SHAREOK Repository
Not a member yet
49261 research outputs found
Sort by
Novel Computer-Aided Diagnosis Schemes for Radiological Image Analysis
Get PDFThe computer-aided diagnosis (CAD) scheme is a powerful tool in assisting clinicians (e.g., radiologists) to interpret medical images more accurately and efficiently. In developing high-performing CAD schemes, classic machine learning (ML) and deep learning (DL) algorithms play an essential role because of their advantages in capturing meaningful patterns that are important for disease (e.g., cancer) diagnosis and prognosis from complex datasets. This dissertation, organized into four studies, investigates the feasibility of developing several novel ML-based and DL-based CAD schemes for different cancer research purposes. The first study aims to develop and test a unique radiomics-based CT image marker that can be used to detect lymph node (LN) metastasis for cervical cancer patients. A total of 1,763 radiomics features were first computed from the segmented primary cervical tumor depicted on one CT image with the maximal tumor region. Next, a principal component analysis algorithm was applied on the initial feature pool to determine an optimal feature cluster. Then, based on this optimal cluster, machine learning models (e.g., support vector machine (SVM)) were trained and optimized to generate an image marker to detect LN metastasis. The SVM based imaging marker achieved an AUC (area under the ROC curve) value of 0.841 ± 0.035. This study initially verifies the feasibility of combining CT images and the radiomics technology to develop a low-cost image marker for LN metastasis detection among cervical cancer patients. In the second study, the purpose is to develop and evaluate a unique global mammographic image feature analysis scheme to identify case malignancy for breast cancer. From the entire breast area depicted on the mammograms, 59 features were initially computed to characterize the breast tissue properties in both the spatial and frequency domain. Given that each case consists of two cranio-caudal and two medio-lateral oblique view images of left and right breasts, two feature pools were built, which contain the computed features from either two positive images of one breast or all the four images of two breasts. For each feature pool, a particle swarm optimization (PSO) method was applied to determine the optimal feature cluster followed by training an SVM classifier to generate a final score for predicting likelihood of the case being malignant. The classification performances measured by AUC were 0.79±0.07 and 0.75±0.08 when applying the SVM classifiers trained using image features computed from two-view and four-view images, respectively. This study demonstrates the potential of developing a global mammographic image feature analysis-based scheme to predict case malignancy without including an arduous segmentation of breast lesions. In the third study, given that the performance of DL-based models in the medical imaging field is generally bottlenecked by a lack of sufficient labeled images, we specifically investigate the effectiveness of applying the latest transferring generative adversarial networks (GAN) technology to augment limited data for performance boost in the task of breast mass classification. This transferring GAN model was first pre-trained on a dataset of 25,000 mammogram patches (without labels). Then its generator and the discriminator were fine-tuned on a much smaller dataset containing 1024 labeled breast mass images. A supervised loss was integrated with the discriminator, such that it can be used to directly classify the benign/malignant masses. Our proposed approach improved the classification accuracy by 6.002%, when compared with the classifiers trained without traditional data augmentation. This investigation may provide a new perspective for researchers to effectively train the GAN models on a medical imaging task with only limited datasets. Like the third study, our last study also aims to alleviate DL models’ reliance on large amounts of annotations but uses a totally different approach. We propose employing a semi-supervised method, i.e., virtual adversarial training (VAT), to learn and leverage useful information underlying in unlabeled data for better classification of breast masses. Accordingly, our VAT-based models have two types of losses, namely supervised and virtual adversarial losses. The former loss acts as in supervised classification, while the latter loss works towards enhancing the model’s robustness against virtual adversarial perturbation, thus improving model generalizability. A large CNN and a small CNN were used in this investigation, and both were trained with and without the adversarial loss. When the labeled ratios were 40% and 80%, VAT-based CNNs delivered the highest classification accuracy of 0.740±0.015 and 0.760±0.015, respectively. The experimental results suggest that the VAT-based CAD scheme can effectively utilize meaningful knowledge from unlabeled data to better classify mammographic breast mass images. In summary, several innovative approaches have been investigated and evaluated in this dissertation to develop ML-based and DL-based CAD schemes for the diagnosis of cervical cancer and breast cancer. The promising results demonstrate the potential of these CAD schemes in assisting radiologists to achieve a more accurate interpretation of radiological images
Using Inexpensive Software-Defined Radios as GPS Receivers in a Ground-Based Augmentation System
Get PDFGround Based Augmentation Systems (GBAS) are used to augment Global Positioning Systems (GPS) signals to make the position solutions significantly more accurate and precise. The systems have been studied and demonstrated before, however, they would typically use dedicated GPS receivers. These receivers are typically expensive and lack the ability to be customized for different situations. This thesis attempts to use a software-defined radio (SDR) using a software called Global Navigation Satellite Systems-Software Defined Receiver (GNSS-SDR) to replace these dedicated GPS receivers. Doing this requires multiple GNSS-SDR receivers to output the data in real-time to a central GBAS computer for real-time computations. This is done using the User Datagram Protocol (UDP) output functionality of the GNSS-SDR software. The output then needs to be received on the GBAS computer and decoded. A novel method for decoding the Google Protocol Buffer encoded UDP messages is used for efficient LabVIEW decoding. The outputs are then tested using an existing Closed-Loop Ground Based Augmentation System (CL-GBAS) program. An analysis of the raw pseudorange values of two different SDRs and certified GPS receivers is then performed. The research performed in thesis will ideally be used as a stepping stone for more thorough analysis of different SDRs and the GNSS-SDR program in an attempt to make SDRs an effective receiver for GBAS purposes
Piezoresistive Sensing in Additively Manufactured Prosthetics
Get PDFThis paper attempts to determine whether hobbyist 3D printers can be used to advance
prosthetic capabilities. We attempt to answer this by designing, printing, and testing a
prosthetic hand using a hobbyist 3D printer and hobbyist materials. The prosthetic hand
with an opposable thumb was drafted from scratch and 3D printed collectively across
four different fused deposition modeling printers; A Craftbot Plus Pro, CR-10, Jgaurora,
and Qidi X-Plus. Once assembled, a material study was conducted against three different
materials to identify the plausibility of sensing force using the materials piezoresistivity.
