Functional Ultrasound Imaging Methods for Image-Guided Thermal Surgery

Emad S. Ebbini, Ph.D. ,    Eelectrical and Computer Engineering Department, University of Minnesota, U.S.A.

Ultrasound imaging is routinely used in guidance of RF ablation, cryosurgery, and other forms of minimally invasive thermal surgery. In addition, clinical trials of high intensity focused ultrasound (HIFU) in the treatment of uterine fibroids and prostate cancer are underway in the US, Europe, and Japan. HIFU can be applied non-invasively under guidance of MRI or ultrasound. Ultrasound is attractive due to its portability, cost-effectiveness, and real-time nature. However, ultrasounds low soft-tissue contrast has been a major obstacle preventing ultrasound-guided thermal surgery from widespread clinical practice. Quantitative and functional imaging methods are essential if ultrasound-guided thermal surgery is to continue to be a viable option in modern surgery.

We recently have developed three imaging methods suitable for image guidance of thermal surgery using ultrasound: 1) noninvasive temperature imaging for monitoring and control of tissue temperature during surgery, 2) imaging of blood perfusion defects for assessment of tissue necrosis, and 3) imaging of tissue stiffness for assessment of coagulation. We will show that these three imaging methods allow the ultrasound imaging guidance to evaluate the heating source at the target tissue before treatment, monitor tissue response during treatment, and provide assessment of irreversible tissue damage due to thermal coagulation after treatment. In all of these methods, signal processing of the raw radio frequency (RF) echo data from standard diagnostic scanners is used. Fine tissue displacements are estimated at high frame rates are estimated using speckle tracking methods to produce temperature and stiffness images. Furthermore, RF data is filtered with appropriately designed Volterra filters to separate linear and nonlinear echo components from the treated region. Tissue perfusion images are formed from the quadratic and significant higher-order components from the Volterra filter.

 In this paper, we present the mathematical bases leading to the imaging algorithms and requisite signal processing.  In vivo imaging results using a high frequency ultrasound in a small animal model demonstrating the feasibility of the new imaging methods will be presented. New opportunities and outstanding challenges in the fast growing field of image-guided thermal surgery will be discussed in light of presented data and results from other research groups working on other image-guidance modalities, e.g. MRI.

Novel Stochastic Models for Medical Image analysis

Aly A. Farag, Ph.D., Department of Electrical and Computer Engineering, University of Louisville,  Louisville, Kentucky, U.S.A.

Image models attempt to capture the visual characteristics of the images in a few parameters so as to understand the nature of the phenomenon generating the images. Models provide a quantitative methodology to represent and analyze the information content of the images; and various modeling approaches have been introduced in the past three decades.  In this talk, novel approaches for modeling the intensity distribution of the gray levels and the spatial interaction between the pixels in the observed image will be presented. In particular, we will introduce a generalized linear model for representing the joint probability density function of the class of objects in an image or volume. Most importantly, we present a novel approach for model identification and estimation of the marginal probability densities of the dominant classes/modes in the image. All parameters in the model are estimated on-line; thus an automatic approach for statistical inference is possible. We illustrate the power of this approach on a number of image analysis algorithms.

The second part of the talk will deal with image analysis of low dose chest CT scanning (LDCT) in order to create an automatic computer-assisted diagnosis (CAD) approach for early detection of lung cancer. There exist a number of LDCT screening studies in the US and abroad and the University of Louisville is among the active centers in world in lung cancer research.  During the past five years the CVIP Lab has developed and validated a comprehensive approach for lung nodule detection which is the backbone of image-based CAD models. This talk will describe the UofL Lung Nodule CAD system. In particular, special focus will be on the algorithms and validation of the nodule detection approach. The talk will also address few aspects of lung cancer as a major healthcare issue in the US and elsewhere.

The talk will illustrate the multidisciplinary nature of biomedical imaging and that significant progress in image-guided interventions, cancer diagnosis and treatment, and other image-related research require a close collaboration between biomedical engineers, physicians, life scientists and technologists.

Understanding and Approaching the Limits of Functional Magnetic Resonance Imaging

 Essa Yacoub, Center for Magnetic Resonance Research, Department of Radiology, University of Minnesota, U.S.A.

