165 research outputs found
Wide-field modulated imaging for non-invasive quantification of tissue properties: a method development study
No full textModulated Imaging (MI) is a recently reported method for rapid, non-invasive quantification of tissue optical properties (reduced scattering, µs and absorption, µa), which can be performed across a range of optical wavelengths to determine chromophore concentrations. In this thesis, development and characterization of a compact, low cost MI system is reported, using off-the-shelf hardware components with a custom software interface capable of easy modification for specific applications. This prototype setup consists of a color CCD camera which captures the diffusely reflected light from an object illuminated with patterns generated by a miniature projector. Broadband white light from the projector is delivered through a filter wheel containing narrowband filters for measurement at 420nm, 570nm, and 620nm wavelengths. A software application in MATLAB was written to control and synchronize the phase-shifted illumination patterns with image acquisition, and perform processing of image data into optical property maps. System accuracy was characterized by measuring a series of tissue simulating phantoms fabricated with varying µs and µa, with both the prototype platform and a commercially available MI system as a reference. The overall error of the prototype system, for µs ranging from 0.93-2.23mm-1 and µa ranging from 0.009-0.049mm-1, was approximately 10% and 16%, respectively. Utilizing a lookup table that requires measurements at two illumination spatial frequencies instead of performing a least-squares fit to diffuse reflectance measurements at ten frequencies reduced the acquisition and processing time by 80%, while reducing the accuracy of optical property determination by approximately 3%. In summary, a prototype MI platform was developed and shown to be capable of quantifying the optical properties within biologically relevant µs and µa ranges. The system was assembled for less than 10% of the cost of commercially available systems while enabling individual components to be upgraded for a wider range of accurate optical property determination. Scattering and absorption maps obtained at multiple wavelengths can subsequently be used to quantify the concentrations of various tissue chromophores including hemoglobin, water, and lipids. Non-invasive, image based acquisition of such information may have impact in medical applications, ultimately improving patient health through disease characterization and monitoring progress of treatment.M.S.Includes bibliographical referencesby Vipul Atulkumar Bax
Kinematic primitives in action similarity judgments : A human-centered computational model
No full textThis paper investigates the role that kinematic features play in human action similarity judgments. The results of three experiments with human participants are compared with the computational model that solves the same task. The chosen model has its roots in developmental robotics and performs action classification based on learned kinematic primitives. The comparative experimental results show that both model and human participants can reliably identify whether two actions are the same or not. Specifically, most of the given actions could be similarity judged based on very limited information from a single feature domain (velocity or spatial). Both velocity and spatial features were however necessary to reach a level of human performance on evaluated actions. The experimental results also show that human performance on an action identification task indicated that they clearly relied on kinematic information rather than on action semantics. The results show that both the model and human performance are highly accurate in an action similarity task based on kinematic-level features, which can provide an essential basis for classifying human actions. CC BY 4.0Corresponding author: Vipul Nair.This work has been partially carried out at the Machine Learning Genoa (MaLGa) center, Università di Genova (IT). It has been partially supported by AFOSR, grant n. FA8655-20-1-7035, and research collaboration between University of Skövde and Istituto Italiano di Tecnologia, Genoa.</p
DataSpread: scaling spreadsheets using relational databases
No full textSpreadsheet software is the tool of choice for ad-hoc tabular data management, manipulation, querying, and visualization with adoption by billions of users. However, spreadsheets are not scalable, unlike database systems. We develop DataSpread, a system that holistically unifies databases and spreadsheets with a goal to work with massive spreadsheets: DataSpread retains all of the advantages of spreadsheets, including ease of use, ad-hoc analysis and visualization capabilities, and a schema-free nature, while also adding the scalability and collaboration abilities of traditional relational databases. We design DataSpread with a spreadsheet front-end and a regular relational database back-end. To integrate spreadsheets and databases, in this thesis, we develop a storage and indexing engine for spreadsheet data. We first formalize and study the problem of representing and manipulating spreadsheet data within a relational database. We demonstrate that identifying the optimal representation is NP-Hard via a reduction from partitioning of rectangles; however, under certain reasonable assumptions, can