The heterogeneity of breast cancer (disease stage and phenotype) makes it challenging to differentiate between each subtype; luminal A, luminal B, HER2, basal-like and claudin-low, on the basis of a single gene or protein. Therefore, a collection of markers is required that can serve as a signature for diagnosing different types of breast cancer. New developments in proteomics have provided the opportunity to look at phenotype-specific breast cancer cell lines and stage-specific liquid biopsies (nipple aspirate fluid [NAF], plasma samples) to identify disease and phenotype specific signature. An 8-plex iTRAQ quantification strategy was employed to compare proteomic profiles of a range of breast cancer and ‘normal-like’ cell lines with primary breast epithelial cells. From this, 2467 proteins were identified on Orbitrap Fusion and Ultraflex II, of which 1430 were common. Matched pairs of NAF samples from four patients with different stages of breast cancer, were analysed by SCX-LC-MS and a total of 1990 unique gene products were identified. More than double the number of proteins previously published data, were detected in NAF, including 300 not detected in plasma. The NAF from the diseased patients have 138 potential phenotype biomarkers that were significantly changed compared to the healthy volunteer (7 for luminal A, 9 for luminal B, 11 for HER2, 14 for basal-like and 52 for claudin-low type). The average coefficient of variation for triplicate analyses by multiple reaction monitoring mass spectrometry (MRM-MS), was 9% in cell lines, 17 % in tissue biopsies, 22% in serum samples and 24% in NAF samples. Overall, the results provide a strong paradigm to develop a clinical assay based on proteomic changes in NAF samples for the early detection of breast cancer supplementary to established mammography programmes.

Breast cancer is the most crucial cancer type among all other cancer types. There are many imaging techniques used to screen breast carcinoma. These are mammography, ultrasound, computed tomography, magnetic resonance imaging, infrared imaging, positron emission tomography and electrical impedance tomography. However, there is no gold standard in breast carcinoma diagnosis. The object of this study is to create a hybrid system that uses thermal and electrical imaging methods together for breast cancer diagnosis. Body tissues have different electrical conductivity values depending on their state of health and types. Consequently, one can get information about the anatomy of the human body and tissue&rsquo / s health by imaging tissue conductivity distribution. Due to metabolic heat generation values and thermal characteristics that differ from tissue to tissue, thermal imaging has started to play an important role in medical diagnosis. To increase the temperature contrast in thermal images, the characteristics of the two imaging modalities can be combined. This is achieved by implementing thermal imaging applying electrical currents from the body surface within safety limits (i.e., thermal imaging in active mode). Electrical conductivity of tissues changes with frequency, so it is possible to obtain more than one thermal image for the same body. Combining these images, more detailed information about the tumor tissue can be acquired. This may increase the accuracy in diagnosis while tumor can be detected at deeper locations. Feasibility of the proposed technique is investigated with analytical and numerical simulations and experimental studies. 2-D and 3-D numerical models of the female breast are developed and feasibility work is implemented in the frequency range of 10 kHz and 800 MHz. Temporal and spatial temperature distributions are obtained at desired depths. Thermal body-phantoms are developed to simulate the healthy breast and tumor tissues in experimental studies. Thermograms of these phantoms are obtained using two different infrared cameras (microbolometer uncooled and cooled Quantum Well Infrared Photodetectors). Single and dual tumor tissues are determined using the ratio of uniform (healthy) and inhomogeneous (tumor) images. Single tumor (1 cm away from boundary) causes 55 &deg / mC temperature increase and dual tumor (2 cm away from boundary) leads to 50 &deg / mC temperature contrast. With multi-frequency current application (in the range of 10 kHz-800 MHz), the temperature contrast generated by 3.4 mm3 tumor at 9 mm depth can be detected with the state-of-the-art thermal imagers.

