Image Processing Routines Research Articles

Neural network based automated algorithm to identify joint locations on hand/wrist radiographs for arthritis assessment.

Arthritis is a significant and costly healthcare problem that requires objective and quantifiable methods to evaluate its progression. Here we describe software that can automatically determine the locations of seven joints in the proximal hand and wrist that demonstrate arthritic changes. These are the five carpometacarpal (CMC1, CMC2, CMC3, CMC4, CMC5), radiocarpal (RC), and the scaphocapitate (SC) joints. The algorithm was based on an artificial neural network (ANN) that was trained using independent sets of digitized hand radiographs and manually identified joint locations. The algorithm used landmarks determined automatically by software developed in our previous work as starting points. Other than requiring user input of the location of nonanatomical structures and the orientation of the hand on the film, the procedure was fully automated. The software was tested on two datasets: 50 digitized hand radiographs from patients participating in a large clinical study, and 60 from subjects participating in arthritis research studies and who had mild to moderate rheumatoid arthritis (RA). It was evaluated by a comparison to joint locations determined by a trained radiologist using manual tracing. The success rate for determining the CMC, RC, and SC joints was 87%-99%, for normal hands and 81%-99% for RA hands. This is a first step in performing an automated computer-aided assessment of wrist joints for arthritis progression. The software provides landmarks that will be used by subsequent image processing routines to analyze each joint individually for structural changes such as erosions and joint space narrowing.

Fibre-orientation measurements in short-glass-fibre composites. Part I: automated, high-angular-resolution measurement by confocal microscopy

In this first paper, the confocal technique for fibre-orientation distribution measurement is described and the associated errors discussed in detail. With this technique, fibres are followed for short distances into the bulk of the sample enabling a complete description of the fibre-orientation distribution tensor. Novel image processing and management routines have been implemented in order fully to automate the data acquisition, and consequently significant sample areas (several square mm) can be analysed within hours. In a follow-up paper the fibre-orientation measurements from the confocal technique will be used to evaluate independently the errors in the standard 2D image analysis technique of fibre-orientation measurement.

Three-dimensional quantification of color-marked occlusal paths on anatomically oriented casts

Statement of problem. The analysis of occlusal contacts on mounted diagnostic casts is an important task in prosthetic diagnostics. However, it is still restricted mainly to qualitative aspects because existing measuring techniques fail to provide 3-dimensional data for more than a few single points in acceptable time. Purpose. The aim of this study was to develop a method for quantifying occlusal contacts and paths by using anatomically oriented diagnostic casts and to design the method to allow fast acquisition of digital 3-dimensional coordinates. Material and methods. Plaster casts with color-marked contacts were digitized optically with a profilometric system based on the fringe projection technique. Digital camera images taken simultaneously were used to define the contact areas by marking them either manually or automatically with the help of image-processing routines. Three-dimensional coordinates were determined by finding the corresponding points within the profilometric data set. Results. Color-marked contact areas on diagnostic casts were successfully digitized with a computer-controlled, automatic setup in approximately 30seconds. The accuracy of the acquired 3-dimensional data was estimated to be better than 60 μm in lateral and 30 μm in height resolution. The data set was visualized and evaluated in a skullrelated coordinate system. Summary. This study verified the use of a new tool to quantify color-marked occlusal contacts on diagnostic casts in terms of spatial coordinates. The resulting digital data may be stored easily and analyzed numerically as well as visualized 3-dimensionally with computer graphic equipment. Because the anatomic orientation of the casts is maintained throughout the measurement process, it is possible to compare the data with electronically registered condyle paths and therefore to investigate, for example, their relation to the corresponding guidance paths of the frontal teeth and the canines in dynamic occlusion. (J Prosthet Dent 2001;85:156-61.)

Three-Dimensional Visualization and Measurement of Temperature Field Using Liquid Crystal Scanning Thermography

A Liquid Crystal Scanning Thermography (LCST) technique is developed and implemented to visualize and measure three-dimensional temperature fields. Results are reported for natural convection in a differentially heated vertical cavity. Experiments are conducted under steady state conditions at Rayleigh numbers of about 104 with glycerin as the working fluid. The scanning arrangement is described in detail together with the calibration scheme, and image processing routines that enable the processing of the thermographs and extraction of quantitative temperature information from them, including local and global heat transfer rates. The resultant temperature fields and Nusselt number values are computed and validated against both standard empirical relations and numerical predictions. To date, LCT has not been used to determine Nusselt number for enclosed fluid systems. The ability of LCST to perform this task makes this novel experimental technique a useful tool.

