Automated classification of medical images for colorectal and prostate cancer diagnosis is a crucial tool for improving routine diagnosis decisions. Therefore, in the last few decades, there has been an increasing interest in refining and adapting machine learning algorithms to classify microscopic images of tumour biopsies. Recently, multispectral imagery has received a significant interest from the research community due to the fast-growing development of high-performance computers. This thesis investigates novel algorithms for automatic classification of colorectal and prostate cancer using multispectral imagery in order to propose a system outperforming the state-of-the-art techniques in the field. To achieve this objective, several feature extraction methods based on image texture have been investigated, analysed and evaluated. A novel texture feature for multispectral images is also constructed as an adaptation of the local binary pattern texture feature to multispectral images by expanding the pixels neighbourhood to the spectral dimension. It has the advantage of capturing the multispectral information with a limited feature vector size. This feature has demonstrated improved classification results when compared against traditional texture features. In order to further enhance the systems performance, advanced classification schemes such as bag-of-features - to better capture local information - and stacked generalisation - to select the most discriminative texture features - are explored and evaluated. Finally, the recent years have seen an accelerated and exponential rise of deep learning, boosted by the advances in hardware, and more specifically graphics processing units. Such models have demonstrated excellent results for supervised learning in multiple applications. This observation has motivated the employment in this thesis of deep neural network architectures, namely convolutional neural networks. Experiments were also carried out to evaluate and compare the performance obtained with the features extracted using convolutional neural networks with random initialisation against features extracted with pre-trained models on ImageNet dataset. The analysis of the classication accuracy achieved with deep learning models reveals that the latter outperforms the previously proposed texture extraction methods. In this thesis, the algorithms are assessed using two separate multiclass datasets: the first one consists of prostate tumour multispectral images, and the second contains multispectral images of colorectal tumours. The colorectal dataset was acquired on a wide domain of the light spectrum ranging from the visible to the infrared wavelengths. This dataset was used to demonstrate the improved results produced using infrared light as well as visible light.

Wound healing is a complex and coordinated response where the skin repairs itself. It can be divided into four phases: inflammation, proliferation, maturation and remodelling. The prolongation or failure of one of these phases may result in a chronic wound condition due to a healing delay or a non-healing wound. Failure of wound healing is an unsolved and significant public health problem, associated with economic and human costs. However, skin tissue regeneration may provide therapies which enhance wound healing. Despite the fact that a variety of scaffolds have been fabricated for wound healing, an increasingly aging and diabetic (as major risk factor for non-healing wounds) population is now creating a further extensive demand for new regenerative products. The first body of research examined in the present thesis evaluated the effects of propolis and its compounds, quercetin, rutin, and artepillin C on fibroblast proliferation and migration rates. In order to investigate this, cytotoxicity, membrane disruption, and in vitro scratch experiments were conducted. However, findings showed that there is not a compelling argument to pursue the use of propolis and its isolated components related to cell migration and proliferation; as a result, another natural product, Lucilia sericata maggot excretion/secretion (L. sericata maggot E/S) was investigated. The second and third bodies of research have involved the development of electrospun polycaprolactone (PCL) scaffolds in the micron scale range (2-14 μm) to allow good fibroblast infiltration. In vitro biocompatibility and infiltration behaviour of 3T3 GFP dermal fibroblasts, as an exemplar cell type on electrospun PCL scaffolds, were investigated. Additionally, synthesis of polycapolactone-poly(ethylene glycol)–block copolymer (PCL-b-PEG) was included in this study, where it was hypothesized that PCL-b-PEG containing scaffolds may