Matching Items (21)

134706-Thumbnail Image.png

Open-Source Feature Selection Tool for Medical Imaging Diagnosis

Description

Open source image analytics and data mining software are widely available but can be overly-complicated and non-intuitive for medical physicians and researchers to use. The ASU-Mayo Clinic Imaging Informatics Lab

Open source image analytics and data mining software are widely available but can be overly-complicated and non-intuitive for medical physicians and researchers to use. The ASU-Mayo Clinic Imaging Informatics Lab has developed an in-house pipeline to process medical images, extract imaging features, and develop multi-parametric models to assist disease staging and diagnosis. The tools have been extensively used in a number of medical studies including brain tumor, breast cancer, liver cancer, Alzheimer's disease, and migraine. Recognizing the need from users in the medical field for a simplified interface and streamlined functionalities, this project aims to democratize this pipeline so that it is more readily available to health practitioners and third party developers.

Contributors

Agent

Created

Date Created
  • 2016-12

133439-Thumbnail Image.png

Ambassador Handbook

Description

Arizona State University and Banner Thunderbird Hospital have partnered to provide pre-med students with an internship at a local emergency department. Students entering into this program will have access to

Arizona State University and Banner Thunderbird Hospital have partnered to provide pre-med students with an internship at a local emergency department. Students entering into this program will have access to each patient's vital signs, medical imaging, lab tests, and medications. This access presents students with an opportunity to learn about a variety of tools used in the assessment and treatment of emergency room patients. In order to enhance the amount of knowledge students take away from the program, I created a handbook summarizing a variety of diagnostic tests and medications. The first section of the handbook (assessment) is spilt up into the three following categories: vital signs, medical imaging, and lab tests. The second section (treatment) consists of one category, medications. Each section was written with emphasis on basic physiology, and is intended to provide pre-med students with a foundation for building further medical knowledge. Although this handbook was tailored to the information students are most likely to encounter working in Banner Thunderbird Hospital's emergency department, it is still appropriate for any student interested in learning about emergency medicine.

Contributors

Agent

Created

Date Created
  • 2018-05

155389-Thumbnail Image.png

Scaling Up Large-scale Sparse Learning and Its Application to Medical Imaging

Description

Large-scale $\ell_1$-regularized loss minimization problems arise in high-dimensional applications such as compressed sensing and high-dimensional supervised learning, including classification and regression problems. In many applications, it remains challenging to apply

Large-scale $\ell_1$-regularized loss minimization problems arise in high-dimensional applications such as compressed sensing and high-dimensional supervised learning, including classification and regression problems. In many applications, it remains challenging to apply the sparse learning model to large-scale problems that have massive data samples with high-dimensional features. One popular and promising strategy is to scaling up the optimization problem in parallel. Parallel solvers run multiple cores on a shared memory system or a distributed environment to speed up the computation, while the practical usage is limited by the huge dimension in the feature space and synchronization problems.

In this dissertation, I carry out the research along the direction with particular focuses on scaling up the optimization of sparse learning for supervised and unsupervised learning problems. For the supervised learning, I firstly propose an asynchronous parallel solver to optimize the large-scale sparse learning model in a multithreading environment. Moreover, I propose a distributed framework to conduct the learning process when the dataset is distributed stored among different machines. Then the proposed model is further extended to the studies of risk genetic factors for Alzheimer's Disease (AD) among different research institutions, integrating a group feature selection framework to rank the top risk SNPs for AD. For the unsupervised learning problem, I propose a highly efficient solver, termed Stochastic Coordinate Coding (SCC), scaling up the optimization of dictionary learning and sparse coding problems. The common issue for the medical imaging research is that the longitudinal features of patients among different time points are beneficial to study together. To further improve the dictionary learning model, I propose a multi-task dictionary learning method, learning the different task simultaneously and utilizing shared and individual dictionary to encode both consistent and changing imaging features.

Contributors

Agent

Created

Date Created
  • 2017

157226-Thumbnail Image.png

Confocal Laser Endomicroscopy Image Analysis with Deep Convolutional Neural Networks

Description

Rapid intraoperative diagnosis of brain tumors is of great importance for planning treatment and guiding the surgeon about the extent of resection. Currently, the standard for the preliminary intraoperative tissue

Rapid intraoperative diagnosis of brain tumors is of great importance for planning treatment and guiding the surgeon about the extent of resection. Currently, the standard for the preliminary intraoperative tissue analysis is frozen section biopsy that has major limitations such as tissue freezing and cutting artifacts, sampling errors, lack of immediate interaction between the pathologist and the surgeon, and time consuming.