be solved in PTIME. We develop a collection of mechanisms for representing spreadsheet data, and evaluate these representations on a workload of typical data manipulation operations. We augment our mechanisms with novel positionally-aware indexing structures that further improve performance. DataSpread can scale to billions of cells, returning results for common operations within seconds. Lastly, to motivate our research questions, we perform an extensive survey of spreadsheet use for ad-hoc tabular data management.Submission published under a 24 month embargo labeled 'U of I Access', the embargo will last until 2019-05-01The student, Vipul Venkataraman, accepted the attached license on 2017-04-12 at 12:19.The student, Vipul Venkataraman, submitted this Thesis for approval on 2017-04-12 at 12:20.This Thesis was approved for publication on 2017-04-12 at 15:22.DSpace SAF Submission Ingestion Package generated from Vireo submission #10722 on 2017-08-10 at 15:05:27Made available in DSpace on 2017-08-10T20:32:52Z (GMT). No. of bitstreams: 2
VENKATARAMAN-THESIS-2017.pdf: 1068361 bytes, checksum: d09adf1e73a044e0e5e4ea1dc393e475 (MD5)
LICENSE.txt: 4215 bytes, checksum: f35d5860899c10463eff04f36479665c (MD5)
Previous issue date: 2017-04-12Embargo set by: Colleen Fallaw for item 102745
Lift date: 2019-08-10T21:27:21Z
Reason: Author requested U of Illinois access only (OA after 2yrs) in Vireo ETD systemU of I Only Restriction Lifted for Item 102745 on 2019-08-11T09:15:28Z
Book Review: Fundamentals of Operative Surgery, 2nd edition
No full textBook Title: Fundamentals of Operative SurgeryBook Author: Vipul YagnikPublisher: Wolters Kluwer, Year 2019, 435 pages Format: PDF and EPUB, Hardcover ISBN 9789387506817
GuavaNet: A deep neural network architecture for automatic sensory evaluation to predict degree of acceptability for Guava by a consumer
No full textThis thesis is divided into two parts:Part I: Analysis of Fruits, Vegetables, Cheese and Fish based on Image Processing using Computer Vision and Deep Learning: A Review. It consists of a comprehensive review of image processing, computer vision and deep learning techniques applied to carry out analysis of fruits, vegetables, cheese and fish.This part also serves as a literature review for Part II.Part II: GuavaNet: A deep neural network architecture for automatic sensory evaluation to predict degree of acceptability for Guava by a consumer. This part introduces to an end-to-end deep neural network architecture that can predict the degree of acceptability by the consumer for a guava based on sensory evaluation
Special session honoring the contributions of Dr. T. Kevin O'Brien, NASA Langley Research Center
No full textAeronautical composite stiffened structures have the capability to carry loads deep into postbuckling, yet they are typically designed to operate below the buckling load to avoid potential issues with durability and structural integrity. Large out-of-plane postbuckling deformation of the skin can result in the opening of the skin-stringer interfaces, especially in the presence of defects, such as impact damage. To ensure that skin-stringer separation does not propagate in an unstable mode that can cause a complete collapse of the structure, a deeper understanding of the interaction between the postbuckling deformation and the development of damage is required. The present study represents a first step towards a methodology based on analysis and experiments to assess and improve the strength, life, and damage tolerance of stiffened composite structures subjected to postbuckling deformations. Two regions were identified in a four-stringer panel in which skin-stringer separation can occur, namely the region of maximum deflection and the region of maximum twisting. Both regions have been studied using a finite element model of a representative single-stringer specimen. For the region of maximum deflection, a seven-point bending configuration was used, in which five supports and two loading points induce buckling waves to the specimen. The region of maximum twisting was studied using an edge crack torsion configuration, with two supports and two loading points. These two configurations were studied by changing the positions of the supports and the loading points. An optimization procedure was carried out to minimize the error between the out-of-plane deformation of the representative single-stringer specimen and the corresponding region of the four-stringer panel
Mapping Techniques And Performance Analysis For An Interconnection Cached Multiprocessor Network
No full textMany parallel applications exhibit a behavior in which each computation entity communicates with a small set of other entities and the communication pattern changes slowly with respect to time. We call this phenomenon switching locality. The Interconnection Cached Network (ICN) is a reconfigurable network suitable for exploiting such locality. The ICN contains a number of small crossbar switches that connect clusters of processing elements to the input/output ports of a single large crossbar. Technology restrictions impose a trade-o between the size of a switch and its switching speed. By using a large crossbar for topology configuration, and small crossbars for the more frequent task of message switching, the ICN effectively combines the connectivity of the large switch with the speed of the smaller switches. This is analogous to the concept of memory caching.