enhance degradation, wettability, and cell-compatibility as previous research has reported. Caprolactone-poly(ethylene glycol) block copolymer were blended with commercial caprolactone (PCL-b-PEG/PCL) and electrospun scaffolds were developed and fully characterised. Selected scaffolds (PCL and PCL-b-PEG/PCL with an average fibre diameter of 4 μm) were compared in terms of: i) in vitro biocompatibility and cell infiltration, ii) mechanical properties, iii) physiochemical properties, iv) degradation rate, and v) porosity. The generated evidence suggested that PCL-b-PEG/PCL electrospun scaffolds with an average fibre diameter of 4 μm, with a porosity of~70% (mCT), average pore size of 40 μm and a Young’s modulus of ~9 MPa possess suitable properties to enhance the fibroblasts’ (murine and human) growth. In addition, PCL-b-PEG/PCL scaffolds allowed significantly higher fibroblast migration (60% after 5 days) compared with PCL scaffolds (40%). The third and fourth bodies of research focused on scaffolds as a platform for Lucilia sericata maggot excretion/secretion release as bioactive molecule for wound healing. The protein models (lysozyme and avidin) and the Lucilia sericata maggot excretion/secretion were absorbed onto the PCL, H-PCL and PCL-b-PEG/PCL scaffolds with and without cationic polymer poly (trimethylamino) ethyl methacrylate hydrochloride-block-Poly(N-(2-hydroxypropyl)methacrylamide (PTMAEMA33-b-PHPMA180) modification. The comparison of the release profiles of avidin from PCL, H-PCL and PCL-b-PEG/PCL scaffolds with and without out cationic polymer treatment showed that the release of avidin from cation polymer modified scaffolds, tend to exhibit zero order kinetics and is not time dependent. Meanwhile, scaffolds without cationic treatment showed a greater and prolonged burst release (first 6 h). Also, the release profile of lysozyme (smaller protein than avidin) shows that lysozyme released from PCL, H-PCL and PCL-b-PEG/PCL scaffolds without cationic polymer treatment was lower than the release from non-modified scaffolds. In addition, it was found that the bioactivity of lysozyme was not statistically reduced by the cationic modification of the scaffold surface. Finally, L. sericata maggot E/S was incorporated more efficiently on cationically modified PCL-b-PEG/PCL scaffolds than on scaffolds without cationic polymer. The loading and release of L. sericata maggot E/S from electrospun scaffolds has not been reported before. The release profiles of L. sericata maggot E/S revealed that the burst release and the cumulative release of L. sericata maggot E/S was significantly lower when cationic polymer was incorporated into the scaffolds. After 21 days, cumulative amount of released L. sericata maggot E/S was ~14 μg/mL from cationicallly modified scaffolds. No statistical differences were found between materials. In addition, the protease activity of Lucilia sericata maggot ES was preserved when PCL-b-PEG/PCL and cationically modified PCL-b-PEG/PCL scaffolds were used as release platform.

Current conventional cell culture methods for the expansion of cells to large quantities for tissue engineering and regenerative medicine application relies on the use of enzymatic digestion passaging. However, this method significantly reduces the cell viability, due to the destruction of important cell-surface proteins and therefore considered undesirable for the subsequent cell therapies [1]. Research has led to the development of thermo-responsive surfaces for the continuous culture of cells. These thermo-responsive materials properties can be used to passage cells from the surface when the cell culture temperature is reduced. Conventional cell culture is performed on a 2D platform, however this can result in undesired de-differentiation, due to lack of similarities to the natural extracellular matrix (ECM) environment. This thesis aims to develop a thermo-responsive 3D fibre-based scaffold, fabricated by electrospinning, to create an enzymatic-free 3D cell culture platform for the expansion of mammalian cells with the desired phenotype for clinical use. In particular the expansion of human corneal stromal cells (hCSCs), which have been observed to de-differentiate to their active myofibroblastic phenotype with conventional 2D cell culture methods were examined here. Although experimentally the electrospinning method is relatively simple, at the theoretical level the interactions between process parameters and their influence on the fibre morphology is not yet fully understood. In this study a design of experiments (DoE) model was proposed to determine combinations of process parameters that result in significant effects on poly-D,L-lactic acid (PDLLA) fibre morphology. The process parameters used in this study were applied voltage, needle-to-collector distance, flow rate and polymer concentration. Data obtained for mean fibre diameter, standard deviation of the fibre diameter (stdev, measure of fibre morphology) and presence of ‘beading’ on the fibres (beads per μm2) were evaluated as a measure of PDLLA fibre morphology. Uniform fibres occurred at standard deviations of ± 500 nm, ‘beads-on-string’ morphologies were apparent between ± 500-1300 nm and large beads were observed at ± 1300-1800 nm respectively. Mean fibre diameter was significantly influenced by the applied voltage and interaction between flow rate and polymer concentration. Fibre morphology was mainly influenced by the polymer concentration, while bead distribution was significantly influenced by the polymer concentration as well as the flow rate. The resultant DoE model regression equations were tested and considered suitable for the prediction of parameter combinations needed for the desired PDLLA fibre diameter and additionally, provided information regarding the expected fibre morphology. The thermo-responsive behaviour of the scaffolds was achieved by blend electrospinning of the thermo-responsive polymer poly(di(ethylene glycol) methyl ether methacrylate (PDEGMA) and the supportive polymer poly-D,L-Lactic Acid (PLA), which gives the support needed for cell culture. PDEGMA was synthesised by free radical polymerisation and has a lower critical solution temperature of 28°C. The optimised electrospinning process was taken forward for the fabrication of the blend electrospun PLA/PDEGMA scaffolds. Different weight (wt) % concentrations to PLA was incorporated and distribution of the PDEGMA was confirmed by fluorescent labeled PBIPDEGMA. Initial thermo-responsive passaging method was investigated with 3T3 fibroblasts (3T3s). No significant detachment of 3T3s was observed below 10 wt% of PDEGMA in the blend electrospun PLA/PDEGMA scaffolds, while a maximum of 25% of cells detached when cooled for 30 mins at 8°C as observed for 10 wt% PDEGMA. An increase in fibre diameter from 900 nm to 2-3 μm and different cooling methods (1 and 3 h at rt and controlled cooling of -1°C per minute) did not result in an increase of 3T3 detachment. These findings indicate possible insufficient presence of the PDEGMA thermo-responsive brushes at the surface of the fibres. However, when hCSCs were cultured on these scaffolds significantly higher cell detachment was observed (approx. 100% compared to 25% for 3T3s). These findings indicate the application of the blend electrospun PLA/PDEGMA scaffolds for the culture and passaging of hCSCs. hCSCs were observed to not adhere and proliferate on the PLA/PDEGMA scaffolds. Therefore, a peptide thermo-responsive co-polymer conjugated by photo-initiated thiolene chemistry was proposed. Poly(di(ethylene glycol) methyl ether methacrylate-copoly(di(ethylene-glycol) vinyl ether (PDEGMA/ PDEGOH) was synthesised by free radical polymerisation and subsequently functionalised with a free thiol group by a condensation reaction. The obtained poly(diethylene glycol) methyl ether methacrylate co-poly(diethylene glycol) thiol (PDEGMA/PDEGSH) was blend electrospun and the presence of free thiols on the electrospun scaffolds were determined by ToF-SIMS and FITC labeling. These free thiol containing scaffolds were functionalised by a norbornene acid functionalised peptide sequence GGG-YIGSR by the photo-initiated thiolene reaction and its presence was confirmed by ToF-SIMS and ATTO labeling. The biocompatibility of the peptide containing scaffolds was assessed by the adhesion, proliferation and immuno-staining of hCSCs. Significant increase in hCSCs adherence and proliferation was observed on the peptide containing scaffolds. Immuno-staining showed maintained expression of the desired phenotypic markers ALDH, CD34 and CD105, while showing no or low expression of the undesired phenotype expressing a-SMA marker. This desired expression was observed to be maintained after thermo-responsive passaging and observed higher when cells were cultured on PLA scaffolds with 10 wt% PDEGMA/4 mole% PDEGS-Nor-GGG-YIGSR. In this thesis, the fabrication and application of a first generation, biocompatible thermo-responsive peptide conjugate fibrous scaffolds is described. The ease of fabrication, successful adherence and expansion of a therapeutically relevant cell type make these scaffolds a promising new class of materials for the application of cell culture expansion platforms. The scaffolds developed and reported in this thesis are believed to represent a promising contribution to the fields of biomaterials and tissue engineering.