Submitted by Fabio Sobreira Campos da Costa (fabio.sobreira@ufpe.br) on 2016-12-01T14:33:08Z No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação_Kamila_Queiroz.pdf: 2031256 bytes, checksum: 11934292ef99e7d010a7bdd8c3309096 (MD5) / Made available in DSpace on 2016-12-01T14:33:08Z (GMT). No. of bitstreams: 2 license_rdf: 1232 bytes, checksum: 66e71c371cc565284e70f40736c94386 (MD5) Dissertação_Kamila_Queiroz.pdf: 2031256 bytes, checksum: 11934292ef99e7d010a7bdd8c3309096 (MD5) Previous issue date: 2016-02-19 / CNPQ / A inspeção termográfica tem emergido como um método potencial para melhorar a eficiência da detecção precoce do câncer de mama. A técnica não utiliza radiação ionizante e possui a vantagem de facilitar a realização de exames de mama em homens e detectar alterações nas mamas de mulheres mais jovens. Sistemas de diagnóstico auxiliado por computador (CAD) são métodos importantes no subsídio à decisão médica e são usados para melhorar a consistência da interpretação das imagens. Normalmente, estes sistemas são associados a interfaces gráficas para facilitar o trabalho dos usuários e tornar o programa desenvolvido acessível a pesquisadores e/ou médicos ligados à área. O objetivo desta dissertação é desenvolver uma interface gráfica de usuário (GUI – Graphical User Interface) prática e que possibilite a detecção de anormalidades a partir de termogramas de mamas. Para isto foram implementados sistemas de CAD baseados em classificadores estatísticos, além de análises relacionadas ao quantitativo de casos clínicos e sua relação com a idade das pacientes. As regiões de interesse foram segmentadas tanto de forma semiautomática quanto de forma automática, as quais estão associadas, respectivamente, ao classificador SVM (Support Vector Machine) e ao classificador baseado na distância de Mahalanobis. Com o intuito de identificaras anormalidades das mamas, participaram noventa e oito pacientes do Hospital das Clínicas da Universidade Federal de Pernambuco, as quais serviram para construir a base de dados para a classificação individual de determinada paciente. A eficácia da classificação para esta amostra foi medida através da sensibilidade e da especificidade ao grupo Maligno, e das taxas de acerto das classes Benigno, Cisto e Normal. A GUI desenvolvida foi avaliada através do estudo das imagens termográficas de cinco pacientes pertencentes às diferentes classes. No presente trabalho, apresentam-se resultados para o classificador Mahalanobis e para o classificador SVM, além de suas variações. / The infrared thermography has emerged as a potential method to improve the efficiency of the early detection of the breast cancer. This technique does not use ionizing radiation and is also suitable to breast screening in men, as well as is able to detect changes in the younger women’s breasts. Computer-aided diagnosis (CAD) systems are important to medical decision and are used to improve image interpretation. Typically, these systems are associated with graphical interfaces to facilitate users’ work. Furthermore the developed framework can be an important tool (GUI - Graphical User Interface) for the people interested in breast abnormalities detection. In the sense, the CAD systems were implemented based on statistical classifiers. Some statistical analyses associated to quantitative clinical cases were performed. The relation to patients’ age was also analyzed. The regions of interest were segmented in automatic and semiautomatic manners, which are respectively associated with the SVM (Support Vector Machine) classifier and the Mahalanobis classifier. Ninety eight patients images from the Hospital das Clínicas (HC) of Federal University of Pernambuco (UFPE) participated in the tests. The classification efficiency for this sample was measured using the sensitivity and the specificity to the malignant group, and to the accuracy of classifying the classes: Benign, Cyst and Normal. The GUI created was evaluated through the study of thermographic images of five patients with the different referred classes. In the present work, the results for the Mahalanobis classifier and SVM classifier are presented.