Since its introduction, functional magnetic resonance imaging (fMRI) has evolved into the most commonly used methodology for mapping brain function, particularly in humans. This has resulted because it is non-invasive and offers an unmatched combination of spatial and temporal resolution that continues to be pushed to new levels. The primary technique used in fMRI is the blood oxygen level dependent (BOLD) signal, which is a measure of the regional paramagnetic deoxyhemoglobin content in the brain. BOLD fMRI relies on the presence of paramagnetic deoxyhemoglobin sequestered in blood vessels, which in turn depends on the coupling between cerebral blood flow (CBF), blood volume (CBV), oxygen consumption (CMRO2), and neural activity. Non-commensurate changes in CBF, CBV, and/or CMRO2 will alter the local deoxyhemoglobin concentration, and subsequently the local transverse relaxation rates.  CBF increases following neural activity more than compensate for any increases in CMRO2 or CBV, resulting in an overall decrease in the local deoxyhemoglobin content and, consequently, increased signal intensity in T2 or T2* weighted MR images. Manipulation of acquisition strategies (i.e. magnetic field strength, preparation pulses, refocusing pulses, delay times, etc.) can significantly affect the observed BOLD signals in terms of vascular specificity as well as sensitivity. Because of its ease of implementation and high contrast to noise ratio (CNR), the most commonly used technique relies on the detection of the delayed positive BOLD signal change, which is associated with the CBF increases following neural activity, using gradient echo (GE) or T2* weighted imaging. In numerous studies, however, this approach was demonstrated to be highly sensitive to large draining veins especially at lower fields. In contrast to GE BOLD approaches, Hahn Spin Echo (HSE) based contrast preparation in fMRI is expected to have significantly better accuracy, specifically at high magnetic fields. This is ascribed to the increased sensitivity of HSE BOLD contrast to the microvasculature. Static B0 inhomogeneities around large vessels are refocused via radio frequency pulses and blood signals are effectively suppressed at high magnetic fields due to the short T2 of venous blood. Both GE and HSE BOLD images were however both shown to have increased specificity and sensitivity when going to higher field strengths such as 7 Tesla. For the vast majority of routine fMRI studies specificity and sensitivity are generally not terribly critical since the neuroscience questions being asked do not require high spatial resolutions. These studies aim to map more coarse functional processes. Recently, however, the limits of BOLD fMRI have been challenged to the level of sub-millimeter functional structures. The reliability and reproducibility of these maps, as well as the optimization of the fMRI methods to acquire these maps has not yet been established, especially in humans. Differences in the spatial specificity to neural activity between GE and HSE BOLD fMRI signals at high fields are pertinent to experimental design considerations at high magnetic fields. In addition, the intrinsic limits of the hemodynamic mapping signals observed via fMRI are also critical in understanding brain function at fundamental levels. Optimized vascular signals to map brain function could allow for mapping of functional architecture which is not yet known. This would be an invaluable tool for many neuroscience applications because many columnar organizations remain undiscovered. The primary aim of this research is to understand the BOLD mechanisms and their intrinsically imposed limits on the BOLD signal as a functional mapping tool while thereby developing strategies for addressing high spatial resolution questions.

Multi-Modality Image Analysis for Epilepsy

Hamid Soltanianzadeh, Ph.D., Henry Ford Health System, Michigan, U.S.A. and the University of Tehran, Iran

One in about every 200 individuals suffers from a neurological disease referred to as “epilepsy.”  Two-thirds of all epileptic patients have a specific focal area of seizure onset within the brain.  More than 20% of the epileptic patients undergo surgery when treatment with medication becomes ineffective.  The conventional method for determining the focal area of the seizure onset and consequently evaluating the patient for surgical candidacy is lengthy, painful, and costly.  It requires electroencephalography (EEG) exams to detect “irritative zones.”  A phase one EEG exam requires admittance to the hospital for a period of five to seven days. During this hospital stay, the patient undergoes 24 hour video monitoring and EEG recording and analysis (with electrodes placed at several sites on the head).  If the epileptic focus is not sufficiently localized in phase one, the patient will undergo phase two study which involves implantation of electrodes intracranially and monitoring the patient for nearly two weeks.   

