No description available.

Convolution neural networks have achieved a state of the art accuracy for multi-class segmentation in biomedical image science. In this thesis, a 2-Stage binary 2D UNet and MultiResUNet are used to segment the 3D cardiac CT Volumes. 3D volumes have been sliced into 2D images. The 2D networks learned to classify the pixels by transforming the information about the segmentation into latent feature space in a contracting path and upsampling them to semantic segmentation in an expanding path. The network trained on diastole and systole timestamp volumes will be able to handle much more extreme morphological differences between the subjects. Evaluation of the results is based on the Dice coefficient as a segmentation metric. The thesis work also explores the impact of the various loss function in image segmentation for the imbalanced dataset. Results show that2-Stage binary UNet has higher performance than MultiResUnet considering segmentation done in all planes. In this work, Convolution neural network prediction uncertainty is estimated using Monte Carlo dropout estimation and it shows that 2-Stage Binary UNet has lower prediction uncertainty than MultiResUNet.

<p>Vocal fold vibration results in voice production. Optimal hydration levels contribute to self-sustained vocal fold vibration and preservation of voice quality. Adequate hydration is implicated as a factor in maintaining voice and preventing voice problems. Voice problems affect up to one third of adults during their lifetime. But whether altered hydration state adversely affects vocal fold biology and biomechanics is still unclear. To untangle the effects of systemic dehydration on vocal fold biology, we developed a water restriction protocol on lab animals that can translate to humans. Our results showed that dehydration induced by restricted water access downregulated the gene expression of IL-1α and desmoglein-1, upregulated the gene expression level of hyaluronidase-2, and downregulated hyaluronic acid (HA).</p> <p>Clinically, hydration treatments are hypothesized to maintain the viscoelastic properties of vocal folds. However, our understanding of the relationship between vocal fold tissue hydration level and biomechanical properties is still evolving. To investigate the effects of dehydration on biomechanical properties we used an ex vivo experimental design. We hypothesized that the optimal stiffness of vocal folds would be impacted after dehydration via losing both water and HA, but that the stiffness properties would recover through rehydration. To test this hypothesis, we experimentally treated porcine vocal fold samples using two different approaches: 1) immersion in hypertonic solution (15% NaCl in ddH2O) and PBS sequentially to mimic dehydration and rehydration, and 2) incubation with hyaluronidase (Hyal) to mimic HA loss during dehydration. Our results showed that loss of water increased tissue stiffness and could be recovered through rehydration in a certain degree. In addition, loss of HA increased tissue stiffness. </p> <p>In While dehydration decreases total body blood volume, different tissues and organs of the body may be impacted in different ways from dehydration. Therefore, it is important to investigate the hemodynamic alterations during changes to hydration status. Magnetic resonance angiography (MRA) and ultrasound imaging were employed to identify the delicate vascular geometry and hemodynamics of the laryngeal blood supply. Animals underwent both MRA and ultrasound imaging at baseline, dehydration and rehydration time points. Our results showed that dehydration impacted the blood supply to larynx. This blood supply was restored through rehydration treatment.</p> <p>Overall, this research has been successful in establishing a mild dehydration animal model, providing evidence from gene and protein levels that dehydration affects cytokine production and extracellular matrix components (ECM) in vocal fold, demonstrating the vocal fold tissue biomechanical behavior after dehydration and loss of HA, and offering a combination application of MRA and ultrasound imaging to study vascular geometry and hemodynamics of the blood supply to the vocal fold region.</p>