Handheld, portable confocal laser endomicroscopy (CLE) is being explored in neurosurgery for its ability to image histopathological features of tissue at cellular resolution in real time during brain tumor surgery. Over the course of examination of the surgical tumor resection, hundreds to thousands of images may be collected. The high number of images requires significant time and storage load for subsequent reviewing, which motivated several research groups to employ deep convolutional neural networks (DCNNs) to improve its utility during surgery. DCNNs have proven to be useful in natural and medical image analysis tasks such as classification, object detection, and image segmentation.

This thesis proposes using DCNNs for analyzing CLE images of brain tumors. Particularly, it explores the practicality of DCNNs in three main tasks. First, off-the shelf DCNNs were used to classify images into diagnostic and non-diagnostic. Further experiments showed that both ensemble modeling and transfer learning improved the classifier’s accuracy in evaluating the diagnostic quality of new images at test stage. Second, a weakly-supervised learning pipeline was developed for localizing key features of diagnostic CLE images from gliomas. Third, image style transfer was used to improve the diagnostic quality of CLE images from glioma tumors by transforming the histology patterns in CLE images of fluorescein sodium-stained tissue into the ones in conventional hematoxylin and eosin-stained tissue slides.

These studies suggest that DCNNs are opted for analysis of CLE images. They may assist surgeons in sorting out the non-diagnostic images, highlighting the key regions and enhancing their appearance through pattern transformation in real time. With recent advances in deep learning such as generative adversarial networks and semi-supervised learning, new research directions need to be followed to discover more promises of DCNNs in CLE image analysis.

Contributors

Agent

Created

Date Created
  • 2019

157380-Thumbnail Image.png

Ultrahigh Field Functional Magnetic Resonance Electrical Impedance Tomography (fMREIT) in Neural Activity Imaging

Description

A direct Magnetic Resonance (MR)-based neural activity mapping technique with high spatial and temporal resolution may accelerate studies of brain functional organization.

The most widely used technique for brain functional

A direct Magnetic Resonance (MR)-based neural activity mapping technique with high spatial and temporal resolution may accelerate studies of brain functional organization.

The most widely used technique for brain functional imaging is functional Magnetic Resonance Image (fMRI). The spatial resolution of fMRI is high. However, fMRI signals are highly influenced by the vasculature in each voxel and can be affected by capillary orientation and vessel size. Functional MRI analysis may, therefore, produce misleading results when voxels are nearby large vessels. Another problem in fMRI is that hemodynamic responses are slower than the neuronal activity. Therefore, temporal resolution is limited in fMRI. Furthermore, the correlation between neural activity and the hemodynamic response is not fully understood. fMRI can only be considered an indirect method of functional brain imaging.

Another MR-based method of functional brain mapping is neuronal current magnetic resonance imaging (ncMRI), which has been studied over several years. However, the amplitude of these neuronal current signals is an order of magnitude smaller than the physiological noise. Works on ncMRI include simulation, phantom experiments, and studies in tissue including isolated ganglia, optic nerves, and human brains. However, ncMRI development has been hampered due to the extremely small signal amplitude, as well as the presence of confounding signals from hemodynamic changes and other physiological noise.

Magnetic Resonance Electrical Impedance Tomography (MREIT) methods could have the potential for the detection of neuronal activity. In this technique, small external currents are applied to a body during MR scans. This current flow produces a magnetic field as well as an electric field. The altered magnetic flux density along the main magnetic field direction caused by this current flow can be obtained from phase images. When there is neural activity, the conductivity of the neural cell membrane changes and the current paths around the neurons change consequently. Neural spiking activity during external current injection, therefore, causes differential phase accumulation in MR data. Statistical analysis methods can be used to identify neuronal-current-induced magnetic field changes.

Contributors

Agent

Created

Date Created
  • 2019

158670-Thumbnail Image.png

Neural Activity Mapping Using Electromagnetic Fields: An In Vivo Preliminary Functional Magnetic Resonance Electrical Impedance Tomography (fMREIT) Study

Description

Electromagnetic fields (EMFs) generated by biologically active neural tissue are critical in the diagnosis and treatment of neurological diseases. Biological EMFs are characterized by electromagnetic properties such as electrical conductivity,

Electromagnetic fields (EMFs) generated by biologically active neural tissue are critical in the diagnosis and treatment of neurological diseases. Biological EMFs are characterized by electromagnetic properties such as electrical conductivity, permittivity and magnetic susceptibility. The electrical conductivity of active tissue has been shown to serve as a biomarker for the direct detection of neural activity, and the diagnosis, staging and prognosis of disease states such as cancer. Magnetic resonance electrical impedance tomography (MREIT) was developed to map the cross-sectional conductivity distribution of electrically conductive objects using externally applied electrical currents. Simulation and in vitro studies of invertebrate neural tissue complexes demonstrated the correlation of membrane conductivity variations with neural activation levels using the MREIT technique, therefore laying the foundation for functional MREIT (fMREIT) to detect neural activity, and future in vivo fMREIT studies.

