Interactive algorithms for the segmentation and quantitation of 3-D MRI brain scans

Interactive algorithms are an attractive approach to the accurate segmentation of 3D brain scans as they potentially improve the reliability of fully automated segmentation while avoiding the labour intensiveness and inaccuracies of manual segmentation. We present a 3D image analysis package (MIDAS) with a novel architecture enabling highly interactive segmentation algorithms to be implemented as add on modules. Interactive methods based on intensity thresholding, region growing and the constrained application of morphological operators are also presented. The methods involve the application of constraints and freedoms on the algorithms coupled with real time visualisation of the effect. This methodology has been applied to the segmentation, visualisation and measurement of the whole brain and a small irregular neuroanatomical structure, the hippocampus. We demonstrate reproducible and anatomically accurate segmentations of these structures. The efficacy of one method in measuring volume loss (atrophy) of the hippocampus in Alzheimer's disease is shown and is compared to conventional methods.

Investigation of acoustic noise on 15 MRI scanners from 0.2 T to 3 T

Acoustic noise levels for fast MRI pulse sequences were surveyed on 14 systems with field strengths ranging from 0.2 T to 3 T. A microphone insensitive to the magnetic environment was placed close to the magnet isocenter and connected via an extension cable to a sound level meter outside the scan room. Measured noise levels varied from 82.5 +/- 0.1 dB(A) for a 0.23 T system to 118.4 +/- 1.3 dB(A) for a 3 T system. Further measurements on four of the closed-bore systems surveyed showed that: 1) pulse sequence parameters (particularly FOV and TR) were more influential in determining noise level than field strength, 2) the noise level was found to vary along the z-direction with a maximum near the bore entrance, and 3) in one of two systems tested there was a significant increase in noise with a volunteer present instead of a test object. The results underline the importance of hearing protection for patients and for staff spending extended periods in the scan room.

Biomedical signal processing (in four parts). Part 1. Time-domain methods

This is the first of a series of four tutorial papers on biomedical signal processing. It provides an introduction to terminology and basic ideas for testing for randomness and trend, and for the determination of basic signal properties in the time domain, given the uncertainties associated with the estimation process. Techniques outlined in the paper are: the coherent average, cross-correlation and covariance, autocorrelation and phase-shift averaging.

Biomedical signal processing (in four parts). Part 3. The power spectrum and coherence function

This is the third in a series of four tutorial papers on biomedical signal processing and concerns the estimation of the power spectrum (PS) and coherence function (CF) od biomedical data. The PS is introduced and its estimation by means of the discrete Fourier transform is considered in terms of the problem of resolution in the frequency domain. The periodogram is introduced and its variance, bias and the effects of windowing and smoothing are considered. The use of the autocovariance function as a stage in power spectral estimation is described and the effects of windows in the autocorrelation domain are compared with the related effects of windows in the original time domain. The concept of coherence is introduced and the many ways in which coherence functions might be estimated are considered.

Transient interactions between blood pressure, respiration and heart rate in man

Auto regressive spectral estimation techniques have been used to follow transient interactions between mean blood pressure, respiration and heart rate. This demonstrates that these inter-relationships are variable. It is concluded that while central modulation of heart rate is the major factor in the interactions, when the heart rate is fixed, peripheral modulation of the blood pressure by respiration is clearly demonstrated.

Neonatal heart rate variability and its relation to respiration

The heart rate and respiration signals from nine healthy full term neonates were studied using autoregressive spectral analysis and cross-correlation techniques. The heart rate spectra could be divided into three regions of activity: a very low frequency (VLF) region from 0-0.04 Hz; a low frequency (LF) band from 0.04-0.20 Hz; and a high frequency (HF) region above 0.20 Hz. The newborns exhibited very little respiratory sinus arrhythmia in their heart rate variability in contrast to the situation for adults and older infants. However, variations in heart rate correlated strongly with changes in the breath amplitude, leading to what may be termed a breath amplitude sinus arrhythmia. The neonatal heart rate behaviour under stable conditions of oscillation could be simulated with a nonlinear control model provided the delay time in the baroreceptor loop of the model was taken to be approximately 2 seconds longer than in adults. This is consistent with the immature neurological status of neonates.

