Examining surface shape appearance by touching and observing a lesion from different points of view is a part of the clinical process for skin lesion diagnosis. Motivated by this, we hypothesise that surface shape embodies important information that serves to represent lesion identity and status. A new sensor, Dense Stereo Imaging System (DSIS) allows us to capture 1:1 aligned 3D surface data and 2D colour images simultaneously. This thesis investigates whether the extra surface shape appearance information, represented by features derived from the captured 3D data benefits skin lesion analysis, particularly on the tasks of segmentation and classification. In order to validate the contribution of 3D data to lesion identification, we compare the segmentations resulting from various combinations of images cues (e.g., colour, depth and texture) embedded in a region-based level set segmentation method. The experiments indicate that depth is complementary to colour. Adding the 3D information reduces the error rate from 7:8% to 6:6%. For the purpose of evaluating the segmentation results, we propose a novel ground truth estimation approach that incorporates a prior pattern analysis of a set of manual segmentations. The experiments on both synthetic and real data show that this method performs favourably compared to the state of the art approach STAPLE [1] on ground truth estimation. Finally, we explore the usefulness of 3D information to non-melanoma lesion diagnosis by tests on both human and computer based classifications of five lesion types. The results provide evidence for the benefit of the additional 3D information, i.e., adding the 3D-based features gives a significantly improved classification rate of 80:7% compared to only using colour features (75:3%). The three main contributions of the thesis are improved methods for lesion segmentation, non-melanoma lesion classification and lesion boundary ground-truth estimation.

No description available.

No description available.

Patients with skin lesions of the temple region that have spread to the parotid gland are commonly treated with three-dimensional conformal radiation therapy (3DCRT). 3DCRT has associated limitations when treating this disease. 3DCRT requires this disease site to be treated with two junction regions, resulting in poor dose conformity to the tumour target. Proximity of critical structures to the target volume can make dosimetry difficult, "especially for concave-shaped targets in close proximity to sensitive normal structures" (Saw.C et al., 2002, p76). Intensity modulated radiation therapy (IMRT) is a relatively new treatment technology that has potential to overcome limitations associated with 3DCRT (Garden.A et al., 2004). IMRT has been reported to have significant advantages over conventional 3DCRT treatment, by improving dose to the tumour and lowering doses to critical structures (Adams.E et al., 2001). Research has been conducted into the optimal IMRT treatment for specific head and neck carcinomas. They are identified as class solutions. "A class solution can be defined as the historical experience in designing RT plans for a particular site" (Intensity Modulated Radiation Therapy Collaborative Working, 2001, p913). This study was performed to establish an optimal IMRT class solution for patients with skin lesions of the temple region that have spread to the parotid gland, and to determine if it is the superior treatment option over 3DCRT treatment. Dosimetry planning was performed on computerised tomography data sets of nine patients with this disease site. One optimised 3DCRT dosimetry plan and eight optimised IMRT plans with specific beam arrangements were calculated. Clinical and statistical analysis was performed on; critical structures, conformity indices (CI) and dose volume histogram (DVH) range analysis of the planning target volume (PTV). Analysis of IMRT plans revealed that the 7-beam arrangement and 4-beam ipsilateral arrangement produced significantly lower doses to the majority of critical structures (P < 0.05). The 7-beam IMRT arrangement produced the best and second best CI and DVH PTV results, but these were not significantly different to the majority of other beam arrangements. This indicates that the 7-beam arrangement with defined beam angles of; 40°,120°,160°,200°,240°,300°,0°, is the superior IMRT treatment plan, and thus class solution for this disease site. Clinical analysis confirmed results. Analysis was performed on IMRT class solution results compared with 3DCRT results. CI was significance higher and DVH PTV range was significantly lower for the IMRT class solution (P < 0.05). The class solution delivered significantly higher doses to the majority of critical structures in comparison to the 3DCRT plan (P < 0.05). This indicates that the IMRT class solution is superior to 3DCRT in terms of PTV conformity and homogeneity, but not in terms of doses to critical structures. Skin lesions of the temple region with tumour extension to the parotid gland, is a complicated disease site. Investigations into current and potential radiation therapy treatments will guide treatment options and facilitate outcomes for patients with this disease.