In Sweden, women of age between of 40 and 74 go through regular screening of their breasts every 18-24 months. The screening mainly involves obtaining a mammogram and having radiologists analyze them to detect any sign of breast cancer. However reading a mammography image requires experienced radiologist, and the lack of radiologist reduces the hospital's operating efficiency. What's more, mammography from different facilities increases the difficulty of diagnosis. Our work proposed a deep learning segmentation system which could adapt to mammography from various facilities and locate the position of the tumor. We train and test our method on two public mammography datasets and do several experiments to find the best parameter setting for our system. The test segmentation results suggest that our system could play as an auxiliary diagnosis tool for breast cancer diagnosis and improves diagnostic accuracy and efficiency. / I Sverige går kvinnor i åldrarna mellan 40 och 74 igenom regelbunden screening av sina bröst med 18-24 månaders mellanrum. Screeningen innbär huvudsakligen att ta mammogram och att låta radiologer analysera dem för att upptäcka tecken på bröstcancer. Emellertid krävs det en erfaren radiolog för att tyda en mammografibild, och bristen på radiologer reducerar sjukhusets operativa effektivitet. Dessutom, att mammografin kommer från olika anläggningar ökar svårigheten att diagnostisera. Vårt arbete föreslår ett djuplärande segmenteringssystem som kan anpassa sig till mammografi från olika anläggningar och lokalisera tumörens position. Vi tränar och testar vår metod på två offentliga mammografidataset och gör flera experiment för att hitta den bästa parameterinställningen för vårt system. Testsegmenteringsresultaten tyder på att vårt system kan fungera som ett hjälpdiagnosverktyg vid diagnos av bröstcancer och förbättra diagnostisk noggrannhet och effektivitet.

Yes / An ultra-wideband microstrip antenna design is considered with respect to applications in breast cancer detection. The underlying design concept is based on ground penetrating radar (GPR). Simulated and measured prototype performance show excellent performance in the input impedance and radiation pattern over the target range from 4 GHz to 8 GHz. The 4 GHz to 8GHz frequency band for microwave imaging perform better in comparison with other microwave frequencies. The antenna also shows a reasonable uniform radiation performance in the broadside direction which contributes to the reduction of clutter levels, thus aiding the reconstruction quality of the final image.

The CD-ROM disc containing supplementary material is kept in the cardboard box in the Systems Office. / The heterogeneity of breast cancer (disease stage and phenotype) makes it challenging to differentiate between each subtype; luminal A, luminal B, HER2, basal-like and claudin-low, on the basis of a single gene or protein. Therefore, a collection of markers is required that can serve as a signature for diagnosing different types of breast cancer. New developments in proteomics have provided the opportunity to look at phenotype-specific breast cancer cell lines and stage-specific liquid biopsies (nipple aspirate fluid [NAF], plasma samples) to identify disease and phenotype specific signature. An 8-plex iTRAQ quantification strategy was employed to compare proteomic profiles of a range of breast cancer and ‘normal-like’ cell lines with primary breast epithelial cells. From this, 2467 proteins were identified on Orbitrap Fusion and Ultraflex II, of which 1430 were common. Matched pairs of NAF samples from four patients with different stages of breast cancer, were analysed by SCX-LC-MS and a total of 1990 unique gene products were identified. More than double the number of proteins previously published data, were detected in NAF, including 300 not detected in plasma. The NAF from the diseased patients have 138 potential phenotype biomarkers that were significantly changed compared to the healthy volunteer (7 for luminal A, 9 for luminal B, 11 for HER2, 14 for basal-like and 52 for claudin-low type). The average coefficient of variation for triplicate analyses by multiple reaction monitoring mass spectrometry (MRM-MS), was 9% in cell lines, 17 % in tissue biopsies, 22% in serum samples and 24% in NAF samples. Overall, the results provide a strong paradigm to develop a clinical assay based on proteomic changes in NAF samples for the early detection of breast cancer supplementary to established mammography programmes. / The supplementary material submitted with the thesis is not available online.