During the central nervous system (CNS) morphogenesis, chemical gradients of morphogens such as retinoic acid (RA) or sonic hedgehog play a central role in regulating CNS patterning and differentiation of neural subtypes. Recapitulation of these gradients in a 3D hydrogel matrix may provide a model for CNS tissue formation in vitro. 3D Printing technology offers an opportunity to reproduce the complex architecture of cell microenvironment. We have developed a 3D-printable alginate hydrogel bioink suitable for extrusion-based bioprinting. The bioink was characterised by shear thinning, high printing resolution and minimal adverse effects on cell viability. The bioink was successfully used to print mouse embryonic stem cells (mESCs)-laden constructs and supported their differentiation into neural-like cells. Extrusion-based bioprinting was used to 3D-print hybrid polycaprolactone (PCL)-alginate tubular scaffolds functionalised with a fluorescein isothiocyanate-conjugated bovine serum albumin (FITC-BSA) concentration gradient pattern. Quantification of the FITC-BSA concentrations in the scaffold showed a linear reduction in concentration as a function of scaffold’s distance (length). Tubular scaffolds printed with fibroblast-laden alginate supported cell viability and proliferation up to 6 days after printing. Next, the developed model was used to replicate the in vivo RA-induced directed differentiation of mESCs into spinal cord neurons. RA-concentration-dependent acquisition of neural identity was investigated using immunocytochemistry and flow cytometry analysis. RA promoted the formation of neurons with hindbrain and spinal cord identity and supressed the forebrain identity in a concentration-dependent manner. Among the investigated hydrogels, gelatine methacrylate (GelMA) supported neural differentiation and neurite outgrowth of the mESCs-derived embryoid bodies (EBs). Subsequently EBs-laden GelMA (5%) was successfully used as a bioink to print the hybrid PCL-hydrogel scaffolds. 3D Printing of EBs and RA-loaded GelMA in PCL scaffold induced differentiation of EBs into neurons with spinal cord positional identity. In conclusion, the model can be used for effective morphogens gradients delivery to replicate some of the complex processes of CNS development in vitro.

<p>Ultralyd er i dag et standardverktøy innen hjertediagnostikk. Analyse av ventrikkelfunksjon gjøres subjektivt ved å se på gråskala ultralydopptak av hjertet. Dette gjør dokumentasjon og opplæring vanskelig. Det er derfor ønskelig med kvantitative mål på ventrikkelfunksjon. Kvantitative metoder kan benytte høy tidsoppløsning og dermed også registrere flere detaljer.</p><p>To teknikker for kvantitativ ventrikkelanalyse er vevs-Doppler og speckle tracking (mønsterfølging). Ved bruk av disse teknikkene er det viktig å sikre kvaliteten av resultatene samtidig som analyseprosessen kan gjennomføres så effektivt som mulig. I denne avhandlingen presenteres flere metoder for å sikre kvalitet og øke effektivitet ved kvantitativ ventrikkelanalyse som alle er relatert til bruk av høy tidsoppløsning.