Melanoma is the most deadly form of skin cancer worldwide, which causes the 75% of deaths related to skin cancer. National Cancer Institute estimated that 91,270 new case and 9,320 deaths are expected in 2018 caused by melanoma. Early detection of melanoma is the key for the treatment. The image technique to diagnose skin cancer is dermoscopy, which leads to improved diagnose accuracy compared to traditional ABCD criteria. But reading and examining dermoscopic images is a time-consuming and complex process. Therefore, computerized analysis methods of dermoscopic images have been developed to assist the visual interpretation of dermoscopic images. The automatic segmentation of skin lesion attributes is a key step in computerized analysis of dermoscopic images. The International Skin Imaging Collaboration (ISIC) hosted the 2018 Challenges to help the diagnosis of melanoma based on dermoscopic images. In this thesis, I develop a deep learning based approach to automatically segment the attributes from dermoscopic skin lesion images. The approach described in the thesis achieved the Jaccard index of 0.477 on the official test dataset, which ranked 5th place in the challenge. / Master of Science / Melanoma is the most deadly form of skin cancer worldwide, which causes the 75% of deaths related to skin cancer. Early detection of melanoma is the key for the treatment. The image technique to diagnose skin cancer is called dermoscopy. It has become increasingly conveniently to use dermoscopic device to image the skin in recent years. Dermoscopic lens are available in the market for individual customer. When coupling the dermoscopic lens with smartphones, people are be able to take dermoscopic images of their skin even at home. However, reading and examining dermoscopic images is a time-consuming and complex process. It requires specialists to examine the image, extract the features, and compare with criteria to make clinical diagnosis. The time-consuming image examination process becomes the bottleneck of fast diagnosis of melanoma. Therefore, computerized analysis methods of dermoscopic images have been developed to promote the melanoma diagnosis and to increase the survival rate and save lives eventually. The automatic segmentation of skin lesion attributes is a key step in computerized analysis of dermoscopic images. In this thesis, I developed a deep learning based approach to automatically segment the attributes from dermoscopic skin lesion images. The segmentation result from this approach won 5th place in a public competition. It has the potential to be utilized in clinic application in the future.

No description available.

Malignant melanoma (MM) is a type of skin cancer that is associated with a very poor prognosis and can often lead to death. Early detection is crucial in order to administer the right treatment successfully but currently requires the expertise of a dermatologist. In the past years, studies have shown that automatic detection of MM is possible through computer vision and machine learning methods. Skin lesion segmentation and classification are the key methods in supporting automatic detection of different skin lesions. Compared with traditional computer vision as well as other machine learning methods, deep neural networks currently show the greatest promise both in segmentation and classification. In our work, we have implemented several deep neural networks to achieve the goals of skin lesion segmentation and classification. We have also applied different training schemes. Our best segmentation model achieves pixel-wise accuracy of \textbf{0.940}, Dice index of \textbf{0.867} and Jaccard index of \textbf{0.765} on the ISIC 2017 challenge dataset. This surpassed the official state of the art model whose pixel-wise accuracy was 0.934, Dice index 0.849 and Jaccard Index 0.765. We have also trained a segmentation model with the help of adversarial loss which improved the baseline model slightly. Our experiments with several neural network models for skin lesion classification achieved varying results. We also combined both segmentation and classification in one pipeline meaning that we were able to train the most promising classification model on pre-segmented images. This resulted in improved classification performance. The binary (melanoma or not) classification from this single model trained without extra data and clinical information reaches an area under the curve (AUC) of 0.684 on the official ISIC test dataset. Our results suggest that automatic detection of skin cancers through image analysis shows significant promise in early detection of malignant melanoma.