<p>The work in this thesis deals with pre-clinical development of rapid in vivo ³¹P mag- netic resonance spectroscopy (MRS) techniques. Current MRI literature of ³¹P spec- troscopy presents evidence of increased concentrations of phosphomonoesters (PME), and phosphodiester (PDE) as well as inorganic phosphate concentrations in tumor tissue. Human breast cancer studies have demonstrated correlation between disease progression and both PME and PDE peaks. Furthermore, ³¹P MRS can be used to detect, grade tumours and monitor response to chemo and radiation therapy.<br />Tumor measurements are typically static (i.e. single measurement per scan). In other experiments, on muscle for example, dynamic measures are required the purpose of which is to assess temporal function and recovery. In all ³¹P acquisitions there are problems surrounding RF coil design, pulse sequence speed, localization and system calibration. The work presented here focused on improving all these aspects and provide easy and reliable work flow to use ³¹P MRS in a clinical setting.<br />One of the aspects of this thesis lies in designing and construction of an RF coil that is well suited for integration with a clinical MRI breast imaging and biopsy system. The designed coil was tuned for simultaneous operation at ³¹P (51.73 MHz) and ¹H (127.88MHz) Larmor frequencies. This design has advantages in the fact that complex pulse sequences with heteronuclear decoupling could be performed easily. The additional features of the coil design is that it is possible to swap it into the breast imaging system without moving the patient. Along with the designed coil, custom software was written to assist with transmit gain calibration of ³¹P RF pulses, to ensure maximum MR signal. The automated prescan ensures easy work flow and minimizes the operator variability and patient time inside the MR scanner.<br />Another aspect of this thesis deals with rapid pulse sequence development, to further speed up the ³¹P MRS data acquisition. Echo planar spectroscopic imaging (EPSI) with a fly–back gradient trajectory is currently one of the most reliable and robust techniques for speeding up chemical shift imaging (CSI) acquisitions. A ³¹P EPSI sequence was written to acquire spectroscopic imaging data at 1, 2 and 2.6 cm spatial resolution and spectral bandwidth of 3125 Hz. The sequence showed an ability to speed up data acquisition up to 16 times, where SNR permits.<br />Phantom studies were used to verify the double tuned coil and EPSI sequence en- suring proper and safe operation. In vivo measurements of an exercising muscle demonstrated the ability of ³¹P EPSI to play an important role in rapidly acquiring spatially localized ³¹P spectroscopic data.<br />With these preclinical developments in place a clinical trial is possible using ³¹P MRS rapidly and efficiently. Furthermore the increased usability of ³¹P MRS provided by the tools developed in this thesis can prove to be beneficial by integrating ³¹P MRS into existing clinical protocols.</p> / Doctor of Science (PhD)

Selection of an appropriate structure for storage and analysis of complex datasets is a vital but often overlooked decision in the design of data mining and machine learning experiments. Most present techniques impose a matrix structure on the dataset, with rows representing observations and columns representing features. While this assumption is reasonable when features are scalar and do not exhibit co-dependence, the matrix data model becomes inappropriate when dependencies between non-target features must be modeled in parallel, or when features naturally take the form of higher-order multilinear structures. Such datasets particularly abound in functional medical imaging modalities, such as fMRI, where accurate integration of both spatial and temporal information is critical. Although necessary to take full advantage of the high-order structure of these datasets and built on well-studied mathematical tools, tensor analysis methodologies have only recently entered widespread use in the data mining community and remain relatively absent from the literature within the biomedical domain. Furthermore, naive tensor approaches suffer from fundamental efficiency problems which limit their practical use in large-scale high-order mining and do not capture local neighborhoods necessary for accurate spatiotemporal analysis. To address these issues, a comprehensive framework based on wavelet analysis, tensor decomposition, and the WaveCluster algorithm is proposed for addressing the problems of preprocessing, classification, clustering, compression, feature extraction, and latent concept discovery on large-scale high-order datasets, with a particular emphasis on applications in computer-assisted diagnosis. Our framework is evaluated on a 9.3 GB fMRI motor task dataset of both high dimensionality and high order, performing favorably against traditional voxelwise and spectral methods of analysis, discovering latent concepts suggestive of subject handedness, and reducing space and time complexities by up to two orders of magnitude. Novel wavelet and tensor tools are derived in the course of this work, including a novel formulation of an r-dimensional wavelet transform in terms of elementary tensor operations and an enhanced WaveCluster algorithm capable of clustering real-valued as well as binary data. Sparseness-exploiting properties are demonstrated and variations of core algorithms for specialized tasks such as image segmentation are presented. / Computer and Information Science