Diagnosing early formation of tumors or lumps, particularly those caused by cancer, has been a challenging problem. To help physicians detect tumors more efficiently, various imaging techniques with different imaging modalities such as computer tomography, ultrasonic imaging, nuclear magnetic resonance imaging, and mammography, have been developed. However, each of these techniques has limitations, including exposure to radiation, excessive costs, and complexity of machinery. Tissue elasticity is an important indicator of tissue health, with increased stiffness pointing to an increased risk of cancer. In addition to increased tissue elasticity, geometric parameters such as size of a tissue inclusion are also important factors in assessing the tumor. The combined knowledge of tissue elasticity and its geometry would aid in tumor identification. In this research, we present a tactile sensation imaging system (TSIS) and algorithms which can be used for practical medical diagnostic experiments for measuring stiffness and geometry of tissue inclusion. The TSIS incorporates an optical waveguide sensing probe unit, a light source unit, a camera unit, and a computer unit. The optical method of total internal reflection phenomenon in an optical waveguide is adapted for the tactile sensation imaging principle. The light sources are attached along the edges of the waveguide and illuminates at a critical angle to totally reflect the light within the waveguide. Once the waveguide is deformed due to the stiff object, it causes the trapped light to change the critical angle and diffuse outside the waveguide. The scattered light is captured by a camera. To estimate various target parameters, we develop the tactile data processing algorithm for the target elasticity measurement via direct contact. This algorithm is accomplished by adopting a new non-rigid point matching algorithm called "topology preserving relaxation labeling (TPRL)." Using this algorithm, a series of tactile data is registered and strain information is calculated. The stress information is measured through the summation of pixel values of the tactile data. The stress and strain measurements are used to estimate the elasticity of the touched object. This method is validated by commercial soft polymer samples with a known Young's modulus. The experimental results show that using the TSIS and its algorithm, the elasticity of the touched object is estimated within 5.38% relative estimation error. We also develop a tissue inclusion parameter estimation method via indirect contact for the characterization of tissue inclusion. This method includes developing a forward algorithm and an inversion algorithm. The finite element modeling (FEM) based forward algorithm is designed to comprehensively predict the tactile data based on the parameters of an inclusion in the soft tissue. This algorithm is then used to develop an artificial neural network (ANN) based inversion algorithm for extracting various characteristics of tissue inclusions, such as size, depth, and Young's modulus. The estimation method is then validated by using realistic tissue phantoms with stiff inclusions. The experimental results show that the minimum relative estimation errors for the tissue inclusion size, depth, and hardness are 0.75%, 6.25%, and 17.03%, respectively. The work presented in this dissertation is the initial step towards early detection of malignant breast tumors. / Electrical and Computer Engineering

Near field imaging using microwave in medical applications has gain much attention recently as various researches show its high ability and accuracy in illuminating object comparing to the well-known screening tools such as Magnetic Resonance Imaging (MRI), digital mammography, ultrasound etc. This has encourage and motivate scientists continue to exploit the potential of microwave imaging so that a better and more powerful sensing tools can be developed. This thesis documents the development of antenna design for microwave imaging application such as breast cancer detection. The application is similar to the concept of Ground Penetrating Radar (GPR) but operating at higher frequency band. In these systems a short pulse is transmitted from an antenna to the medium and the backscattered response is investigated for diagnose. In order to accommodate such a short pulse, a very wideband antenna with a minimal internal reflection is required. Printed monopole and planar metal plate antenna is implemented to achieve the necessary operating wide bandwidth. The development of new compact printed planar metal plate ultra wide bandwidth antenna is presented. A generalized parametric study is carried out using two well-known software packages to achieve optimum antenna performance. The Prototype antennas are tested and analysed experimentally, in which a reasonable agreement was achieved with the simulations. The antennas present an excellent relative wide bandwidth of 67% with acceptable range of power gain between 3.5 to 7 dBi. A new compact size air-dielectric microstrip patch-antenna designs proposed for breast cancer detection are