Browsing by Author "Zhang, Yunyan"
Now showing 1 - 10 of 10
Results Per Page
Sort Options
- ItemOpen AccessAdvanced MRI methods for probing disease severity and functional decline in multiple sclerosis(2023-12-14) Oladosu, Olayinka Adeoluwa; Zhang, Yunyan; Joshi, Manish; Dunn, Jeffrey Frank; Frayne, Richard; Le, Lawrence Trong-HuanMultiple sclerosis (MS) is a chronic and severe disease of the central nervous system characterized by complex pathology including inflammatory demyelination and neurodegeneration. MS impacts >2.8 million people worldwide, with most starting with a relapsing-remitting form (RRMS) in young adulthood, and many of them worsening to a secondary-progressive course (SPMS) despite treatment. So, there is a clear need for improved disease characterization. MRI is an ideal tool for non-invasive assessment of MS pathology, but there is still no established measure of disease activity and functional consequences. This project aims to overcome the challenge by developing novel imaging measures based on brain diffusion MRI and phase congruency texture analysis of conventional MRI. Through advanced modeling and analysis of clinically feasible brain MRI, this thesis investigates whether and how the derived measures differentiate MS pathology types and disease severity and predict functional outcomes in MS. The overall process has led to important technical innovations in several aspects. These include: innovative modeling of simple diffusion acquisitions to generate high angular resolution diffusion imaging (HARDI) measures; new optimization and harmonization techniques for diffusion MRI; innovative neural network models to create new diffusion data for comprehensive HARDI modeling; and novel methods and a graphic user interface for optimizing phase congruency analyses. Assisted by different machine learning methods, collective findings show that advanced measures from both diffusion MRI and phase congruency are highly sensitive to subtle differences in MS pathology, which differentiate disease severity between RRMS and SPMS through multi-dimensional analyses including chronic active lesions, and predict functional outcomes especially in physical and neurocognitive domains. These results are clinically translational and the new measures and techniques can help improve the evaluation and management of both MS and similar diseases.
- ItemOpen AccessAn Investigation of Deep Learning Methods to Shorten GABA-edited Magnetic Resonance Spectroscopy Scan Times(2023-09-18) Pommot Berto, Rodrigo; Medeiros de Souza, Roberto; Harris, Ashley; Lebel, Marc; Zhang, Yunyan; Frayne, Richard; McCreary, CherylEdited magnetic resonance spectroscopy (MRS) can provide localized information on gamma-aminobutyric acid (GABA) concentration in vivo. However, GABA-edited MRS data has a low spectral quality, and many measurements, known as transients, need to be collected and averaged to obtain a high-quality spectrum, resulting in long scan times. This work investigated using deep learning (DL) with only a quarter of the number of conventionally acquired transients to shorten scan times by four while maintaining or improving spectral quality. A proof of concept was demonstrated by reconstructing GABA-edited spectra with only 80 transients and different configurations of DL-based pipelines. The best-performing pipeline used a proposed dimension-reducing 2D U-NET variation and it obtained better spectral quality metrics than conventionally reconstructed spectra with 320 transients. Simulated data was also shown to be useful in pre-training DL model weights. An open data challenge for reconstructing GABA-edited spectra from reduced transients was organized, and various DL models from different participating teams were compared. The challenge results reinforced the proof of concept conclusions that higher spectral quality can be achieved with DL reconstructions. However, the challenge metric evaluation also showed that DL models are able to undesirably exploit the limitations of conventional MRS metrics when using those as the training loss for the models, leading to good metric values but poor reconstructed spectra quality. DL reconstructions of GABA-edited MRS with 80 transients were also quantified and had significant differences from results from conventional reconstructions with 320 transients. However, given the lack of ground truths in the quantified data, it is not possible to conclude which results are closer to the actual concentrations. This work showed that DL methods can reduce GABA-edited MRS scan times while increasing spectral quality. Due to the lack of ground truths for the in vivo data, further studies are necessary to validate the concentrations obtained from the DL-based GABA-edited MRS reconstructions in comparison to conventional methods. This work was developed in the spirit of open science, and the data and code to reproduce the results were made publicly available.
- ItemOpen AccessDeep Learning for Domain-Invariant Magnetic Resonance Carotid Artery Wall Segmentation(2019-04-26) Danko, Anna M.; Frayne, Richard; Souza, Roberto Medeiros De; Rittner, Leticia; Zhang, YunyanSegmentation of the carotid arteries is a prerequisite to image processing techniques that are applied to medical images to assess the features of atherosclerosis, a disease which can lead to ischemic stroke. Carotid artery segmentation is currently mainly done manually in a time-consuming processing. In this work deep learning approaches were applied to carotid artery segmentation. Additionally, the influence of image contrast on segmentation performance was explored, and whether a network could be taught to learn domain-invariant features including the use of adversarial methods. Non-adversarial and adversarial methods were successfully demonstrated.