Fêmeas suínas gestantes são frequentemente submetidas a uma restrição alimentar, que pode comprometer o bem-estar e produtividade. Existe pouca informação em relação as consequências da fome nas fêmeas gestantes nas medidas de bem-estar dos leitões. Dietas ricas em fibra podem minimizar a sensação de fome e, consequentemente, melhorar o bem-estar e a produtividade. Os objetivos deste estudo foram: 1) avaliar o impacto de uma dieta alta fibra (AF) para marrãs gestantes em medidas de bem-estar e 2) avaliar as consequências da dieta contendo AF durante a gestação no comportamento agonístico e indicadores de medo nos leitões ao desmame. Vinte e oito marrãs gestantes foram alimentadas com diferentes dietas: dieta AF, contendo 12,86% de fibra bruta (n=16), ou dieta baixa fibra (BF), contendo 2,53% de fibra bruta (n=12). Investigamos o impacto da AF ou BF nos seguintes parâmetros nas marrãs gestantes: comportamento; concentração do cortisol salivar; desempenho; motivação alimentar durante um teste de consumo ad libitum. Também avaliamos medidas de comportamento e desempenho da prole de 22 fêmeas (AF=14, BF=8). Lesões de pele foram avaliadas antes e após o desmame em 156 (100 AF e 56 BF), e 142 leitões foram submetidos ao teste de campo aberto e objeto novo (87 AF e 55 BF). Houve uma interação entre tratamento e tempo de alimentação para duração e frequência do comportamento mastigação em falso, indicando que as fêmeas que receberam a dieta BF apresentaram o comportamento mastigação em falso por mais tempo e com maior frequência antes da alimentação comparada com após a alimentação. Isso não foi observado nas fêmeas alimentadas com dieta AF. Para a maioria dos comportamentos avaliados, houve um efeito do momento de alimentação na duração e frequência. Não houve diferença na concentração de cortisol salivar entre os tratamentos. Os dados de desempenho indicam que as fêmeas AF foram mais pesadas no terço final de gestação e aos 107 dias de gestação, quando elas foram transferidas para a maternidade, comparadas com fêmeas BF. Não houve diferença nos outros parâmetros de desempenho. Não houve efeito do tratamento no total de alimento consumido durante o teste ad libitum. Não houve efeito do tratamento no desempenho dos leitões. Leitões nascidos de marrãs que receberam a dieta AF apresentaram menor número de lesões de pele antes do desmame comparados à prole de fêmeas BF. No teste de campo aberto e objeto novo, não houve efeito do tratamento no comportamento dos leitões. Esses resultados indicam que uma dieta AF foi eficaz em reduzir o comportamento anormal em marrãs e os leitões nascidos de fêmeas alimentadas com dieta AF demonstraram menor comportamento agressivo antes do desmame. / Pregnant sows are often subjected to food restriction, which can compromise their welfare and performance. Limited information is available on the consequences of sow hunger during pregnancy on welfare outcomes for their piglets. High fibre diets can mitigate the feeling of hunger and, consequently, improving welfare and productivity measures. The aims of this study were: 1) to measure the impact of feeding pregnant gilts with high fibre diet (HFD) on welfare measures, and 2) to assess the consequences of feeding gilts with HFD during pregnancy on agonistic behaviour and indicators of fear in their piglets at weaning. Twenty-eight pregnant gilts were fed either HFD, 12,86% of crude fibre (n=16) or low fibre diet (LFD), 2,53% of crude fibre (n=12). We investigated the impact of HFD and LFD on the following parameters in pregnant gilts: behaviour; salivary cortisol concentration; performance; and feeding motivation during an ad libitum test. We also assessed some behavior and performance measures in the offspring of 22 sows (HFD=14 LFD=8). Skin lesions were evaluated before and after weaning in 156 piglets (100 HFD and 56 LFD), and 142 piglets were subjected to an open field and novel object test (87HFD and 55 LFD). There was an interaction between treatment and feeding time, for duration and frequency of sham chewing abnormal behaviour, indicating that sows that received LFD performed the behaviour for longer and more often before feeding than after feeding. This was not observed in HFD fed sows. For most behaviours evaluated, there was an effect of feeding time in their duration and frequency. There was no difference in salivary cortisol concentration among treatments. The performance data indicated that HFD fed pregnant gilts were heavier in the last third of gestation and at 107th days of gestation, when they were moved to farrowing pens, than LFD fed pregnant gilts. There was no difference in other performance parameters. There was no effect of treatment on total food consumption during the ad libitum test. There was no treatment effect on the piglet\'s performance. Piglets born from gilts that received HFD had less skin lesions before weaning that the offspring of LFD pregnant gilts. In the open field and novel object tests, there was no treatment effect on the behaviour of piglets. These results indicate that HFD was beneficial in reducing abnormal behaviour is sows and that piglets born from gilts fed with HFD showed less aggressive behaviour prior to weaning.

