The presented ParaView plugin allows easy access to the algorithm described in Ref 1. It enables analysis of faceted objects that exhibit distortions in their digital representation, e.g. due to tomographic reconstruction artifacts.
The contributed functionality can also be used outside ParaView in e.g. command-line programs. The code, data, a test and an example program are included.

Facet Analyser: ParaView plugin for automated facet detection and measurement of interplanar angles of tomographic objects

Nonlinear optimization is a key component of many image registration algorithms. Improving registration speed is almost always desirable. One way to do this is to accelerate the optimization cost function using a parallel implementation. The purpose of this document is to provide a tutorial on how to combine the CUDA GPU computing framework with standard nonlinear optimization libraries (VNL) using CMake. The provided code can be used as a starting template for programmers looking for a relatively painless introduction to CUDA-accelerated medical image registration and other nonlinear optimization problems.

A Tutorial on Combining Nonlinear Optimization with CUDA

Anisotropic Non-Linear Diffusion is a powerful image processing technique, which allows to simultaneously remove the noise and enhance sharp features in two or three dimensional images. Anisotropic Diffusion is understood here in the sense of Weickert, meaning that diffusion tensors are anisotropic and reflect the local orientation of image features. This is in contrast with the non-linear diffusion filter of Perona and Malik, which only involves scalar diffusion coefficients, in other words isotropic diffusion tensors. 
In this paper, we present an anisotropic non-linear diffusion technique we implemented in ITK. This technique is based on a recent adaptive scheme making the diffusion stable and requiring limited numerical resources. 

Anisotropic Diffusion in ITK

Although phase data can take on any value, it is generally only possible to measure phase as a principle value, i.e., wrapped within the range (-pi,pi]. Determining the unwrapped phase from its principle value is a topic of considerable interest in magnetic resonance imaging (MRI), as well as many non-medical disciplines. Despite their importance in image analysis, filters for manipulating phase information have not been incorporated into ITK. This article introduces the ITKPhase module, containing filters useful for understanding, analyzing, and unwrapping n-dimensional phase data, and also serves as a practical introduction to phase unwrapping.

N-Dimensional Phase Unwrapping

This document describes the derivation of the mixture models commonly used in the literature to describe the probabilistic nature of speckle: The Gaussian Mixture Model, the Rayleigh Mixture Model, the Gamma Mixture Model and the Generalized Gamma Mixture Model. New algorithms were implemented using the Insight Toolkit
ITK for tissue characterization by means of a mixture model. 
The source code is composed of a set of reusable ITK filters and classes. In addition to an overview of our implementation, we provide the source code, input data, parameters and output data that the authors used for validating the different probabilistic tissue characterization variants described in this paper. This adheres to the fundamental principle that scientific publications must facilitate reproducibility of the reported results.

Probabilistic Tissue Characterization for Ultrasound Images

The Fast Marching algorithm is an efficient numerical method for computing the distance and shortest path between points of a domain. For that purpose, it solves a front propagation problem, which can be of interest in itself. The method has numerous applications, ranging from motion planning to image segmentation. The unit of length, for computing the path length, may vary on the domain.
Motivated by applications, we generalize the algorithm to the case where the unit of length also depends on the path direction. Segmentation methods can take advantage of this flexibility to achieve greater sensitivity and specificity, for a comparable computation time.


Anisotropic Fast Marching in ITK

Isotropic volumetric acquisition in Magnetic Resonance Imaging (MRI) is often challenging. A large number of factors such as patient or physiological motion, signal-to-noise ratio (SNR), available static magnetic field B0 and total scanning
time limit the acquired resolution. Super-resolution (SR) is a post-processing tech-
nique that optimally combines several anisotropic scans into a single isotropic volume that was not - or could not be - acquired in practice. If necessary conditions are met, the resulting isotropic volume offers clear improvements over the initial acquisitions such as reduced partial volume effect, oblique visualization and improved sharpness and SNR. This paper details the misoSR implementation of a SR isotropicreconstruction algorithm using the Insight Toolkit Library (ITK) library. It is a generic implementation that reconstructs an isotropic volume from any number of anisotropic volumes acquired from any orientation. The algorithm takes advantage of the inputs header information to handle the different scans properties such as field of view (FOV), resolution parameters and orientation. Step by step details on the implementation are given, parameters are individually detailed, and results are shown on different applications as an example of SR reconstruction. The algorithm is hosted on the Creatis Virtual Imaging Platform (VIP), which allows users to run misoSR without having to install the software on their system.

MisoSR: Medical Image Isotropic Super-Resolution Reconstruction

Deformable (2D or 3D) medical image registration is a challenging problem. Existing approaches assume that the underlying deformation is smooth. This smoothness assumption allows for solving the deformable registration at a coarse resolution and interpolate for finer resolutions. However, sliding of organs and breathing motion, exhibit discontinuities. We propose a discrete optimization approach to preserve these discontinuities. Solving continuous deformations using discrete optimization requires a fine distribution of the discrete labels. Coupled with the typical size of medical image datasets, this poses challenges to compute solutions efficiently. In this paper we present a practical, multi-scale formulation. We describe how discontinuities can be preserved, and how the optimization problem is solved. Results on synthetic 2D, and real 3D data show that we can well approximate the smoothness of continuous optimization, while accurately maintaining discontinuities.

Deformable Registration with Discontinuity Preservation using Multi-Scale MRF

Diffuse Reflectance Spectroscopy (DRS) is a leading technique for the detection of head and neck cancers. It can capture information regarding tissue absorption and scattering. In this research work, we propose a novel method for the identification of normal and Squamous Cell Carcinoma (SCC) mucosa tissues using the Bag Of Words (BOW) approach. The study included 70 spectra from normal mucosa tissue sites and 70 spectra from SCC mucosa tissue sites. First, the spectra are preprocessed by extracting the useful wavelength range, denoising and reducing the inter and intra patient variability. Subsequently, features are extracted from each spectrum by continuously sliding a window with a predefined length along each spectrum to extract a group of local segments. Discrete Wavelet transform (DWT) is then employed for each segment. Next, we construct the codebook to represent each spectrum by a histogram of codewords at which each bin in the histogram is a count of a codeword appeared in the spectrum. Finally, the histogram representation is used as input for classification. The maximum accuracy reported is 94.28% with sensitivity and specificity of 91.42% and 97.14% respectively.

The Bag of Words Approach for the Classification of Head and Neck Cancers Using Diffuse Reflectance Spectroscopy

Transitive consistency of pairwise transformations is a desirable property of groupwise image registration procedures. The transformation synchronisation method (Bernard et al., 2015) is able to retrieve transitively consistent pairwise transformations from pairwise transformations that are initially not transitively consistent. In the present paper, we present a numerically stable implementation of the transformation synchronisation method for affine transformations, which can deal with very large translations, such as those occurring in medical images where the coordinate origins may be far away from each other. By using this method in conjunction with any pairwise (affine) image registration algorithm, a transitively consistent and unbiased groupwise image registration can be achieved.
Experiments involving the average template generation from 3D brain images demonstrate that the method is more robust with respect to outliers and achieves higher registration accuracy compared to reference-based registration.