The development of fMREIT for the direct detection of neural activity using conductivity contrast in in vivo settings has been the focus of the research work presented here. An in vivo animal model was developed to detect neural activity initiated changes in neuronal membrane conductivities under external electrical current stimulation. Neural activity was induced in somatosensory areas I (SAI) and II (SAII) by applying electrical currents between the second and fourth digits of the rodent forepaw. The in vivo animal model involved the use of forepaw stimulation to evoke somatosensory neural activations along with hippocampal fMREIT imaging currents contemporaneously applied under magnetic field strengths of 7 Tesla. Three distinct types of fMREIT current waveforms were applied as imaging currents under two inhalants – air and carbogen. Active regions in the somatosensory cortex showed significant apparent conductivity changes as variations in fMREIT phase (φ_d and ∇^2 φ_d) signals represented by fMREIT activation maps (F-tests, p <0.05). Consistent changes in the standard deviation of φ_d and ∇^2 φ_d in cortical voxels contralateral to forepaw stimulation were observed across imaging sessions. These preliminary findings show that fMREIT may have the potential to detect conductivity changes correlated with neural activity.

Contributors

Agent

Created

Date Created
  • 2020

157651-Thumbnail Image.png

Magnetic resonance parameter assessment from a second order time-dependent linear model

Description

This dissertation develops a second order accurate approximation to the magnetic resonance (MR) signal model used in the PARSE (Parameter Assessment by Retrieval from Single Encoding) method to recover information

This dissertation develops a second order accurate approximation to the magnetic resonance (MR) signal model used in the PARSE (Parameter Assessment by Retrieval from Single Encoding) method to recover information about the reciprocal of the spin-spin relaxation time function (R2*) and frequency offset function (w) in addition to the typical steady-state transverse magnetization (M) from single-shot magnetic resonance imaging (MRI) scans. Sparse regularization on an approximation to the edge map is used to solve the associated inverse problem. Several studies are carried out for both one- and two-dimensional test problems, including comparisons to the first order approximation method, as well as the first order approximation method with joint sparsity across multiple time windows enforced. The second order accurate model provides increased accuracy while reducing the amount of data required to reconstruct an image when compared to piecewise constant in time models. A key component of the proposed technique is the use of fast transforms for the forward evaluation. It is determined that the second order model is capable of providing accurate single-shot MRI reconstructions, but requires an adequate coverage of k-space to do so. Alternative data sampling schemes are investigated in an attempt to improve reconstruction with single-shot data, as current trajectories do not provide ideal k-space coverage for the proposed method.

Contributors

Agent

Created

Date Created
  • 2019

155261-Thumbnail Image.png

Improved spatial coverage of high-temporal resolution dynamic susceptibility contrast-MRI through 3D spiral-based acquisition and parallel imaging

Description

Dynamic susceptibility contrast MRI (DSC-MRI) is a powerful tool used to quantitatively measure parameters related to blood flow and volume in the brain. The technique is known as a “bolus-tracking”

Dynamic susceptibility contrast MRI (DSC-MRI) is a powerful tool used to quantitatively measure parameters related to blood flow and volume in the brain. The technique is known as a “bolus-tracking” method and relies upon very fast scanning to accurately measure the flow of contrast agent into and out of a region of interest. The need for high temporal resolution to measure contrast agent dynamics limits the spatial coverage of perfusion parameter maps which limits the utility of DSC-perfusion studies in pathologies involving the entire brain. Typical clinical DSC-perfusion studies are capable of acquiring 10-15 slices, generally centered on a known lesion or pathology.

The methods developed in this work improve the spatial coverage of whole-brain DSC-MRI by combining a highly efficient 3D spiral k-space trajectory with Generalized Autocalibrating Partial Parallel Acquisition (GRAPPA) parallel imaging without increasing temporal resolution. The proposed method is capable of acquiring 30 slices with a temporal resolution of under 1 second, covering the entire cerebrum with isotropic spatial resolution of 3 mm. Additionally, the acquisition method allows for correction of T1-enhancing leakage effects by virtue of collecting two echoes, which confound DSC perfusion measurements. The proposed DSC-perfusion method results in high quality perfusion parameter maps across a larger volume than is currently available with current clinical standards, improving diagnostic utility of perfusion MRI methods, which ultimately improves patient care.