3-D visualization of arterial structures using ultrasound and Voxel modelling

In this paper, a new type of vascular imaging system is presented which is designed for use in conjunction with percutaneous transluminal treatment techniques (balloon and laser angioplasty, atherectomy etc). Three dimensional computer models of arterial sections are reconstructed in full voxel space from data acquired using a purpose-built, catheter-mounted ultrasound probe. The system is standalone, using commercially available computer hardware and specially written software. The software is equally compatible with source data from other modalities (e.g. CT and MR), and the system can therefore be incorporated into a PACS environment.

Breath amplitude modulation of heart rate variability in normal full term neonates

The relationship between heart rate variability and respiration patterns was investigated using spectral analysis techniques in nine full-term infants whose ages ranged from 39-75 h. All the infants were studied during sleep, although no attempt was made to classify rapid eye movement or nonrapid eye movement states prospectively. The data obtained were examined to determine which aspects of neonatal breathing patterns are correlated with heart rate variability. Three spectral regions of heart rate variability could be identified: a very low frequency region below 0.02 Hz; a low frequency region from 0.02-0.20 Hz; and a high frequency region above 0.20 Hz. The dominant heart rate variability activity in these neonates was seen in the very low and low frequency regions, with little activity in the high frequency regions. In contrast to older infants and adults, respiration and heart rate variability were not strongly related through a high frequency region respiratory sinus arrhythmia but rather through a breath amplitude sinus arrhythmia which occurs in the low frequency region of the spectrum. The prominent very low frequency activity and the low frequency activity ascribed to breath amplitude modulation may result from autonomic nervous system mediation of chemoregulation.

Biomedical signal processing (in four parts). Part 2. The frequency transforms and their inter-relationships

This is the second in a series of four tutorial papers on biomedical signal processing, and it concerns the relationships between commonly used frequency transforms. It begins with the Fourier series and Fourier transform for continuous time signals and extends these concepts for aperiodic discrete time data and then periodic discrete time data. The Laplace transform is discussed as an extension of the Fourier transform. The z-transform is introduced and the ideas behind the chirp-z transform are described. The equivalence between the time and frequency domains is described in terms of Parseval's theorem and the theory of convolution. The use of the FFT for fast convolution and fast correlation is described for both short recordings and long recordings that must be processed in sections.

A nonlinear model for studying oscillations in the blood pressure control system

The idea that physiological systems sometimes include oscillations as part of their control function is a relatively new idea; the paper discusses the role of such oscillations in the regulation of arterial blood pressure and how they can be studied by modelling techniques. A nonlinear model is proposed which can be related directly to the physiology. This is seen as an important step because mainly previous studies of biological oscillations have used lumped equations, e.g. the Van der Pol equation, which do not possess the property. On the basis of the model, the oscillatory behaviour of the blood pressure control system is analysed by the application of the dual input describing function technique, a powerful analytical method which has wide application to the study of nonlinear oscillatory phenomena in physiology. The final section of the paper considers the digital simulation of the model by Z transform techniques and how the simulation can be used to study the interaction of respiratory and blood pressure control systems.

Real-time heart rate variability extraction using the Kaiser window

A new method for real-time heart rate variability (HRV) detection from the R-wave signal, based on the integral pulse frequency modulation (IPFM) model and its similarity to pulse position modulation, is presented. The proposed method exerts low-pass filtering with a Kaiser window. It can also be used for off-line HRV analysis in both the time and frequency domains. Real-time bandpass filtering as a new HRV investigation method and as a by-product of the proposed algorithm is also introduced. Furthermore, the discrete time domain version of the French-Holden algorithm is developed, and it is thoroughly proved that low-pass filtering is an ideal method for detection of HRV.