The overall aim of this thesis is to evaluate the risk for increased morbidity and mortality due to long-term arsenic exposure via drinking water by use of epidemiological and spatial approaches in studies performed in Matlab, Bangladesh. A total of 166,934 individuals aged 4 years and above were screened for skin lesions in 2002-2003. Another sample of 115,903 adults aged 15 years or more and a third sample of 26,972 pregnancies in 1991-2000 were considered in a historical cohort and an ecological study, respectively, where risk of adult mortality and spatial clusters of foetal loss and infant death were analysed in relation to arsenic exposure. More than 70% of the tube-wells in the study area exceeded the threshold for arsenic contamination according to the WHO guideline (10 µg/L). The prevalence of arsenic-induced skin lesions was 3/1000 and men had significantly higher prevalence of skin lesions (SMR 158, 95% CI: 133-188) compared to women. There was an increased risk for death in adulthood due to all non-accidental causes (hazards ratio = 1.16, [95% CI 1.06-1.26]) even at a low level of arsenic contamination (10-49 µg/L). Slightly lower risks were observed for death in cancers (1.44 [1.06-1.95]) and infectious diseases (1.30 [1.13-1.49]) at 50-149 µg/L, but for cardiovascular diseases, it was evident (1.23 [1.01-1.51]) from the level 150-299 µg/L. A dose-response relationship was observed for each of these causes. We identified high and low risk clusters of foetal loss and infant death that coincided with identified high and low clusters of arsenic exposure. Water arsenic concentration of the reported main water source was significantly correlated with arsenic concentration in urine, which reflects current arsenic intake from all sources (R2=0.41, ρ&lt; 0.0001), and the influence of neighbouring water sources was minimal. The study findings underlines that the ongoing arsenic exposure has resulted in a series of severe public health consequences in Bangladesh that call for reinforcement in the mitigation efforts. Knowledge about the spatial distribution of exposure and health effects may be of value in that process.