Recent developments in multi-modality medical imaging (magnetic resonance imaging - MRI, positron emission tomography - PET, single photon emission computed tomography - SPECT, magnetoencephalography - MEG) have provided the opportunity to analyze them for the purpose of non-invasive evaluation of epileptic patients. In particular, the union of MRI analysis with other imaging studies (ictal and interictal SPECT and MEG) and pre- or early admission electrographic study enable more efficient rendering of the epileptic focus. Moreover, establishment of a database containing similar cases with outcomes provide predictive weight in regards to therapy by offering a comparison with cases manifesting similar characteristics. The results of these multi-modality image analysis methods will improve the treatment and reduce the cost for majority of patients.   

 In this talk, we will present highlights of recent image analysis research conducted for noninvasive determination of structural and volumetric changes in the human brain using multi-modality images that can be used for the determination of focal abnormality and surgical candidacy of patients with temporal lobe epilepsy.  To this end, development, evaluation, and validation of novel knowledge-based approaches to localize, segment, and characterize anatomical structures from three-dimensional (3D) medical images will be presented. These methods include 3D deformable models and wavelet and multi-wavelet feature extractors. In addition, integration of the image analysis methods in a web-based database system named Multi-Modality Human Brain Image Database System (HBIDS) will be explained. Through its query tools, this system will facilitate and improve diagnosis, treatment planning, and treatment evaluation of epilepsy patients.  At the end, new opportunities and remaining challenges in this field will be discussed.

Detection and Monitoring of Brain Injury ِAfter Cardiac Arrest.

Hasan Al-Nashash, Ph.D., Department of Electrical Engineering at the American University of Sharjah, United Arab Emirates

Cardiovascular disease is the major cause of death in UAE and the Arab Gulf Countries. Although recent technological developments of defibrillators have resulted in successful resuscitation of many patients in or outside the hospital, a large majority of resuscitated patients are left with significant neurological impairment. Dramatic improvements in quantitative measures of brain injury severity have been demonstrated. Still however, the overall compelling goal is to bring to the bedside state-of-the art electronic instrumentation for rapid and accurate detection of brain injury severity and progression in cardiac arrest victims by using novel engineering approaches.  

The electroencephalograph (EEG) is a useful tool in clinical neurophysiology and can be used for the identification of cerebral injury.  Following global cerebral ischemia by hypoxic-ischemic cardiac arrest, an important observation related to the process of neurological recovery is the presence of spikes and bursts pattern manifested in the EEG. To study the different statistical distributions associated with spiking and random background activity of EEG rhythms, we use quantitative measures related to the amount of “information” content of the signal. Subband Wavelet Entropy (SWE) is used to characterize the interactions between the bursting and random background of EEG rhythms. The wavelet transform is useful for progressively and systematically ‘decomposing’ the EEG into multi-scaled components. For EEG signal sampled at 250 Hz, a five level decomposition results in a good match to the standard clinical bands of interest: Gamma (31.2–62.5Hz), Beta  (15.6–31.2Hz), Alpha  (7.8–15.6Hz), Theta  (3.9–7.8Hz) and Delta  (1.9-3.9Hz).

 The EEG is measured from rodent brains in a controlled experimental brain injury model by hypoxic-ischemic cardiac arrest. Results show that while the relative EEG power fails to reveal the order of bursting activity associated with recovery, wavelet entropy is used to segment the EEG and delineate the initial bursting periods in each subband. 

 Markov process amplitude algorithm is also used to model and simulate the EEG signal to identify pathophysiological EEG changes. The EEG signal from the injured brain during various phases of injury and recovery is modeled. Results show that the model is accurate in simulating EEG signal variations following brain injury. The dynamics of the model coefficients successfully capture the presence of spiking and bursting in EEG.

FROM LIFE SCIENCES TO ENGINEERING LIVING TISSUES

Mona K. Marei, BDS, MSCD, Ph.D., Tissue Engineering Lab, Alexandria University, Egypt

Tissue and organ failure produced as a result of injury or other type of damage, is a major health problem accounting for about half of the total annual expenditure in healthcare all over the world.

    Treatment options include transplantation (human or xenotransplantation), surgical repair, artificial prostheses, mechanical devices, in a few cases drug therapy. Ultimately, however, major damage to a tissue or organ can neither be repaired nor long-term  recovery effected in a truly satisfactory way by these method.Working in life sciences, as in all technical fields is becoming increasingly cross-disciplinary, as evidenced in new hybrid disciplines like Tissue Engineering.