Contributors

Agent

Created

Date Created
  • 2017

3D - Patch Based Machine Learning Systems for Alzheimer’s Disease classification via 18F-FDG PET Analysis

Description

Alzheimer’s disease (AD), is a chronic neurodegenerative disease that usually starts slowly and gets worse over time. It is the cause of 60% to 70% of cases of dementia. There

Alzheimer’s disease (AD), is a chronic neurodegenerative disease that usually starts slowly and gets worse over time. It is the cause of 60% to 70% of cases of dementia. There is growing interest in identifying brain image biomarkers that help evaluate AD risk pre-symptomatically. High-dimensional non-linear pattern classification methods have been applied to structural magnetic resonance images (MRI’s) and used to discriminate between clinical groups in Alzheimers progression. Using Fluorodeoxyglucose (FDG) positron emission tomography (PET) as the pre- ferred imaging modality, this thesis develops two independent machine learning based patch analysis methods and uses them to perform six binary classification experiments across different (AD) diagnostic categories. Specifically, features were extracted and learned using dimensionality reduction and dictionary learning & sparse coding by taking overlapping patches in and around the cerebral cortex and using them as fea- tures. Using AdaBoost as the preferred choice of classifier both methods try to utilize 18F-FDG PET as a biological marker in the early diagnosis of Alzheimer’s . Addi- tional we investigate the involvement of rich demographic features (ApoeE3, ApoeE4 and Functional Activities Questionnaires (FAQ)) in classification. The experimental results on Alzheimer’s Disease Neuroimaging initiative (ADNI) dataset demonstrate the effectiveness of both the proposed systems. The use of 18F-FDG PET may offer a new sensitive biomarker and enrich the brain imaging analysis toolset for studying the diagnosis and prognosis of AD.

Contributors

Agent

Created

Date Created
  • 2017

158352-Thumbnail Image.png

Visualization of Brain Tumors with Intraoperative Confocal Laser Endomicroscopy

Description

Intraoperative diagnosis in neurosurgery has traditionally relied on frozen and formalin-fixed, paraffin-embedded section analysis of biopsied tissue samples. Although this technique is considered to be the “gold standard” for establishing

Intraoperative diagnosis in neurosurgery has traditionally relied on frozen and formalin-fixed, paraffin-embedded section analysis of biopsied tissue samples. Although this technique is considered to be the “gold standard” for establishing a histopathologic diagnosis, it entails a number of significant limitations such as invasiveness and the time required for processing and interpreting the tissue. Rapid intraoperative diagnosis has become possible with a handheld confocal laser endomicroscopy (CLE) system. Combined with appropriate fluorescent stains or labels, CLE provides an imaging technique for real-time intraoperative visualization of histopathologic features of the suspected tumor and healthy tissues.

This thesis scrutinizes CLE technology for its ability to provide real-time intraoperative in vivo and ex vivo visualization of histopathological features of the normal and tumor brain tissues. First, the optimal settings for CLE imaging are studied in an animal model along with a generational comparison of CLE performance. Second, the ability of CLE to discriminate uninjured normal brain, injured normal brain and tumor tissues is demonstrated. Third, CLE was used to investigate cerebral microvasculature and blood flow in normal and pathological conditions. Fourth, the feasibility of CLE for providing optical biopsies of brain tumors was established during the fluorescence-guided neurosurgical procedures. This study established the optimal workflow and confirmed the high specificity of the CLE optical biopsies. Fifth, the feasibility of CLE was established for endoscopic endonasal approaches and interrogation of pituitary tumor tissue. Finally, improved and prolonged near wide-field fluorescent visualization of brain tumor margins was demonstrated with a scanning fiber endoscopy and 5-aminolevulinic acid.

These studies suggested a novel paradigm for neurosurgery-pathology workflow when the noninvasive intraoperative optical biopsies are used to interrogate the tissue and augment intraoperative decision making. Such optical biopsies could shorten the time for obtaining preliminary information on the histological composition of the tissue of interest and may lead to improved diagnostics and tumor resection. This work establishes a basis for future in vivo optical biopsy use in neurosurgery and planning of patient-related outcome studies. Future studies would lead to refinement and development of new confocal scanning technologies making noninvasive optical biopsy faster, convenient and more accurate.

Contributors

Agent

Created

Date Created
  • 2020