- ItemOpen AccessDeformable Image Registration Using Attentional Generative Adversarial Networks(2021-01-22) Zhou, Hanchong; Leung, Henry; MacDonald, M. Ethan; Hemmati, Hadi; Zhang, YunyanDeformable image registration is a fundamental process that aims to estimate non-linear spatial correspondence between input images. Medical image registration serves wildly on clinical treatment evaluation, monitoring disease and tracking disease. Conventional registration algorithms iteratively optimize a similarity function for each pair of images, which result in long registration time. Learning based registration method typically use convolutional neural networks to learn features automatically during training and register an image pair in one shot. Generative adversarial network is a novel structure that involves a generator and a discriminator, where the discriminator encourages the former to generate better results. Predicting deformation between brain magnetic resonance images is a complicated task because of their high-dimensional non-linear transform. A generative model leveraging is proposed using an attentional mechanism to estimate the complicated deformation field, and train the model using perceptual cyclic constraints. As an unsupervised method, our model dose not need any labels for training. Experimental results show quantitative evidence that the proposed method can predict reliable deformation field at a fast speed.
- ItemOpen AccessDeveloping TomoNet: A Deep Learning Feature Extractor for Medical Image Classification(2023-06) Guerra - Librero Camacho, Javier; Frayne, Richard; Pinheiro Bento, Mariana; Pichardo, Samuel; Zhang, Yunyan; Garcia Flores, Julio Garcia; Whelan, PatrickIn this thesis, I investigated whether incorporating medical images in the training of feature extractors would enhance their performance in medical imaging tasks. To achieve this objective, I conducted two studies. In the first study, I compared the performance of a feature extractor pretrained with natural images from ImageNet (referred to as ImageNet-derived) with a feature extractor trained from scratch without any pre-existing feature extraction capabilities. Both feature extractors were trained to classify medical images into nineteen categories. The results demonstrated that the performance of the feature extractor trained from scratch approached the performance of the ImageNet-derived feature extractor as the size of the training set increased. Additionally, I introduced a metric called the "transfer learning gap" to quantify the relative amount of knowledge transfer in transfer learning. The experiments revealed that training the feature extractor twice, first with the ImageNet dataset and then with 160,000 medical images, resulted in a better-performing deep learning model. This was an unexpected result and likely due to training with an insufficient number of medical images. Thus, I named the better performing, twice-trained feature extractor "TomoNet" and used it in my subsequent studies. In the second study, I compared the performance of TomoNet with that of the ImageNet-derived feature extractor in classifying medical images based on sex. This study failed to provide conclusive evidence that TomoNet, the twice-trained feature extractor, outperformed the ImageNet-derived feature extractor. Both models exhibited signs of overfitting, and there was potential evidence of catastrophic interference in TomoNet results. Finally, I presented an initial business model for a company aiming to commercialize products derived from the TomoNet studies. The business model focused on the commercialization of "Petal-Blue," a deep learning model designed to detect white matter hyperintensities associated with dementia in brain imaging. The Canvas Business Model framework was utilized to evaluate crucial aspects of the startup, including value proposition, customer segments, revenue streams, and partnerships.
- ItemOpen AccessEvaluating the utility of advanced MRI methods for monitoring structural changes in demyelinated lesions using two models of multiple sclerosis(2019-07-30) Hossain, Md. Shahnewaz; Zhang, Yunyan; Goodyear, Bradley Gordon; Burton, Jodie M.The ability to precisely evaluate tissue pathologies and their functional correlates has been an ongoing challenge in patients with multiple sclerosis (MS). MS pathology is complex; however, much of them start from or are mediated by a demyelinating event. In this thesis, I have studied 2 common models of MS: a cuprizone mouse model, and an optic neuritis (ON) human model. I have particularly focused on the investigation of the potential of novel advanced magnetic resonance imaging (MRI) techniques including neurite orientation dispersion and density imaging (NODDI), diffusion tensor imaging (DTI), and MRI texture analysis. The first model evaluates the ability of these methods to assess the time course and regional preference of MS-like pathology following induced demyelination and spontaneous remyelination in mouse brain. The second model tests the feasibility of select imaging measures for detecting structural changes in the optic nerves and correlating clinical measures in acute optic neuritis (AON) as part of a clinical trial of high dose vitamin D. Through a focused study of the corpus callosum over an extended time series, the animal study shows that all MRI metrics have detected the expected changes over the de- and remyelination periods, consistent with histology quantified using a texture method, structure tensor analysis. The NODDI metric neurite density index is specific to myelin integrity, NODDI orientation dispersion index to axonal changes, and texture angular entropy specific to both myelin and axonal changes. Moreover, early de- and remyelination seems to occur in the genu of corpus callosum featuring relatively thin and high-density axons and early demyelination but relatively late repair in the splenium showing large calibre and comparably low-density axons. All NODDI metrics appear to outperform DTI metrics. In a clinical setting, the advanced imaging measures have successfully detected the structural changes in the affected versus non-affected optic nerves and angular entropy correlates with patient disability. Collectively, this thesis suggests that advanced MRI measures are powerful indices of MS-like pathology and could be used clinically for monitoring disease development and treatment responses, deserving further validation.