Dynamic measurement of human capillary blood pressure

1. Capillary blood pressure was measured in man using a dynamic servo-nulling system and direct micropuncture. This enabled assessments of the normal variations in pressure which influence fluid filtration and reabsorption. 2. Seventy-eight capillaries in 19 subjects were punctured in one of three positions around the capillary loop with the hand at the level of the sternal angle. Mean pressure around the loop fell from 37.7 +/- 3.7 mmHg (arteriolar limb, mean +/- SEM, n = 12) to 19.4 +/- 1.0 mmHg (apex, n = 25) to 14.6 +/- 0.5 mmHg (venular limb, n = 41) at skin temperatures of 18.7-33.1 degrees C. These values agree closely with Landis' original studies in 1930 [E. Landis (1930) Heart, 15, 209-228]. 3. The mean filtration/reabsorption state of any particular capillary limb was not static because of cardiac, vasomotor and respiratory fluctuations in capillary pressure. From a total of 38 capillaries in which recordings were analysed for 30 s, the fluctuations in pressure were such that 27 capillaries probably had periods of both filtration and reabsorption. 4. Computerized superimposition and coherent averaging of trains of capillary pulses enabled an accurate description of the pulse waveform to be made in three capillaries. This was remarkably similar to waveforms from the radial artery, albeit at reduced amplitude (average 3.6 +/- 3.4 mmHg, mean +/- SD overall). The time for the pulse to travel between the radial artery and the finger capillary was approximately 10 ms, which implies a propagation velocity of several metres per second.(ABSTRACT TRUNCATED AT 250 WORDS)

Transient phase locking patterns among respiration, heart rate and blood pressure during cardiorespiratory synchronisation in humans

The interactions between respiration, heart rate and blood pressure variability (HRV, BPV), are considered to be of paramount importance for the study of the functional organisation of the autonomic nervous system (ANS). The aim of the reported study is to detect and classify the intermittent phase locking (PL) phenomena between respiration, HRV and BPV during cardiorespiratory synchronisation experiments, by using the following time-domain techniques: Poincaré maps, recurrence plots, time-space separation plots and frequency tracking locus. The experimental protocol consists of three stages, with normal subjects in paced breathing at 15, 12 and 8 breaths min-1. Transient phenomena of coordination between respiration and the major rhythms of HRV and BPV (low and high frequency, LF and HF) have been detected and classified: no interaction between LF and HF rhythms at 15 breaths min-1; short time intervals of stable 1:2 frequency and phase synchronisation during the 12 breaths min-1 stage; 1:1 PL during the 8 breaths min-1 stage. 1:1 and 1:2 PL phenomena occurred when the respiration frequency was quite close to the LF frequency or when it was about twice the LF frequency, respectively. The complex organisation of the ANS seems to provoke transient rather than permanent PL phenomena between the co-ordinating components of respiration and cardiovascular variability series.

Synthetic biology in the UK - An outline of plans and progress

Synthetic biology is capable of delivering new solutions to key challenges spanning the bioeconomy, both nationally and internationally. Recognising this significant potential and the associated need to facilitate its translation and commercialisation the UK government commissioned the production of a national Synthetic Biology Roadmap in 2011, and subsequently provided crucial support to assist its implementation. Critical infrastructural investments have been made, and important strides made towards the development of an effectively connected community of practitioners and interest groups. A number of Synthetic Biology Research Centres, DNA Synthesis Foundries, a Centre for Doctoral Training, and an Innovation Knowledge Centre have been established, creating a nationally distributed and integrated network of complementary facilities and expertise. The UK Synthetic Biology Leadership Council published a UK Synthetic Biology Strategic Plan in 2016, increasing focus on the processes of translation and commercialisation. Over 50 start-ups, SMEs and larger companies are actively engaged in synthetic biology in the UK, and inward investments are starting to flow. Together these initiatives provide an important foundation for stimulating innovation, actively contributing to international research and development partnerships, and helping deliver useful benefits from synthetic biology in response to local and global needs and challenges.