</p><p>For å kunne gjøre effektiv ventrikkelanalyse behøves analyseverktøy. Et slikt verktøy som fungerer med både vevs-Doppler og speckle tracking presenteres i denne avhandlingen. Dette verktøyet har gjort flere store kliniske studier mulig. Ved analyse av hendelsene som skjer innen en hjertesykel er det viktig å dele opp hjertesykelen i ulike faser. Venstre ventrikkels tømmingsfase avsluttes ved at aortaklaffen lukkes, og denne hendelsen må derfor bestemmes før kvantitative parametre i slutten av tømmingsfasen og i overgangen til fyllingsfasen kan bestemmes. I denne avhandlingen vurderes og testes ulike kandidater for bestemmelse av lukkingen. To gunstige kandidater i hastighetskurver fra basale deler av ventrikkelen ble funnet: andre nullkryssing etter tømming og høyeste positive akselerasjon etter tømming før fylling. Et eget bidrag viser at den siste av disse kan bestemmes automatisk. For å oppnå høy tidsoppløsning kreves høy datarate. En populær metode for å øke datarate er flere parallelle mottaksstråler for hver sendestråle. I vanlig gråtoneavbilding av hjertet er det imidlertid observert at dette medfører uønskede linjelignende strukturer i bildet. I denne avhandlingen vises det at årsaken til dette er at de parallelle strålene deformeres når de passerer strukturene mellom hjertet og proben. Avhandlingen inneholder også et bidrag som viser hvordan bruk av parallelle mottakstråler kan medføre at vevs-Doppler kurver blir hentet fra et annet sted i hjertet enn det intensjonen var. Samlet er avhandlingen et bidrag til fortsatt utvikling og forbedring av kvantitative ultralydmetoder innen hjertediagnostikk.</p>

Ultralyd er i dag et standardverktøy innen hjertediagnostikk. Analyse av ventrikkelfunksjon gjøres subjektivt ved å se på gråskala ultralydopptak av hjertet. Dette gjør dokumentasjon og opplæring vanskelig. Det er derfor ønskelig med kvantitative mål på ventrikkelfunksjon. Kvantitative metoder kan benytte høy tidsoppløsning og dermed også registrere flere detaljer. To teknikker for kvantitativ ventrikkelanalyse er vevs-Doppler og speckle tracking (mønsterfølging). Ved bruk av disse teknikkene er det viktig å sikre kvaliteten av resultatene samtidig som analyseprosessen kan gjennomføres så effektivt som mulig. I denne avhandlingen presenteres flere metoder for å sikre kvalitet og øke effektivitet ved kvantitativ ventrikkelanalyse som alle er relatert til bruk av høy tidsoppløsning. For å kunne gjøre effektiv ventrikkelanalyse behøves analyseverktøy. Et slikt verktøy som fungerer med både vevs-Doppler og speckle tracking presenteres i denne avhandlingen. Dette verktøyet har gjort flere store kliniske studier mulig. Ved analyse av hendelsene som skjer innen en hjertesykel er det viktig å dele opp hjertesykelen i ulike faser. Venstre ventrikkels tømmingsfase avsluttes ved at aortaklaffen lukkes, og denne hendelsen må derfor bestemmes før kvantitative parametre i slutten av tømmingsfasen og i overgangen til fyllingsfasen kan bestemmes. I denne avhandlingen vurderes og testes ulike kandidater for bestemmelse av lukkingen. To gunstige kandidater i hastighetskurver fra basale deler av ventrikkelen ble funnet: andre nullkryssing etter tømming og høyeste positive akselerasjon etter tømming før fylling. Et eget bidrag viser at den siste av disse kan bestemmes automatisk. For å oppnå høy tidsoppløsning kreves høy datarate. En populær metode for å øke datarate er flere parallelle mottaksstråler for hver sendestråle. I vanlig gråtoneavbilding av hjertet er det imidlertid observert at dette medfører uønskede linjelignende strukturer i bildet. I denne avhandlingen vises det at årsaken til dette er at de parallelle strålene deformeres når de passerer strukturene mellom hjertet og proben. Avhandlingen inneholder også et bidrag som viser hvordan bruk av parallelle mottakstråler kan medføre at vevs-Doppler kurver blir hentet fra et annet sted i hjertet enn det intensjonen var. Samlet er avhandlingen et bidrag til fortsatt utvikling og forbedring av kvantitative ultralydmetoder innen hjertediagnostikk.