- ItemOpen AccessEvaluation of Injury and Repair in Multiple Sclerosis Using Advanced MRI Methods(2023-01-16) Hosseinpour, Zahra; Zhang, Yunyan; Pike, Bruce G.; Yong, Wee V.Multiple sclerosis (MS) is an inflammatory demyelinating and neurodegenerating disease of the central nervous system impacting more than 2.8 million people worldwide. Many of the people experience paramount disability after 10-15 years of disease onset. Optimal management requires accurate measurement of tissue pathology, including the likelihood of repair in lesions. However, there is no established marker of lesion severity in vivo. This project aimed to develop methods to characterize tissue injury and repair as seen in focal lesions based on brain magnetic resonance imaging (MRI) of relapsing-remitting MS (RRMS). The focus was on image processing techniques ranging from development, validation, to application. Initially, based on histology-informed MRI of postmortem MS brains, I conducted texture analysis using an optimized method known as gray level co-occurrence matrix (GLCM) and compared texture analysis to advanced MRI measures using machine learning models for classifying MS pathology, including de- and re-myelinated lesions. Based on the selected MRI measures, I then developed a percentile approach for characterizing MS lesion severity in clinical MRI, and for assessing lesion recovery in clinical trial MS participants. Overall, brain MRI texture measures performed the best in differentiating de- and re-myelination. These measures characterized 2 extreme types of MS lesions on de- and re-myelination, which differentiated men from women, and detected significant recovery in acute MS lesions with treatment. Collectively, advanced texture analysis in clinical MRI is promising for characterizing lesion injury and repair in MS. This ability is critical for improved evaluation of both disease activity and treatment response for MS participants.
- ItemOpen AccessMagnetic resonance imaging derived biomarkers in multiple sclerosis(2007) Zhang, Yunyan; Mitchell, J. RossSeveral mechanisms contribute to the complex pathophysiology of multiple sclerosis (MS). This calls for biomarkers to characterize the heterogeneity of MS in individual subjects. The advancement in magnetic resonance imaging (MRI) technology over past decades makes it the most attractive biomarker candidate in MS. Texture analysis was applied to quantify subtle abnormalities on the MRI from people with MS. A classical statistical method (gray level co-occurrence matrix) was able to detect a tread yet not significant difference of therapeutic responses between different tissue types. A novel time-frequency based technique (polar Stockwell Transform, PST) was then investigated. The increased PST texture was shown to represent inflammation and demyelination in a murine model of MS. PST texture altered significantly when tissue MRI evolved from normal appearing white matter to MS lesions. It decreased significantly when an active lesion became inactive. This suggests that the PST texture may be a MRI surrogate to characterize subtle and early abnormalities existed in MS tissue. MS is also a neurodegenerative disease in the central nervous system (CNS) affecting both white matter and grey matter (GM). Abnormally decreased signal intensity (black T2, BT2) in the deep GM of MS patients was noted on 1.5 T MRI. Due to the iron deposition phenomenon in its underlying tissue, 3 T MRI was more sensitive to BT2 detection than 1.5 T MRI. A higher correlation was obtained between patient disability and deep GM BT2 measured at 3 T than that at 1.5 T in MS. The association of 3 T BT2 with EDSS was greater than most of the MRI derived indices in MS. The small sample size limits broad conclusions. However, these results encourage further investigation of the clinical significance of BT2 as a biomarker for disability in MS at high field. This thesis provides evidence for proof of concept that the new developed techniques may have potential to become surrogate measures in MS diagnostics and treatment. However, these techniques need to be combined with other MRI matrices in order to obtain an integrated picture of MS.
- ItemOpen AccessMRI Texture Analysis in Multiple Sclerosis(Hindawi Publishing Corporation, 2011-09-06) Zhang, Yunyan
- ItemOpen AccessMRI Texture Analysis in Multiple Sclerosis(2011-11-16) Zhang, YunyanMultiple sclerosis (MS) is a complicated disease characterized by heterogeneous pathology that varies across individuals. Accurate identification and quantification of pathological changes may facilitate a better understanding of disease pathogenesis and progression and help identify novel therapies for MS patients. Texture analysis evaluates interpixel relationships that generate characteristic organizational patterns in an image, many of which are beyond the ability of visual perception. Given its promise detecting subtle structural alterations texture analysis may be an attractive means to evaluate disease activity and evolution. It may also become a new tool to assess therapeutic efficacy if technique issues are resolved and pathological correlates are further confirmed. This paper describes the concept, strategies, and considerations of MRI texture analysis; summarizes applications of texture analysis in MS as a measure of tissue integrity and its clinical relevance; then discusses potentially future directions of texture analysis in MS.