The design of digital filters for biomedical signal processing. Part 3: The design of Butterworth and Chebychev filters

The first two papers in this series reviewed the basic concepts which apply to digital filter theory and presented design techniques based on the z plane pole-zero plot. In this paper these methods are used to develop digital versions of Butterworth and Chebychev filters. The basic theory of both filter types is reviewed and the bilinear transformation is used to derive the z-transforms of the filters from their s-plane continuous time descriptions. Recurrence relationships which may be used to implement filters of various orders are developed. The impulse and frequency responses of the elements are illustrated and examples are given of their application to ECG data.

An MRI derived articular cartilage visualization framework

OBJECTIVE

We present a multi-dimensional framework for the visualization of femoral articular cartilage. The framework comprises methods for visualizing and quantifying changes in cartilage thickness and surface morphology derived from MRI based cartilage segmentation. Adequate visualization of cartilage allows accurate and clinically meaningful assessment of cartilage surface morphology and thickness. In current practice the routine use of conventional 2D MR images provides limited qualitative information and is inconvenient because the imaged volume has to be reviewed slice by slice.

METHOD

A Graphical User Interface (GUI) that encapsulates the framework described above was developed. In the first stage of the analysis MR images of the knee are segmented to delineate cartilage boundaries. Cartilage thicknesses are subsequently measured. The detected points and corresponding thickness data are utilized to produce a visualization framework.

RESULTS

The system was tested using data from six example patients. The spatial distribution of cartilage on the articular surface was visualized using a 3D WearMap. The 2D WearMap allowed the entire cartilage surface to be studied at once. Quantitative interaction with the 2D WearMap was assisted by the ability to ascertain cartilage surface dimensions and TrackBack from a point of interest to the original MR image. As a result, the detection of wear patterns and lesions was efficiently carried out.

CONCLUSION

A means of quantitatively visualizing cartilage defects non-invasively is presented. This stands to reduce clinician reporting times, as well as allowing quantitative follow-up that facilitates osteoarthritis (OA) screening and planning/evaluating interventions.

The analysis of blood velocity measurements by autoregressive modelling

Analysis of velocity disturbances arising in poststenotic flow is examined using the technique of autoregressive modeling. The essential elements of this method are described including relevant criteria for selecting the model order. Velocity data employed in the analysis are taken from in vivo measurements in the dog aorta, and the results indicate that the autoregressive method improves the resolution of coherent features in disturbed flow patterns. By applying homomorphic filtering to individual beats, the occurrence of organized structures convected from their origin in the shear layer is readily identified.

Intravascular ultrasound imaging and three-dimensional modeling of arteries

In this article, we describe the reconstruction of arterial structures using solid modeling. The alternative approaches to three-dimensional modeling are discussed and the voxel space system we use for intra-arterial imaging, based on ultrasonic data, is described. The complete three-dimensional ultrasonic imaging system comprising a purpose-built, catheter-mounted ultrasound probe, ultrasonic transceiver, and computer system is presented. This system has been used to recreate three-dimensional computer models of arterial sections in vitro and in vivo. Examples to illustrate the power and flexibility of voxel space modeling in terms of postprocessing and software manipulation are given. Preliminary work on tissue differentiation, using arterial models and color coding of the image, and three-dimensional presentation of flow data is included.

Miniature ultrasonic probe construction for minimal access surgery

The increasing use of minimal access techniques for surgery has produced a need for imaging technologies that can be used during such interventions. Ultrasound imaging has the advantage that the probe itself can be interventional. Interventional ultrasound probes must be sufficiently small to gain access to the surgical site, and any rigid portion must be limited in length to permit adequate flexibility. In practice this means the ultrasound probes have to operate at high frequencies, and a set of design curves have been produced which relate the number of elements and the ultrasound frequency to the probe dimensions for both linear and cylindrical array configurations. Constructing high-frequency sub-miniature probes presents a number of technical challenges, in particular relating to interconnects and packaging. Solutions to these challenges are discussed using the fabrication of a 1 mm diameter intravascular probe as an example.