Sajjad Baloch

STEP: Spatiotemporal enhancement pattern for MR-based breast tumor diagnosis

The authors propose a spatiotemporal enhancement pattern (STEP) for comprehensive characterization of breast tumors in contrast-enhanced MR images. By viewing serial contrast-enhanced MR images as a single spatiotemporal image, they formulate the STEP as a combination of (1) dynamic enhancement and architectural features of a tumor, and (2) the spatial variations of pixelwise temporal enhancements. Although the latter has been widely used by radiologists for diagnostic purposes, it has rarely been employed for computer-aided diagnosis. This article presents two major contributions. First, the STEP features are introduced to capture temporal enhancement and its spatial variations. This is essentially carried out through the Fourier transformation and pharmacokinetic modeling of various temporal enhancement features, followed by the calculation of moment invariants and Gabor texture features. Second, for effectively extracting the STEP features from tumors, we develop a graph-cut based segmentation algorithm that aims at refining coarse manual segmentations of tumors. The STEP features are assessed through their diagnostic performance for differentiating between benign and malignant tumors using a linear classifier (along with a simple ranking-based feature selection) in a leave-one-out cross-validation setting. The experimental results for the proposed features exhibit superior performance, when compared to the existing approaches, with the area under the ROC curve approaching 0.97.

Rotation invariant topology coding of 2D and 3D objects using Morse theory

In this paper, we propose a numerical algorithm for extracting the topology of a three-dimensional object (2 dimensional surface) embedded in a three-dimensional space /spl Ropf//sup 3/. The method is based on capturing the topology of a modified Reeb graph by tracking the critical points of a distance function. As such, the approach employs Morse theory in the study of translation, rotation, and scale invariant skeletal graphs. The latter are useful in the representation and classification of objects in /spl Ropf//sup 3/.

Object Recognition Through Topo-Geometric Shape Models Using Error-Tolerant Subgraph Isomorphisms

We propose a method for 3-D shape recognition based on inexact subgraph isomorphisms, by extracting topological and geometric properties of a shape in the form of a shape model, referred to as topo-geometric shape model (TGSM). In a nutshell, TGSM captures topological information through a rigid transformation invariant skeletal graph that is constructed in a Morse theoretic framework with distance function as the Morse function. Geometric information is then retained by analyzing the geometric profile as viewed through the distance function. Modeling the geometric profile through elastic yields a weighted skeletal representation, which leads to a complete shape signature. Shape recognition is carried out through inexact subgraph isomorphisms by determining a sequence of graph edit operations on model graphs to establish subgraph isomorphisms with a test graph. Test graph is recognized as a shape that yields the largest subgraph isomorphism with minimal cost of edit operations. In this paper, we propose various cost assignments for graph edit operations for error correction that takes into account any shape variations arising from noise and measurement errors.

Automatic detection of anatomical features on 3D ear impressions for canonical representation

We propose a shape descriptor for 3D ear impressions, derived from a comprehensive set of anatomical features. Motivated by hearing aid (HA) manufacturing, the selection of the anatomical features is carried out according to their uniqueness and importance in HA design. This leads to a canonical ear signature that is highly distinctive and potentially well suited for classification. First, the anatomical features are characterized into generic topological and geometric features, namely concavities, elbows, ridges, peaks, and bumps on the surface of the ear. Fast and robust algorithms are then developed for their detection. This indirect approach ensures the generality of the algorithms with potential applications in biomedicine, biometrics, and reverse engineering.

Robust independent component analysis

Independent component analysis (ICA) attempts to separate independent components present in the mixture signals. Several criteria have been suggested for ICA in the past, including kurtosis and negentropy. Kurtosis suffers from a drawback of being outlier sensitive. As a remedy, we propose robust ICA (RICA), which employs appropriate robust estimators. In this paper, we compare the robustness properties of RICA with kurtosis- and negentropy-based ICA. Since robust estimators are insensitive to outliers in contrast to maximum likelihood estimates (MLE), we demonstrate that in the presence of outliers, RICA works better than kurtosis- and negentropy-based ICA

Layout consistent segmentation of 3-D meshes via conditional random fields and spatial ordering constraints

We address the problem of 3-D Mesh segmentation for categories of objects with known part structure. Part labels are derived from a semantic interpretation of non-overlapping subsurfaces. Our approach models the label distribution using a Conditional Random Field (CRF) that imposes constraints on the relative spatial arrangement of neighboring labels, thereby ensuring semantic consistency. To this end, each label variable is associated with a rich shape descriptor that is intrinsic to the surface. Randomized decision trees and cross validation are employed for learning the model, which is eventually applied using graph cuts. The method is flexible enough for segmenting even geometrically less structured regions and is robust to local and global shape variations.

