learning to significantly reduce annotation effort by making judicious suggestions on the most effective annotation areas. We utilize uncertainty and similarity information provided by FCN and formulate a generalized version of the maximum set cover problem to determine the most representative and uncertain areas for annotation. Extensive experiments using the 2015 MICCAI Gland Challenge dataset and a lymph node ultrasound image segmentation dataset show that, using annotation suggestions by our method, state-of-the-art segmentation performance can be achieved by using only 50% of training data.
  • In this paper, we consider the problem of automatically segmenting neuronal cells in dual-color confocal microscopy images. This problem is a key task in various quantitative analysis applications in neuroscience, such as tracing cell genesis in Danio rerio (zebrafish) brains. Deep learning, especially using fully convolutional networks (FCN), has profoundly changed segmentation research in biomedical imaging. We face two major challenges in this problem. First, neuronal cells may form dense clusters, making it difficult to correctly identify all individual cells (even to human experts). Consequently, segmentation results of the known FCN-type models are not accurate enough. Second, pixel-wise ground truth is difficult to obtain. Only a limited amount of approximate instance-wise annotation can be collected, which makes the training of FCN models quite cumbersome. We propose a new FCN-type deep learning model, called deep complete bipartite networks (CB-Net), and a new scheme for leveraging approximate instance-wise annotation to train our pixel-wise prediction model. Evaluated using seven real datasets, our proposed new CB-Net model outperforms the state-of-the-art FCN models and produces neuron segmentation results of remarkable quality
  • Convolution is a fundamental operation in many applications, such as computer vision, natural language processing, image processing, etc. Recent successes of convolutional neural networks in various deep learning applications put even higher demand on fast convolution. The high computation throughput and memory bandwidth of graphics processing units (GPUs) make GPUs a natural choice for accelerating convolution operations. However, maximally exploiting the available memory bandwidth of GPUs for convolution is a challenging task. This paper introduces a general model to address the mismatch between the memory bank width of GPUs and computation data width of threads. Based on this model, we develop two convolution kernels, one for the general case and the other for a special case with one input channel. By carefully optimizing memory access patterns and computation patterns, we design a communication-optimized kernel for the special case and a communication-reduced kernel for the general case. Experimental data based on implementations on Kepler GPUs show that our kernels achieve 5.16X and 35.5% average performance improvement over the latest cuDNN library, for the special case and the general case, respectively.
  • Segmentation of 3D images is a fundamental problem in biomedical image analysis. Deep learning (DL) approaches have achieved state-of-the-art segmentation perfor- mance. To exploit the 3D contexts using neural networks, known DL segmentation methods, including 3D convolution, 2D convolution on planes orthogonal to 2D image slices, and LSTM in multiple directions, all suffer incompatibility with the highly anisotropic dimensions in common 3D biomedical images. In this paper, we propose a new DL framework for 3D image segmentation, based on a com- bination of a fully convolutional network (FCN) and a recurrent neural network (RNN), which are responsible for exploiting the intra-slice and inter-slice contexts, respectively. To our best knowledge, this is the first DL framework for 3D image segmentation that explicitly leverages 3D image anisotropism. Evaluating using a dataset from the ISBI Neuronal Structure Segmentation Challenge and in-house image stacks for 3D fungus segmentation, our approach achieves promising results comparing to the known DL-based 3D segmentation approaches.
  • The Thomson Problem, arrangement of identical charges on the surface of a sphere, has found many applications in physics, chemistry and biology. Here we show that the energy landscape of the Thomson Problem for $N$ particles with $N=132, 135, 138, 141, 144, 147$ and $150$ is single funnelled, characteristic of a structure-seeking organisation where the global minimum is easily accessible. Algorithmically constructing starting points close to the global minimum of such a potential with spherical constraints is one of Smale's 18 unsolved problems in mathematics for the 21st century because it is important in the solution of univariate and bivariate random polynomial equations. By analysing the kinetic transition networks, we show that a randomly chosen minimum is in fact always `close' to the global minimum in terms of the number of transition states that separate them, a characteristic of small world networks.