presented. The antennas consist of a radiating patch mounted on two vertical plates, fed by coaxial cable. The antennas show a wide bandwidth that were verified by the simulations and also confirmed experimentally. The prototype antennas show excellent performance in terms the input impedance and radiation performance over the target range bandwidth from 4 GHz to 8 GHz. A mono-static model with a homogeneous dielectric box having similar properties to human tissue is used to study the interaction of the antenna with tissue. The numerical results in terms the matching required of new optimised antennas were promising. An experimental setup of sensor array for early-stage breast-cancer detection is developed. The arrangement of two elements separated by short distance that confined equivalent medium of breast tissues were modelled and implemented. The operation performances due to several orientations of the antennas locations were performed to determine the sensitivity limits with and without small size equivalent cancer cells model. In addition, a resistively loaded bow tie antenna, intended for applications in breast cancer detection, is adaptively modified through modelling and genetic optimisation is presented. The required wideband operating characteristic is achieved through manipulating the resistive loading of the antenna structure, the number of wires, and their angular separation within the equivalent wire assembly. The results show an acceptable impedance bandwidth of 100.75 %, with a VSWR < 2, over the interval from 3.3 GHz to 10.0 GHz. Feasibility studies were made on the antenna sensitivity for operation in a tissue equivalent dielectric medium. The simulated and measured results are all in close agreement.

Near field imaging using microwave in medical applications has gain much attention recently as various researches show its high ability and accuracy in illuminating object comparing to the well-known screening tools such as Magnetic Resonance Imaging (MRI), digital mammography, ultrasound etc. This has encourage and motivate scientists continue to exploit the potential of microwave imaging so that a better and more powerful sensing tools can be developed. This thesis documents the development of antenna design for microwave imaging application such as breast cancer detection. The application is similar to the concept of Ground Penetrating Radar (GPR) but operating at higher frequency band. In these systems a short pulse is transmitted from an antenna to the medium and the backscattered response is investigated for diagnose. In order to accommodate such a short pulse, a very wideband antenna with a minimal internal reflection is required. Printed monopole and planar metal plate antenna is implemented to achieve the necessary operating wide bandwidth. The development of new compact printed planar metal plate ultra wide bandwidth antenna is presented. A generalized parametric study is carried out using two well-known software packages to achieve optimum antenna performance. The Prototype antennas are tested and analysed experimentally, in which a reasonable agreement was achieved with the simulations. The antennas present an excellent relative wide bandwidth of 67% with acceptable range of power gain between 3.5 to 7 dBi. A new compact size air-dielectric microstrip patch-antenna designs proposed for breast cancer detection are presented. The antennas consist of a radiating patch mounted on two vertical plates, fed by coaxial cable. The antennas show a wide bandwidth that were verified by the simulations and also confirmed experimentally. The prototype antennas show excellent performance in terms the input impedance and radiation performance over the target range bandwidth from 4 GHz to 8 GHz. A mono-static model with a homogeneous dielectric box having similar properties to human tissue is used to study the interaction of the antenna with tissue. The numerical results in terms the matching required of new optimised antennas were promising. An experimental setup of sensor array for early-stage breast-cancer detection is developed. The arrangement of two elements separated by short distance that confined equivalent medium of breast tissues were modelled and implemented. The operation performances due to several orientations of the antennas locations were performed to determine the sensitivity limits with and without small size equivalent cancer cells model. In addition, a resistively loaded bow tie antenna, intended for applications in breast cancer detection, is adaptively modified through modelling and genetic optimisation is presented. The required wideband operating characteristic is achieved through manipulating the resistive loading of the antenna structure, the number of wires, and their angular separation within the equivalent wire assembly. The results show an acceptable impedance bandwidth of 100.75 %, with a VSWR < 2, over the interval from 3.3 GHz to 10.0 GHz. Feasibility studies were made on the antenna sensitivity for operation in a tissue equivalent dielectric medium. The simulated and measured results are all in close agreement.