Designing ICT systems in the developing world is perceived to be a challenge. But is this really a challenge if we use state-of-the-art development methods? Will the fact that most standards and heuristics are made in the western world render them useless as tools of making modern computer software for the underdeveloped world? In this thesis I will analyze the findings of a software development project I did for my employer, National Center for Fetal Medicine (NCFM) at St. Olav’s Hospital in Trondheim, Norway. In this project we used a user centered development process as defined in ISO 9142-210 to make and test the prototype of an ultrasound help system. We tested the system using two different user groups: midwives in KwaZulu-Natal participating in an ultrasound teaching and training program held by NCFM and midwives in Trondheim, Norway, receiving a post-qualification course in ultrasound diagnostics also held by NCFM at the Norwegian University of Science and Technology (NTNU) in Trondheim. I will analyze the test results and then do an evaluation of the methods used – both the user-centered design process and the usability methods to see if the methods have worked in my project.

The propagation mechanism of signals for multiple input multiple output (MIMO) channels can be explained via a random matrix. Random matrix theory is a very powerful tool to understand behaviour of such channels and analyse their performance measure of MIMO systems. In this work we study: The asymptotic eigenvalue distribution and the mutual information of a multiuser (MU) multiple-input multiple output (MIMO) channel with a certain fraction of users experiencing line-of-sight. It shows that the AED of the channel matrix decomposes into two separate bulks for practically relevant parameter choices and differs very much from the common assumption of independent identically distributed (iid) entries which induces the quarter circle law. This happens even without antenna correlation at either side of the channel. In order to tackle this problem the paper makes use of recent developments in free probability theory which allow to deal with complex-valued eigenvalue distributions of non-Hermitian matrices. Moreover to understand behaviour of MIMO channels we derived asymptotic complex-valued eigenvalue distributions of practically relevant channels models by means of their respective square equivalent and singular equivalent of channel matrices. Finally we derived an explicit mutual information formula which allows us calculate the mutual information (in general) analytically in high signal-to noiseratio (SNR) regime for numerous practical important scenarios. Furthermore the numerical result shows that, high-SNR approximation draws reliable portrait even for quite moderate SNR level.

Reverberation is defined as equally-spaced, bright linear echoes resulting from reflection from specular-type interfaces. They are provoked by the acoustic Impedance change between the tissue and transducer front surface. B. Angelsen developed a mathematical approach to correct this ultrasound artifact by coupling the ultrasound transducer with an ideal electrical load in order to obtain zero reflection coefficients on the transducer from face [1]. However, when analyzing Impedance spectroscopy this approach cannot be achieved by using passive electronics in most simulated cases. Active impedance may be seen as a trend to improve image quality. This research therefore implements and simulates methods for active impedance synthesis [5] and applies Particle Swarm Optimization to the proposed electronics. Finally, the simulations are compared through a LAB experiment.

The commonly used simulation method Field II, which is based on the spatial impulse response approach, has excellent accuracy in linear domain. However the computational time can be up to many days for one simulation. One of the solutions to this problem is a convolution-based methodology called COLE. It is much faster than Field II and has very good approximation. It generates the data by reducing multi-dimensional convolution model to multiple single-dimensional convolutions. This thesis is about implementing COLE on the FieldSim 3 platform and using it for blood flow imaging. This platform is written in MATLAB with object-oriented programming and it is now under development at department of circulation and medical imaging. Both Field II and real scanner have been used to compare with COLE. The simulated phantom for both simulators was a straight tube with scatterers moving inside, whereas a string phantom was used to get the data from the scanner. The computational time of COLE with 2D Doppler mode scan in FieldSim 3 achieved 85 times faster than Field II. The plotted PW Doppler spectra and the 2D power spectra showed that the velocity resolutions of both simulators were at the same level. COLE had higher noise floor than Field II and scanner in Doppler mode scan. COLE had relatively high sampling frequency requirement compared with Field II. If the sampling frequency was not high enough, COLE would produce side lobes in the PW Doppler spectra.