      Tissue Engineering is emerging as a significant potential alternative or complementary  solution, whereby tissue and organ failure is addressed by implanting natural, synthetic, or semisynthetic tissue and organ mimics that are fully functional from the start, or that grow into the required functionality. The term Tissue Engineering was officially coined at NSF 1988 as the application of principles and methods of engineering and life sciences toward fundamental understanding in normal and pathological mammalian tissues and the development of biological substitutes to restore, maintain or improve tissue function.

         Much of the current research in the field involves growing cells in 3/D structures instead of in laboratory dishes. In 3/D culture method, scaffolds dissolve once the cells reach a certain mass, hoping that this mass will mature into fully differentiated tissues and organs. Biomedicine, along with technological advances and years of research, has led scientists and engineers to attempt to solve the problem of organ and tissue transplantation in human body. Organ donation is very scarce in comparison to the millions of people in need of these surgical procedures and transplants.

         There is increasing need for kidney, liver, pancreas, and skin transplants, along with cartilage to repair knee damage, bone for facial reconstruction and muscle repair……..etc.

        Tissue Engineering draws on experts from chemical and  mechanical engineering, materials science, surgery, genetics and related disciplines from engineering and life sciences.

Monitoring Fluid Shifts During Hemodialysis (HD) Using Electrical Bio-Impedance Techniques

Omar Al-Surkhi, Ph.D., Biomedical Engineering Research Center (CREB), Technical University of Catalonia –UPC, Barcelona , Spain

Numbers of patients suffering from Chronic Kidney Disease (CKD) are very critical. 20 million Americans (1 in 9 US adults) have CKD and another 20 million are at increased risk [National Kidney Foundation, 2006]. Haemodialysis has been considered as an effective therapy for patients with end-state renal disease; however patients can suffer from adverse side effects during haemodialysis.

The main objective of this research work is to help in solving a clinically relevant problem that is suffered by about 25% of patients that need haemodialysis treatment. These patients experience acute complications (haemodynamic instability) during the treatment sessions, which provoke discomfort to the patients due to fainting, vomiting, etc. and usually require relatively long recovery periods. These instabilities are related to the excess shift of fluids between the extracellular- ECW- and intracellular- ICW- as shown in several publications during the past 10 years. In this work we propose a non-invasive method based on local multifrequency bioimpedance measurements that allow us to determine the fluid distribution and variations during haemodialysis. For that purpose a measurement system was developed for the measurement of the body fluid balance changes (ratio ECW/ICW) during haemodialysis sessions. Clinical measurements were done using 10 HD patients during 60 HD sessions. Bioimpedance data, arterial blood pressure, blood volume and blood heamatocrit variations were recorded continuously during the HD sessions. Impedance values at infinite and zero (R,R0) frequencies were extrapolated from Cole-Cole mathematical model. These values represent the impedance of total tissue fluid and the impedance of the extracellular space respectively. Estimators for the extracellular and intracellular fluid volumes were developed using Hanai theory of mixtures with values of (R,R0). Significant decrease in the ECW volumes were recorded during HD session for all patients, however ICW variations were not similar in all patients. In this phase of the research work, the selected patients were stable; however in future work non-stable patients will be selected and the electrical impedance measurements of the fluid variations will be used to develop an indicator of hypovolemic crisis during HD thus allowing a better control of the adverse effects during dialysis sessions.
 

 Imaging the Heart Using MRI

Nael F. Osman, Ph.D., Departments of Radiology and Electrical and Computer Engineering, Johns Hopkins University

Imaging of the heart is one of the most challenging applications of MRI. The constant motion of the heart compounded by breathing motion makes the heart a difficult organ to image. This drives the requirements on MRI performance to new territories of innovation in order to capture the structure, physiology and function of the heart. Because of this, the push to develop new methods and techniques for cardiac MR reaches beyond the heart and benefits other MRI applications. In this talk, the current applications of MRI in cardiac imaging will be presented with special emphasis on the unique capabilities of MRI in quantifying cardiac function. Focus will be on the engineering and technical aspects of cardiac imaging.

 

Send mail to cibec06@k-space.org with questions or comments about this web site.
Last modified: 12/06/06