  • In this paper, we study the problem of moving $n$ sensors on a line to form a barrier coverage of a specified segment of the line such that the maximum moving distance of the sensors is minimized. Previously, it was an open question whether this problem on sensors with arbitrary sensing ranges is solvable in polynomial time. We settle this open question positively by giving an $O(n^2 \log n)$ time algorithm. For the special case when all sensors have the same-size sensing range, the previously best solution takes $O(n^2)$ time. We present an $O(n \log n)$ time algorithm for this case; further, if all sensors are initially located on the coverage segment, our algorithm takes $O(n)$ time. Also, we extend our techniques to the cycle version of the problem where the barrier coverage is for a simple cycle and the sensors are allowed to move only along the cycle. For sensors with the same-size sensing range, we solve the cycle version in $O(n)$ time, improving the previously best $O(n^2)$ time solution.
  • A fundamental problem in computational geometry is to compute an obstacle-avoiding Euclidean shortest path between two points in the plane. The case of this problem on polygonal obstacles is well studied. In this paper, we consider the problem version on curved obstacles, commonly modeled as splinegons. A splinegon can be viewed as replacing each edge of a polygon by a convex curved edge (polygons are special splinegons). Each curved edge is assumed to be of O(1) complexity. Given in the plane two points s and t and a set of $h$ pairwise disjoint splinegons with a total of $n$ vertices, we compute a shortest s-to-t path avoiding the splinegons, in $O(n+h\log^{1+\epsilon}h+k)$ time, where k is a parameter sensitive to the structures of the input splinegons and is upper-bounded by $O(h^2)$. In particular, when all splinegons are convex, $k$ is proportional to the number of common tangents in the free space (called "free common tangents") among the splinegons. We develop techniques for solving the problem on the general (non-convex) splinegon domain, which also improve several previous results. In particular, our techniques produce an optimal output-sensitive algorithm for a basic visibility problem of computing all free common tangents among $h$ pairwise disjoint convex splinegons with a total of $n$ vertices. Our algorithm runs in $O(n+h\log h+k)$ time and $O(n)$ space, where $k$ is the number of all free common tangents. Even for the special case where all splinegons are convex polygons, the previously best algorithm for this visibility problem takes $O(n+h^2\log n)$ time.
  • Let $\mathcal{P}$ be a set of $h$ pairwise-disjoint polygonal obstacles with a total of $n$ vertices in the plane. We consider the problem of building a data structure that can quickly compute an $L_1$ shortest obstacle-avoiding path between any two query points $s$ and $t$. Previously, a data structure of size $O(n^2\log n)$ was constructed in $O(n^2\log^2 n)$ time that answers each two-point query in $O(\log^2 n+k)$ time, i.e., the shortest path length is reported in $O(\log^2 n)$ time and an actual path is reported in additional $O(k)$ time, where $k$ is the number of edges of the output path. In this paper, we build a new data structure of size $O(n+h^2\cdot \log h \cdot 4^{\sqrt{\log h}})$ in $O(n+h^2\cdot \log^{2} h \cdot 4^{\sqrt{\log h}})$ time that answers each query in $O(\log n+k)$ time. Note that $n+h^2\cdot \log^{2} h \cdot 4^{\sqrt{\log h}}=O(n+h^{2+\epsilon})$ for any constant $\epsilon>0$. Further, we extend our techniques to the weighted rectilinear version in which the "obstacles" of $\mathcal{P}$ are rectilinear regions with "weights" and allow $L_1$ paths to travel through them with weighted costs. Our algorithm answers each query in $O(\log n+k)$ time with a data structure of size $O(n^2\cdot \log n\cdot 4^{\sqrt{\log n}})$ that is built in $O(n^2\cdot \log^{2} n\cdot 4^{\sqrt{\log n}})$ time (note that $n^2\cdot \log^{2} n\cdot 4^{\sqrt{\log n}}= O(n^{2+\epsilon})$ for any constant $\epsilon>0$).
  • We consider the problem of finding k centers for n weighted points on a real line. This (weighted) k-center problem was solved in O(n log n) time previously by using Cole's parametric search and other complicated approaches. In this paper, we present an easier O(n log n) time algorithm that avoids the parametric search, and in certain special cases our algorithm solves the problem in O(n) time. In addition, our techniques involve developing interesting data structures for processing queries that find a lowest point in the common intersection of a certain subset of half-planes. This subproblem is interesting in its own right and our solution for it may find other applications as well.