Toward automation in hearing aid design

In the manufacturing of customized medical prostheses, such as in-the-ear hearing aids, the design process often is dictated by a source template representing the anatomy of a patient and a set of work instructions representing the description of surface modifications. Instead of carrying out the work instructions by hand with knife, file or drilling tools, the state-of-the-art relies on modern software tools, such as computer-aided-design and computer-aided-manufacturing. Work instructions are usually defined in terms of anatomical landmarks of a given template. Following the design phase, the virtual model of the customized prosthesis is produced by a rapid prototyping system, like selective laser sintering or stereolithography. An outstanding problem in prostheses design is that the work instructions are often vaguely defined, and a suitable outcome largely depends on the knowledge, experience and skill of the designer. In this paper, we present a solution to minimize the influence of human interaction. Our approach involves the abstraction of the work instructions into expert system rules that exploit a robustly identified canonical set of anatomic features. The versatility of our approach lies in a priori defining an entire design workflow through a rule set, thereby yielding a high degree of automation that is flexible, customizable, consistent, and reproducible. The proposed solution is extensively evaluated in a real world application, and is shown to yield significant improvement in manufacturing. For instance, the consistency of the outcome was improved by about 10% and the design time was reduced by about 8.4%.

Freeform shape clustering for customized design automation

Automation can provide significant performance improvements in digital manufacturing systems that customize shapes of implants and prosthetic devices to the anatomy of a patient. The challenge, however, lies in the ability of an automatic solution to adapt to anatomical variations of a given object category. This paper presents a hierarchical framework that generalizes the digital design of anatomical surface models in terms of a small number of prototypes. The latter are derived from the local shape information of constituent parts via shape matching and clustering and then associated with one operation that dictates how a shape undergoes modification. We demonstrate the proposed technique through application to typical hearing aid design operations with promising results.

Deformable registration of organic shapes via surface intrinsic integrals: application to outer ear surfaces

We propose a method for the deformable registration of organic surfaces. Meaningful correspondences between a source surface and a target surface are established by means of a rich surface descriptor that incorporates three categories of features: (1) local and regional geometry; (2) surface anatomy; and (3) global shape information. First, surface intrinsic, geodesic distance integrals, are exploited to constrain the global geodesic layout. Consequently, the resulting transformation ensures topological consistency. Local geometric features are then introduced to enforce local conformity of various regions. To this end, the extrema of appropriate curvatures -- the extrema of mean curvature, minima of Gauss and minimum principal curvature, and the maxima of maximum principal curvature -- are considered. Regional features are introduced through curvature integrals over various scales. On top of this, explicit anatomical priors are included, thereby resulting in anatomically more consistent registration. The source surface is deformed to the target by minimizing the energy of matching the source features to the target features under a Gaussian propagation model. We validate the proposed method with application to the outer ear surfaces.

2D Shape Modeling using Skeletal Graphs in a Morse Theoretic Framework

Topology and geometry are the attributes that uniquely define a shape. Two objects are said to have the same topological structure if one can be morphed into the other without tearing and gluing, whereas geometry describes the relative position of points on a surface. Existing shape descriptors pay little attention to the topology of shapes and instead operate on a smaller subset, where all shapes are assumed to have a genus of one. In this chapter, we will describe a novel 2D shape modeling method that keeps track of the topology of a shape in combination with its geometry for a robust shape representation. Using a Morse theoretic approach and the 3D shape modeling technique in [2] as an inspiration, we focus on representing planar shapes of arbitrary topology. The proposed approach extends existing modeling techniques in the sense that it encompasses a larger class of shapes.