Piezoresistivity is a measurement of resistance when a mechanical strain is applied. It
was concluded that touch sensing capabilities could be utilized with 3D printed materials
while on a hobbyist grade 3D printer. None of the materials required a heated chamber
however, the argument of a heated bed improving the printability is undeniable. The
likelihood of successfully incorporating this function into a 3D printed prosthetic had
immense potential and promise
Experimental and numerical evaluation of shape memory polymer for lost circulation treatment in geothermal wells
Get PDFLost circulation presents formidable challenges to drilling operations, especially in high-temperature and fractured formations. Lost circulation events increase the nonproductive time (NPT) and the total cost of drilling operations. In some severe cases of complete or high losses, well control is jeopardized, leading to a loss in lives and resources. Besides the conventional lost circulation materials (LCMs), LCM pills, and cement squeeze, several innovative solutions have been introduced to mitigate and treat the lost circulation events, including wellbore strengthening, managed pressure drilling (MPD), and casing while drilling (CwD). However, selecting the suitable LCM and the optimized fluid formulation is vital to the lost circulation remedy's success.
The primary objective of this study is to evaluate the effectiveness of smart material in treating lost circulation in geothermal formations. The smart material is a shape memory polymer (SMP) that can be programmed to activate by formation temperature. Once activated, its particle size increases to plug large fractures and stop mud losses. SMP evaluation is conducted on a large-scale flow loop under high-temperature conditions (above 300℉). The experimental evaluation includes SMP activation with temperature, fluid rheology, wellbore hydraulics, fracture sealing, and SMP transportation in the annulus under a broad range of operating parameters, such as concentration, drill pipe rotational speed, inclination angle, and flow rate. Furthermore, a computational fluid dynamics (CFD) study was conducted to upscale the experimental results and predict the LCM transportation in the annular section of geothermal wells in the field dimensions.
The novelty of this study is the use of shape memory polymer that has the potential to plug large fractures in geothermal formations with minimal risk of plugging downhole drilling and logging tools. Moreover, this study introduces a novel experimental setup to test different lost circulation materials under high-temperature conditions. The introduced large-scale testing is unique and crucial in LCM evaluation to ensure better results in geothermal fields
Top-down Thermal Proteome Profiling using Mass Spectrometry on Human Plasma Proteins
No full textProteins play vital roles in biological systems; hence, the characterization of protein function is crucial for better understanding of complex cellular systems. The study of biological function of proteins at the proteome level is called functional proteomics. One way to probe functionality of proteins is to measure stability via the measurement of protein unfolding as a result of chemical or thermal denaturation. My research is focused on developing and applying novel thermal proteome profiling (TPP) technologies to measure protein thermal stability by heating proteins using a temperature gradient.
Traditionally, TPP is conducted using a bottom-up approach in which proteins are enzymatically digested into peptides for mass spectrometry (MS) analysis. Here, a top-down TPP approach was applied to analyze the stability of intact proteoforms and observe the effect of protein modifications that might be obscured due to protein digestion. We applied this top-down TPP method to human plasma to observe the stability of functional proteoforms of low-abundance, clinically relevant proteins. However, it is challenging to elucidate such proteins in human plasma because high-abundant proteins often mask the detection of low-abundant proteins. We systematically evaluated and optimized a depletion protocol to efficiently remove high-abundant proteins in human plasma samples prior to proteomics analysis. Overall, my research illustrates the feasibility of studying functional proteomes in human plasma through a novel platform that integrates the depletion of high-abundant proteins and top-down TPP applications
"They're Not Little Kids, They're Tiny Humans": Liberating and Humanizing Students Through Culturally Responsive Pedagogy
Get PDFThere is a continuing need for teachers who understand the value that culturally responsive pedagogy can bring to a classroom. To better understand how to develop classrooms that are responsive to their students’ cultural identities, it is important for teachers to learn from those who have done so successfully. Based on this need, I interviewed four elementary teachers who had reputations for effectively engaging in culturally responsive practices with their students. Specifically, I used critical qualitative methods to learn about these teachers’ culturally responsive perspectives and practices. My participants’ responses suggested that culturally responsive teachers: (1) are attuned to students’ needs; (2) critically discern classroom content; (3) affirm student autonomy; (4) possess an asset-based perspective; and (5) utilize an insider’s understanding of historical or institutional disparity. Additionally, my interpretive lenses, based on the work of Paulo Freire and Luis Moll et al., illustrate and explain how culturally responsive teachers utilize local funds of knowledge to liberate and humanize their students. At the conclusion of the study, I explore possible implications for current educators, principals and district administrators, and other people involved with efforts to promote diversity, equity, and inclusion. Implications for this study acknowledge the unique characteristics of the participants that cannot be replicated by every teacher