Diffusion Tensor Imaging (DTI) is a magnetic resonance imaging (MRI) technique used to measure in-vivo anisotropic water diffusion. This can give useful information regarding white matter integrity and has the potential to provide important biomarkers in spinal cord injury. One of the largest challenges in DTI of the spinal cord is the presence of metal which causes geometric distortions, signal pile-up, and signal voids. Because most patients with spinal cord injury have some amount of metal hardware implanted for stabilization, it is important to confront issues involving metal as DTI of the spinal cord becomes more widely examined. This study examined the characteristics of metal artifact in DTI images for several spinal surgical implants via imaging of phantoms constructed with implements suspended in agar gel to provide a homogeneous surrounding medium for analysis. A cervical spine phantom implanted with pedicle screws was also used to simulate in-vivo imaging. Optimization of the DTI sequence was also considered using different metal artifact reduction techniques including view-angle-tilting, slice thickness, and field of view size. Minor reduction in metal artifact was achieved using these techniques. The resulting image data shows that imaging near metal may be feasible in some circumstances, particularly when implantation is minimal. Also, using the cervical spine phantom it was shown that it should be possible to acquire DTI data close to the location of metal implants and thus examine DTI values of the injured spinal cord superior to the injury site. / Mechanical Engineering

Three-dimensional analysis of mandibular growth provides the potential for pedodontists, orthodontists and surgeons to prescribe treatment that works in harmony with the individual growth of the patient. Despite efforts by 3D pioneers, the visualization of growth and development remains reminiscent of the landmark-based cephalometric analyses used in two-dimensional studies. The objective of this study was to identify 3D topographical changes of the mandible during growth and development of adolescent orthodontic patients. Nine pairs of pre- and post-orthodontic cone-beam computed tomography (CBCT) scans were used to generate mandibular surfaces. Surfaces were superimposed on trabecular bone in the anterior mandible using a mutual information algorithm, and topographical changes were visualized and quantified. The intra- and inter-rater intraclass correlation coefficients for surface generation (0.94 and 0.93, respectively) and superimposition (0.96 and 0.82, respectively) demonstrate the reliability of the techniques. The findings of this study support the theories of bone remodeling reported in histological, implant-based, and landmark studies of mandibular growth. / Oral Biology

Application of deep learning in radiology has the potential to automate workflows, support radiologists with decision support, and provide patients a logic-based algorithmic assessment. Unfortunately, medical datasets are often not uniformly distributed due to a naturally occurring imbalance. For this research, a multi-classification of liver MRI sequences for imaging of hepatocellular carcinoma (HCC) was conducted on a highly imbalanced clinical dataset using deep convolutional neural network. We have compared four multi classification classifiers which were Model A and Model B (both trained using imbalanced training data), Model C (trained using augmented training images) and Model D (trained using under sampled training images). Data augmentation such as 45-degree rotation, horizontal and vertical flip and random under sampling were performed to tackle class imbalance. HCC, the third most common cause of cancer-related mortality [1], can be diagnosed with high specificity using Magnetic Resonance Imaging (MRI) with the Liver Imaging Reporting and Data System (LI-RADS). Each individual MRI sequence reveals different characteristics that are useful to determine likelihood of HCC. We developed a deep convolutional neural network for the multi-classification of imbalanced MRI sequences that will aid when building a model to apply LI-RADS to diagnose HCC. Radiologists use these MRI sequences to help them identify specific LI-RADS features, it helps automate some of the LIRADS process, and further applications of machine learning to LI-RADS will likely depend on automatic sequence classification as a first step. Our study included an imbalanced dataset of 193,868 images containing 10 MRI sequences: in- phase (IP) chemical shift imaging, out-phase (OOP) chemical shift imaging, T1-weighted post contrast imaging (C+, C-, C-C+), fat suppressed T2 weighted imaging (T2FS), T2 weighted imaging, Diffusion Weighted Imaging (DWI), Apparent Diffusion Coefficient map (ADC) and In phase/Out of phase (IPOOP) imaging. Model performance for Models A, B, C and D provided a macro average F1 score of 0.97, 0.96, 0.95 and 0.93 respectively. Model A showed higher classification scores than models trained using data augmentation and under sampling. / Thesis / Master of Science (MSc)