Fêmeas suínas gestantes são frequentemente submetidas a uma restrição alimentar, que pode comprometer o bem-estar e produtividade. Existe pouca informação em relação as consequências da fome nas fêmeas gestantes nas medidas de bem-estar dos leitões. Dietas ricas em fibra podem minimizar a sensação de fome e, consequentemente, melhorar o bem-estar e a produtividade. Os objetivos deste estudo foram: 1) avaliar o impacto de uma dieta alta fibra (AF) para marrãs gestantes em medidas de bem-estar e 2) avaliar as consequências da dieta contendo AF durante a gestação no comportamento agonístico e indicadores de medo nos leitões ao desmame. Vinte e oito marrãs gestantes foram alimentadas com diferentes dietas: dieta AF, contendo 12,86% de fibra bruta (n=16), ou dieta baixa fibra (BF), contendo 2,53% de fibra bruta (n=12). Investigamos o impacto da AF ou BF nos seguintes parâmetros nas marrãs gestantes: comportamento; concentração do cortisol salivar; desempenho; motivação alimentar durante um teste de consumo ad libitum. Também avaliamos medidas de comportamento e desempenho da prole de 22 fêmeas (AF=14, BF=8). Lesões de pele foram avaliadas antes e após o desmame em 156 (100 AF e 56 BF), e 142 leitões foram submetidos ao teste de campo aberto e objeto novo (87 AF e 55 BF). Houve uma interação entre tratamento e tempo de alimentação para duração e frequência do comportamento mastigação em falso, indicando que as fêmeas que receberam a dieta BF apresentaram o comportamento mastigação em falso por mais tempo e com maior frequência antes da alimentação comparada com após a alimentação. Isso não foi observado nas fêmeas alimentadas com dieta AF. Para a maioria dos comportamentos avaliados, houve um efeito do momento de alimentação na duração e frequência. Não houve diferença na concentração de cortisol salivar entre os tratamentos. Os dados de desempenho indicam que as fêmeas AF foram mais pesadas no terço final de gestação e aos 107 dias de gestação, quando elas foram transferidas para a maternidade, comparadas com fêmeas BF. Não houve diferença nos outros parâmetros de desempenho. Não houve efeito do tratamento no total de alimento consumido durante o teste ad libitum. Não houve efeito do tratamento no desempenho dos leitões. Leitões nascidos de marrãs que receberam a dieta AF apresentaram menor número de lesões de pele antes do desmame comparados à prole de fêmeas BF. No teste de campo aberto e objeto novo, não houve efeito do tratamento no comportamento dos leitões. Esses resultados indicam que uma dieta AF foi eficaz em reduzir o comportamento anormal em marrãs e os leitões nascidos de fêmeas alimentadas com dieta AF demonstraram menor comportamento agressivo antes do desmame. / Pregnant sows are often subjected to food restriction, which can compromise their welfare and performance. Limited information is available on the consequences of sow hunger during pregnancy on welfare outcomes for their piglets. High fibre diets can mitigate the feeling of hunger and, consequently, improving welfare and productivity measures. The aims of this study were: 1) to measure the impact of feeding pregnant gilts with high fibre diet (HFD) on welfare measures, and 2) to assess the consequences of feeding gilts with HFD during pregnancy on agonistic behaviour and indicators of fear in their piglets at weaning. Twenty-eight pregnant gilts were fed either HFD, 12,86% of crude fibre (n=16) or low fibre diet (LFD), 2,53% of crude fibre (n=12). We investigated the impact of HFD and LFD on the following parameters in pregnant gilts: behaviour; salivary cortisol concentration; performance; and feeding motivation during an ad libitum test. We also assessed some behavior and performance measures in the offspring of 22 sows (HFD=14 LFD=8). Skin lesions were evaluated before and after weaning in 156 piglets (100 HFD and 56 LFD), and 142 piglets were subjected to an open field and novel object test (87HFD and 55 LFD). There was an interaction between treatment and feeding time, for duration and frequency of sham chewing abnormal behaviour, indicating that sows that received LFD performed the behaviour for longer and more often before feeding than after feeding. This was not observed in HFD fed sows. For most behaviours evaluated, there was an effect of feeding time in their duration and frequency. There was no difference in salivary cortisol concentration among treatments. The performance data indicated that HFD fed pregnant gilts were heavier in the last third of gestation and at 107th days of gestation, when they were moved to farrowing pens, than LFD fed pregnant gilts. There was no difference in other performance parameters. There was no effect of treatment on total food consumption during the ad libitum test. There was no treatment effect on the piglet\'s performance. Piglets born from gilts that received HFD had less skin lesions before weaning that the offspring of LFD pregnant gilts. In the open field and novel object tests, there was no treatment effect on the behaviour of piglets. These results indicate that HFD was beneficial in reducing abnormal behaviour is sows and that piglets born from gilts fed with HFD showed less aggressive behaviour prior to weaning.