A CONSERVATION ASSESSMENT OF CENTRAL AMERICAN VEGETATION AND ECOREGIONS

A comprehensive,standardized,and thematically appropriate map of Central American vegetation and landcover types was developed by classifying the remote sensed imagery(AVHRR)1km2 resolution using advanced digital image processing routines and expertise.The map identifies 17 remaining natural vegetation types.The classification accuracy of the map is estimated to exceed 80%.Using a GAP analysis approach,a map of existing and proposed protected areas was overlaid on the vegetation map to analyze the protection status of vegetation types and Ecoregions.

Digital Radiography in Dentistry: What It Should Do for You

Digital radiology will become an important part of dental practice. Manufacturers should develop more sophisticated tools, including software for digital subtraction; image processing routines for the diagnosis of caries, periodontitis and periapical disease; tools for three-dimensional viewing of the teeth and supporting structures; and analysis of bone trabecular pattern for early detection of systemic disease. Hardware improvements should include increased dynamic range and sensitivity to radiation, and improved resolution. Sensors should be made the size of film, and components should be interchangeable across manufacturers. The true opportunity offered by digital imaging, computer-aided diagnosis, should continue to develop with particular attention to development of tools that add value for solving diagnostic problems and ease of use for the dentist and patient.

Automated measurement of single and total lung volume from CT.

The goal of this work was to develop an automated method for calculating single (SLV) and total (TLV) lung volumes from CT images. Patients underwent volumetric CT scanning through the entire chest in a single breath-hold, as well as pulmonary function tests. An automated, knowledge-based system was developed to segment the lungs in the CT images. Image-processing routines were used to extract sets of voxels from the image data that were identified by matching them to anatomical objects defined in a model. SLV and TLV were calculated by summing included voxels. For 43 patients analyzed, TLV from CT and total lung capacity from body plethysmography were strongly correlated (r = 0.90). On average, the CT-derived volume of the left lung accounted for 47.2% of the total. A knowledge-based approach to segmentation of the lungs in CT can be used to automatically estimate SLV and TLV.

Knowledge-based method for segmentation and analysis of lung boundaries in chest X-ray images

We present a knowledge-based approach to segmentation and analysis of the lung boundaries in chest X-rays. Image edges are matched to an anatomical model of the lung boundary using parametric features. A modular system architecture was developed which incorporates the model, image processing routines, an inference engine and a blackboard. Edges associated with the lung boundary are automatically identified and abnormal features are reported. In preliminary testing on 14 images for a set of 18 detectable abnormalities, the system showed a sensitivity of 88% and a specificity of 95% when compared with assessment by an experienced radiologist.

Edge detection in petrographic images using the rotating polarizer stage

A new edge detection procedure which calculates closed edges (grain boundaries) for images of petrographic thin sections is presented. Edges are extracted from a gradient image obtained using the rotating polarizer stage microscope. The gradient image is calculated by passing a simple gradient operator over captured images during the sampling process. A series of image processing routines are passed over the gradient image to extract the edges. The individual routines are: Gaussian smoothing, localized enhancement, nondirectional nonmaxima suppression, double thresholding, dilation, skeletonization and the removal of undesirable boundary fragments. The output of the procedure contains only closed boundaries.

Chapter 5 Introduction to Image Processing

Publisher Summary This chapter presents an introduction to image processing. Recent technological advances in computing and interfaces between video components and computers have allowed scientists to use processing routines to enhance or quantify particular characteristics of images. The use of image processing is not limited to scientists, and the introduction of digital cameras, video cards, and scanners has made image processing available to nearly every user of personal computers. With a minimum investment, one can readily enhance contrast, detect edges, quantify intensity, and apply a variety of mathematical operations to images. As image processing allows the investigator to convert the microscope/camera system into a quantitative device; the chapter focuses on three basic problems: (1) reducing “noise,” (2) enhancing contrast, and (3) quantifying intensity of an image. These techniques can then be applied to a number of different methodologies such as video-enhanced differential interference microscopy (VEDIC), nanovid microscopy, fluorescence recovery after photobleaching, and fluorescence ratio imaging. In all cases, knowledge of the basic principles of microscopy, image formation, and image-processing routines is absolutely required to convert the microscope into a device capable of pushing the limits of resolution and contrast.