Bakgrund Magnetresonanstomografi undersökningar blir mer frekventa och därmed ökar användningen av gadoliniumbaserade kontrastmedel. Dock är gadoliniumretention ett relativt nyupptäckt fenomen och dess påverkan är fortfarande relativt okänt. Syfte Syftet med denna litteraturstudie är att få en djupare förståelse om vad gadoliniumretention är och dess inverkan på människokroppen, samt vilka implementeringar röntgensjuksköterskorkan behöva göra för att anpassa vården efter ny forskning inom området. Metod Denna studie är en systematisk litteraturstudie. Datainsamlingen har gjorts via databaserna PubMed och CINAHL. Endast vetenskapliga artiklar som uppfyllt inklusionskriterierna, exempelvis att de är peer-reviewed och primärpublicerade, har inkluderats. De artiklar som har besvarat frågeställningarna har kvalitetsgranskats och sedan har artiklarna som uppnådde medel eller hög kvalitet analyserats. Resultat Resultatet av denna studie visar att ackumulerad gadoliniumretention kan uppstå efterupprepade administreringar av gadoliniumbaserade kontrastmedel. Inga negativa biverkningar hos människokroppen har påvisats. För att minska risken för gadoliniumretention rekommenderar flera studier att minska kontrastmedelsdoser och vara varsam vid administrering. Slutsats Gadoliniumretention kan ses som förhöjd signalintensitet främst i dentate nucleus och globuspallidus i hjärnan efter upprepade administreringar av gadoliniumkontrastmedel. Dess påverkan på människokroppen är fortfarande okänd och mer forskning behövs om hurröntgensjuksköterskor kan anpassa vården därefter. / Background Magnetic resonance imaging exams is becoming more common and therefore the use of gadolinium-based contrast media increases. However, gadolinium retention is a relatively newly discovered phenomenon, and its impact on the human body is still unknown. Purpose The purpose of this study is to obtain a deeper understanding of what gadolinium retention is and how it affects the human body, and which implementations radiographers may have to do to adapt the care according to new research within the field. Method This study is a systematic literary review. The data collection was made from the databases PubMed and CINAHL. Only scientific studies that fulfilled the inclusion criteria, for example that they were peer-reviewed and primary published, were included. The articles that answered the questions were quality reviewed and the articles that achieved medium or high quality were analysed. Results The result of this study shows thar accumulated gadoliniumretention can arise after repeated administrations of gadolinium-based contrast media. No adverse effects in the human body have been demonstrated. To reduce the risk of accumulating gadoliniumretention several studies recommend reducing the dosage and to be cautious when administrating. Conclusion Gadoliniumretention can be seen as an increased signal intensity, mainly in dentate nucleus and globus palladius in the human brain after multiple administration of gadolinium-based contrast media. Its adverse effects on the human body are still unknown and more research is needed on how radiographer can adapt the care thereafter.

Yes / When used in routine clinical practice, Patient Reported Outcomes Measures (PROMS) can give patients tacit approval to discuss embarrassing topics, which could be particularly useful for urological nursing. The aim of this study was to assess whether it would be feasible to use two such measures for penile cancer; one for body image (the Male Genital Self-Image Scale; MGSIS-5) and another for lymphedema (the Groin and Lower Limb Lymphedema questionnaire; G3L-20). Study packs were posted to penile cancer patients who had received (i) sentinel node biopsy only, (ii) inguinal node dissection only, and (iii) inguinal node dissection with post-operative radiotherapy with or without chemotherapy. The two measures (MGSIS-5 and G3L-20) were complemented with those specific to sexual function (IIEF) and cancer (EORTC-QLQ-C30 version 3) and a modified Lymphoedema Genitourinary Cancer Questionnaire (mLGUCQ). Twenty patients returned questionnaires. Validity and reliability analyses are presented but low participant numbers mean that results need treating with caution. Results show sufficient feasibility for the MGSIS-5 and the G3L-20 to warrant another study to attract larger numbers of participants, either over a longer time frame or at multiple sites. In these further studies, we would recommend adding (1) more Likert responses, (2) the timeframe to the MGIS and (3) exploring either the use of sexual desire psychometric measures or the addition of sexual desire items to the MGSIS for this patient group.