Embedding communication patterns efficiently in an ICN requires finding a special kind of partitioning, called a bounded l-contraction, of the corresponding communication graph. The problem of identifying whether a graph has a bounded l-contraction for a given integer l is NP-complete for l > 2. We extend the class of classical communication graphs that are known to have efficient embeddings in the ICN. For general graphs, we develop a heuristic algorithm based on simulated annealing to solve this partitioning problem. In addition to providing a mapping strategy for assigning processes to processing elements, this partitioning also generates the topology to which the ICN must be configured.
For applications with sufficient switching locality, good mappings combined with topology reconfiguration in the ICN ensure that communication path lengths are uniformly short. Conventional networks are less successful in meeting these objectives. Using both analysis and simulations, we show that the ICN outperforms other networks, such as multistage interconnection networks and low degree k-ary n-cubes, in terms of message latency, highest sustainable throughput, processor utilization and application scalability.Technical report DCS-TR-31
Predicting progressions and clinical subtypes of Alzheimer’s disease using machine learning
No full textAlzheimer’s disease is a degenerative brain disease which impairs a person’s ability to perform day to day activities. Research has shown AD to be a heterogeneous condition, having a high variation in terms of the symptoms and disease progression rate. Treating Alzheimer's disease (AD) is especially challenging due to these variations present in the disease progression stages. The clinical symptoms of AD show marked variability in terms of patients’ age, disease span, progression velocity and types of memory, cognitive and depression related features. Hence, the idea of personalized clinical care, with individualized risk, progression and prediction related patient advice in AD is narrow. This facilitates the yet unfulfilled need for an early prediction of the disease course to assist its treatment and tailor therapy options to the progression rate. Additionally, there are ramifications in clinical trial design when considering the high heterogeneity of disease manifestation and progression.
Recent developments in machine learning techniques provide a huge potential, not only to predict the onset and progression of Alzheimer's disease but also to classify the disease into different etiological subtypes. The advancement of these prediction models have the potential to impact clinical decision making and improve healthcare resource allocation. It will also lead to the development of personalized clinical care and counseling for patients, hopefully reducing AD treatment costs.
The suggested work clusters patients in distinct and multifaceted progression subgroups of Alzheimer's disease and discusses an approach to predict the progression stage from baseline diagnosis through the implementation of machine learning techniques. By applying machine learning algorithms on the extensive clinical observations available in the Alzheimer's Disease Neuroimaging Initiative (ADNI) dataset, we parse the progression space for the Alzheimer’s disease into low, moderate and high disease progressors. This work suggests that the myriad of clinically reported symptoms we summarize in the Alzheimer's Disease progression space correspond directly to memory and cognition measurements classically used to monitor disease onset and progression. The proposed work concludes notably accurate prediction of disease progression after four years from the first 12 months of post-diagnosis data (area under receiver operating characteristic (ROC) curve of 0.90±0.02 for Controls, 0.96±0.04 for High rate, 0.90±0.04 for Moderate rate 0.83±0.06 for Low rate). We validate our model through five-fold cross-validation to obtain a robust prediction of membership into these progression subtypes.