A reappraisal of the use of infrared thermal image analysis in medicine.

Infrared thermal imaging of the skin has been used for several decades to monitor the temperature distribution of human skin. Abnormalities such as malignancies, inflammation, and infection cause localized increases in temperature which show as hot spots or as asymmetrical patterns in an infrared thermogram. Even though it is nonspecific, infrared thermology is a powerful detector of problems that affect a patient's physiology. While the use of infrared imaging is increasing in many industrial and security applications, it has declined in medicine probably because of the continued reliance on first generation cameras. The transfer of military technology for medical use has prompted this reappraisal of infrared thermology in medicine. Digital infrared cameras have much improved spatial and thermal resolutions, and libraries of image processing routines are available to analyze images captured both statically and dynamically. If thermographs are captured under controlled conditions, they may be interpreted readily to diagnose certain conditions and to monitor the reaction of a patient's physiology to thermal and other stresses. Some of the major areas where infrared thermography is being used successfully are neurology, vascular disorders, rheumatic diseases, tissue viability, oncology (especially breast cancer), dermatological disorders, neonatal, ophthalmology, and surgery.

Visual servoing and CAD-driven microassembly

The authors describe current research and development on a robotic visual servoing system for assembly of LIGA (lithography galvonoforming abforming) parts. The workcell consists of an AMTI robot, precision stage, long working distance microscope, and LIGA fabricated tweezers for picking up the parts. Fourier optics methods are used to generate synthetic microscope images from CAD drawings. These synthetic images are used off-line to test image processing routines under varying magnifications and depths of field. They also provide reference image features which are used to visually servo the part to the desired position. Currently, we can visually servo a 100 micron outside diameter LIGA gear to a desired x,y reference position as determined from a synthetic image of the gear.

A novel tool for rapid prototyping and development of simple 3D medical image processing applications on PCs

A 32-bit PC-based 3D medical image processing software package is presented. Its basic functions are the display and manipulation of medical images and the inclusion of user-written image processing routines in C language. This software runs on inexpensive hardware and is easy to learn. It meets the current needs of many research teams working in medical image processing.

Real-space characterization of short-range order in Cu-Pd alloys

Cu-Pd alloys containing 10, 20, 30, 40, and 50 at. % Pd and quenched from a temperature just above the ordering temperature ${\mathrm{T}}_{\mathrm{c}}$ are investigated by electron diffraction and high-resolution electron microscopy (HREM). The results show diffuse electron diffraction intensities at {100} and {110} positions for the alloy with 10 at. % Pd, but with a characteristic twofold and fourfold splitting for the alloys with more than 10 at. % Pd. High-resolution images show the formation of microdomains best developed between 20 and 30 at. % Pd. A real-space characterization has been performed by applying videographic real-structure simulations revealing that the splitting of the diffuse maxima depends on the average distance between microdomains of ${\mathrm{Cu}}_{3}$Au-type in antiphase with each other. By applying image processing routines on the HREM images, correlation vectors are identified which correspond to correlations between microdomains.

Image processing for improved eye-tracking accuracy.

Video cameras provide a simple, noninvasive method for monitoring a subject's eye movements. An important concept is that of the resolution of the system, which is the smallest eye movement that can be reliably detected. While hardware systems are available that estimate direction of gaze in real-time from a video image of the pupil, such systems must limit image processing to attain real-time performance and are limited to a resolution of about 10 arc minutes. Two ways to improve resolution are discussed. The first is to improve the image processing algorithms that are used to derive an estimate. Off-line analysis of the data can improve resolution by at least one order of magnitude for images of the pupil. A second avenue by which to improve resolution is to increase the optical gain of the imaging setup (i.e., the amount of image motion produced by a given eye rotation). Ophthalmoscopic imaging of retinal blood vessels provides increased optical gain and improved immunity to small head movements but requires a highly sensitive camera. The large number of images involved in a typical experiment imposes great demands on the storage, handling, and processing of data. A major bottleneck had been the real-time digitization and storage of large amounts of video imagery, but recent developments in video compression hardware have made this problem tractable at a reasonable cost. Images of both the retina and the pupil can be analyzed successfully using a basic toolbox of image-processing routines (filtering, correlation, thresholding, etc.), which are, for the most part, well suited to implementation on vectorizing supercomputers.