  • In the classic $k$-center problem, we are given a metric graph, and the objective is to open $k$ nodes as centers such that the maximum distance from any vertex to its closest center is minimized. In this paper, we consider two important generalizations of $k$-center, the matroid center problem and the knapsack center problem. Both problems are motivated by recent content distribution network applications. Our contributions can be summarized as follows: 1. We consider the matroid center problem in which the centers are required to form an independent set of a given matroid. We show this problem is NP-hard even on a line. We present a 3-approximation algorithm for the problem on general metrics. We also consider the outlier version of the problem where a given number of vertices can be excluded as the outliers from the solution. We present a 7-approximation for the outlier version. 2. We consider the (multi-)knapsack center problem in which the centers are required to satisfy one (or more) knapsack constraint(s). It is known that the knapsack center problem with a single knapsack constraint admits a 3-approximation. However, when there are at least two knapsack constraints, we show this problem is not approximable at all. To complement the hardness result, we present a polynomial time algorithm that gives a 3-approximate solution such that one knapsack constraint is satisfied and the others may be violated by at most a factor of $1+\epsilon$. We also obtain a 3-approximation for the outlier version that may violate the knapsack constraint by $1+\epsilon$.
  • Given a simple polygon P in the plane, we present new algorithms and data structures for computing the weak visibility polygon from any query line segment in P. We build a data structure in O(n) time and O(n) space that can compute the visibility polygon for any query line segment s in O(k log n) time, where k is the size of the visibility polygon of s and n is the number of vertices of P. Alternatively, we build a data structure in O(n^3) time and O(n^3) space that can compute the visibility polygon for any query line segment in O(k + log n) time.
  • Given a point $s$ and a set of $h$ pairwise disjoint polygonal obstacles of totally $n$ vertices in the plane, we present a new algorithm for building an $L_1$ shortest path map of size O(n) in $O(T)$ time and O(n) space such that for any query point $t$, the length of the $L_1$ shortest obstacle-avoiding path from $s$ to $t$ can be reported in $O(\log n)$ time and the actual shortest path can be found in additional time proportional to the number of edges of the path, where $T$ is the time for triangulating the free space. It is currently known that $T=O(n+h\log^{1+\epsilon}h)$ for an arbitrarily small constant $\epsilon>0$. If the triangulation can be done optimally (i.e., $T=O(n+h\log h)$), then our algorithm is optimal. Previously, the best algorithm computes such an $L_1$ shortest path map in $O(n\log n)$ time and O(n) space. Our techniques can be extended to obtain improved results for other related problems, e.g., computing the $L_1$ geodesic Voronoi diagram for a set of point sites in a polygonal domain, finding shortest paths with fixed orientations, finding approximate Euclidean shortest paths, etc.
  • Given $n$ points in a circular region $C$ in the plane, we study the problems of moving the $n$ points to its boundary to form a regular $n$-gon such that the maximum (min-max) or the sum (min-sum) of the Euclidean distances traveled by the points is minimized. The problems have applications, e.g., in mobile sensor barrier coverage of wireless sensor networks. The min-max problem further has two versions: the decision version and optimization version. For the min-max problem, we present an $O(n\log^2 n)$ time algorithm for the decision version and an $O(n\log^3 n)$ time algorithm for the optimization version. The previously best algorithms for the two problem versions take $O(n^{3.5})$ time and $O(n^{3.5}\log n)$ time, respectively. For the min-sum problem, we show that a special case with all points initially lying on the boundary of the circular region can be solved in $O(n^2)$ time, improving a previous $O(n^4)$ time solution. For the general min-sum problem, we present a 3-approximation $O(n^2)$ time algorithm, improving the previous $(1+\pi)$-approximation $O(n^2)$ time algorithm. A by-product of our techniques is an algorithm for dynamically maintaining the maximum matching of a circular convex bipartite graph; our algorithm can handle each vertex insertion or deletion on the graph in $O(\log^2 n)$ time. This result is interesting in its own right.
  • In this paper, we study the following problem of reconstructing a simple polygon: Given a cyclically ordered vertex sequence of an unknown simple polygon P of n vertices and, for each vertex v of P, the sequence of angles defined by all the visible vertices of v in P, reconstruct the polygon P (up to similarity). An O(n^3 log n) time algorithm has been proposed for this problem. We present an improved algorithm with running time O(n^2), based on new observations on the geometric structures of the problem. Since the input size (i.e., the total number of input visibility angles) is O(n^2) in the worst case, our algorithm is worst-case optimal.