These machine learning techniques will assist the medical practitioners to classify different progression rates within patients and allow for more efficient and unique care delivery. With additional information about the onset rate of AD at hand, doctors may alter their treatments to better suit the patients. The predictive tests discussed in this report not only allow for early detection but also facilitate the characterization of distinct disease subtypes relating to trajectories of disease progression. This will lead to improved clinical trial design and reducing skyrocketing healthcare costs in the future.Submission published under a 24 month embargo labeled 'Closed Access', the embargo will last until 2021-05-01The student, Vipul Satone, accepted the attached license on 2019-04-24 at 19:02.The student, Vipul Satone, submitted this Thesis for approval on 2019-04-24 at 19:58.This Thesis was approved for publication on 2019-04-25 at 14:35.DSpace SAF Submission Ingestion Package generated from Vireo submission #13896 on 2019-08-22 at 16:23:53Made available in DSpace on 2019-08-23T20:48:26Z (GMT). No. of bitstreams: 4
SATONE-THESIS-2019.pdf: 9185404 bytes, checksum: b3c13bcad66fc046f2aaf89b47665601 (MD5)
ADNI_24_MONTH_ANALYSIS.ipynb: 4123768 bytes, checksum: d011bc59e24f28338c7ed5387747da07 (MD5)
FEATURE_DESCRIPTION_TABLE.xlsx: 22715 bytes, checksum: 616ac16ee90729306f0f344caa95582c (MD5)
LICENSE.txt: 4209 bytes, checksum: b1d2d2041a751f194080be81004d6b6e (MD5)
Previous issue date: 2019-04-25Embargo set by: Seth Robbins for item 112385
Lift date: 2021-08-23T20:48:32Z
Reason: Author requested closed access (OA after 2yrs) in Vireo ETD systemLimited Restriction Lifted for Item 112385 on 2021-08-24T09:15:38Z
Evasiveness and the Distribution of Prime Numbers
No full textA Boolean function on variables is called \emph{evasive} if its decision-tree complexity is . A sequence of Boolean functions is \emph{eventually evasive} if is evasive for all sufficiently large .
We confirm the eventual evasiveness of several classes of monotone graph properties under widely accepted number theoretic hypotheses. In particular we show that Chowla's conjecture on Dirichlet primes implies that (a) for any graph , ``forbidden subgraph '' is eventually evasive and (b) all nontrivial monotone properties of graphs with edges are eventually evasive. ( is the number of vertices.)
While Chowla's conjecture is not known to follow from the Extended Riemann Hypothesis (ERH, the Riemann Hypothesis for Dirichlet's functions), we show (b) with the bound under ERH.
We also prove unconditional results: (a) for any graph , the query complexity of ``forbidden subgraph '' is ; (b) for some constant , all nontrivial monotone properties of graphs with edges are eventually evasive.
Even these weaker, unconditional results rely on deep results from number theory such as Vinogradov's theorem on the Goldbach conjecture.
Our technical contribution consists in connecting the topological framework of Kahn, Saks, and Sturtevant (1984), as further developed by Chakrabarti, Khot, and Shi (2002), with a deeper analysis of the orbital structure of permutation groups and their connection to the distribution of prime numbers. Our unconditional results include stronger versions and generalizations of some result of Chakrabarti et al
Cascade Distillation System – A water recovery system for deep space missions
No full textTucson, ArizonaThe 44th International Conference on Environmental Systems was held in Tuscon, Arizona, USA on 13 July 2014 through 17 July 2014.Vipul Patel, Honeywell International, USAHenry Au, Honeywell International, USASarah Shull, NASA Johnson Space Center, USAMiriam J. Sargusingh, NASA Johnson Space Center, USAMichael Callahan, NASA Johnson Space Center, USAHoneywell Aerospace has developed a distillation technology to process wastewater streams in microgravity environments for recovering potable water. The wastewater processing Cascade Distillation System (CDS) utilizes an innovative and proven multi-stage thermodynamic process to produce purified water. The Cascade Distiller (CD) is the core component of the CDS technology. The CD is a Centrifugal Vacuum Distiller (CVD) that processes wastewater as a feed source and purifies it to near potable water. Some volatile substances escape to the purified water. With minimum post processing, the water can be restored as potable for human consumption. The CD was tested at the National Aeronautics and Space Administration (NASA) Johnson Space Center (JSC) with a greater than 90% recovery rate during a technology comparison test. The results were compared with two other technologies. All three systems were challenged with two pretreated test solutions, each intended to represent a feasible wastewater generated in a deep space environment. An expert panel assembled by NASA down-selected the CDS as one of the technologies for further development. NASA internally developed the Vapor Compression Distiller (VCD) technology, which has reached Technology Readiness Level (TRL) 9. The VCD has paved the way for future development of wastewater recovery technologies by identifying critical requirements. However, the VCD has limited distillation capacity when compared to the CD. Currently, Honeywell Aerospace has an Indefinite Delivery, Indefinite Quantity (IDIQ) contract with NASA for further development of the CD
- …