Breast cancer continues to be a major cause of mortality among women. In recent years, machine learning has emerged as a potential tool in detecting and grading cancer. Using machine learning techniques in computational pathology has the potential to improve precision medicine, enabling more personalized and more accurate treatment plans. The machine learning models can even detect structures that cannot be seen with human eyes. The first step is often to identify tissue areas with cancerous cells using machine learning models. Those models often rely solely on Haematoxylin and Eosin slides for training due to the time-consuming and costly nature of annotations by pathologists. Because of that, valuable information for training might be lost since some cancerous cells are more visible in the immunohistochemistry slides. In this thesis, Haematoxylin and Eosin slide annotations are registered to immunohistochemistry slides for training singlestain and multi-stain models. The registration of the annotations is not straightforward since the tissue of the slides is not necessarily from consecutive cuts, and they are sometimes applied to the slide at different angles. An algorithm evaluated during the ACROBAT challenge was used to register the slides. Using the transferred annotations, individual models are trained for each stain (K167, HER2, PGR, ER). Of the single-stain model, the HER2 stain model is showing the most promising results. As a second step, a multistain model is trained using all stains. The multi-stain model performs equally well as the single-stain models specializing in individual stains. This shows that there is no need to train specialized single-stain models. Thus being able to train one model for four different stains makes it possible to detect cancer in whole slide images stained with one of those four stains without the need to train a specialized model and only needing annotations in one stain. While the multi-stain model is a nice addition this thesis shows that it is possible to reuse annotations, which reduces the amount of manual labour from pathologists and allows for training models on immunohistochemistry slides with only having annotations from one stain. / Bröstcancer fortsätter att vara en vanlig orsak till dödlighet bland kvinnor. På senare år har maskininlärning visat sig vara ett värdefullt verktyg för att upptäcka och gradera cancer. Att använda maskininlärningstekniker inom beräkningspatologi har potential att förbättra precisionsmedicinen och möjliggöra mer individanpassade och exakta behandlingsplaner. Maskininlärningsmodellerna kan till och med upptäcka strukturer som inte kan ses med mänskliga ögon. Det första steget är ofta att identifiera vävnadsområden med cancerceller med hjälp av maskininlärningsmodeller. Dessa modeller är ofta helt beroende av hematoxylin- och eosin-slidebilder för träning eftersom det är tidsödande och kostsamt för patologer att göra annoteringar. På grund av detta kan värdefull information för träning gå förlorad eftersom vissa cancerceller är mer synliga på immunohistokemiska objektglas. I den här avhandlingen registreras annoteringar från objektglas med hematoxylin och eosin på immunohistokemiska objektglas för träning av modeller med en och flera infärgningar. Registreringen av annoteringarna är inte okomplicerad eftersom vävnaden på objektglasen inte nödvändigtvis kommer från på varandra följande snitt, och de appliceras ibland på objektglaset i olika vinklar. En algoritm som utvecklades under ACROBAT-utmaningen användes för att registrera bilderna. Med hjälp av de registrerade objektglasen tränas individuella modeller för varje infärgning (K167, HER2, PGR, ER). Av modellerna för enstaka infärgningar visar modellen för HER2-infärgning de mest lovande resultaten. I ett andra steg tränas en modell med flera infärgningar med hjälp av alla infärgningar. Multi-stain-modellen presterar lika bra som single-stain-modellerna som är specialiserade på enskilda infärgningar. Detta visar att det inte finns något behov av att träna specialiserade modeller för enstaka infärgningar. Att kunna träna en modell för fyra olika färgämnen gör det alltså möjligt att upptäcka cancer i hela objektglasbilder som färgats med ett av dessa fyra färgämnen utan att behöva träna en specialiserad modell och utan att behöva göra annoteringar. Möjligheten att endast använda en modell för att förutsäga fyra olika immunohistokemiska helbilder minskade datorkostnaderna för träning och underhåll av modellen.