Quantitative Assessment of Fiber Fracture in Damaged Laminates Using X-Ray Radiography

In this paper, a new technique for extracting quantitative information about fiber fracture from dye penetrant enhanced X-ray radiographs is described. In this new inspection method, the observer marks fiber fracture zones on a digitized version of the radiograph depicting the damage state. Image processing routines are used to isolate the marked fiber fracture zones, and then extract fiber fracture parameters such as the length of the longest fiber fracture zone and the total length of fiber fracture from the marked image. The technique has been used to quantify fiber fracture in graphite epoxy laminates damaged by either quasi-static contact loading or low-velocity impact. The residual strengths of these specimens were found to correlate well with the fiber fracture parameters, with specimens containing more fiber fracture exhibiting lower residual strength.

Microtomographic Measurements of Internal Damage in Portland-Cement-Based Composites

The internal cracking patterns of a series of small mortar specimens were observed using a technique called X-ray microtomography. The experiments were arranged such that three-dimensional images of internal cracks could be made at several different times during the load history of the sample. This arrangement allowed progressive crack growth to be observed at micron-scale resolution in a single specimen. The specimens were loaded in compression. Image processing routines were used to make quantitative measurements of internal crack growth and damage progression. Preliminary measurements indicate that the maximum increase in crack area occurs around the peak load. Three-dimensional renderings qualitatively illustrate features of internal cracking such as the effects of different microstructural features.

Automatic localisation of electron backscattering pattern bands from Hough transform

An important and challenging aspect of the development of a system for fully automated electron backscattering pattern (EBSP) analysis is the design of reliable digital image processing routines for automatic detection and localisation of the bands in the digitised patterns. Procedures based on the Hough transform which permit precise localisation of a large number of bands in digital EBSPs are described in detail. Special attention is paid to the analysis and post-processing of the Hough space which ensure a high insensitivity to noise and a high precision of the measured band positions. The possibilities for further improvements and extensions of the procedures are also discussed, including the possibility of extracting band width information from the Hough transform. To demonstrate these possibilities, a new procedure is outlined and shown to be capable of providing high precision estimates of the EBSP band positions and, in addition, fairly precise estimates of the band widths.MST/3672

Automatic localisation of electron backscattering pattern bands from Hough transform

An important and challenging aspect of the development of a system for fully automated electron backscattering pattern (EBSP) analysis is the design of reliable digital image processing routines for automatic detection and localisation of the bands in the digitised patterns. Procedures based on the Hough transform which permit precise localisation of a large number of bands in digital EBSPs are described in detail. Special attention is paid to the analysis and post-processing of the Hough space which ensure a high insensitivity to noise and a high precision of the measured band positions. The possibilities for further improvements and extensions of the procedures are also discussed, including the possibility of extracting band width information from the Hough transform. To demonstrate these possibilities, a new procedure is outlined and shown to be capable of providing high precision estimates of the EBSP band positions and, in addition, fairly precise estimates of the band widths.MST/3672

Computation reduction of the maximum likelihood classifier using the Winograd identity

The maximum likelihood classifier is one of the most used image processing routines in remote sensing. However, most implementations have exhibited the so-called “Hughes phenomenon” and the computation cost increases quickly as the dimensionality of the feature set increases. Based on the above reasons, the recursive maximum likelihood classification strategy is more suitable for hyperspectral imaging data than the conventional nonrecursive approach. In this paper we derive some computation aspects of quadratic forms by applying the Winograd's method to three previous approaches. The new, modified approaches are approximately four times faster than the conventional nonrecursive approach and two times faster than the existing recursive algorithms.