Automated image tagging through tag propagation

Trabalho apresentado no âmbito do Mestrado em Engenharia Informática, como requisito parcial Para obtenção do grau de Mestre em Engenharia Informática

Towards a classification scheme for musical sounds

Signal Processing: Algorithms, Architectures, Arrangements, and Applications (SPA), 2013

Detecting and evaluating land cover change in the eastern half of East Timor (1972-2011)

Dissertation submitted in partial fulfillment of the requirements for the Degree of Master of Science in Geospatial Technologies.

Optimization of breast tomosynthesis image reconstruction using parallel computing

Breast cancer is the most common cancer among women, being a major public health problem. Worldwide, X-ray mammography is the current gold-standard for medical imaging of breast cancer. However, it has associated some well-known limitations. The false-negative rates, up to 66% in symptomatic women, and the false-positive rates, up to 60%, are a continued source of concern and debate. These drawbacks prompt the development of other imaging techniques for breast cancer detection, in which Digital Breast Tomosynthesis (DBT) is included. DBT is a 3D radiographic technique that reduces the obscuring effect of tissue overlap and appears to address both issues of false-negative and false-positive rates. The 3D images in DBT are only achieved through image reconstruction methods. These methods play an important role in a clinical setting since there is a need to implement a reconstruction process that is both accurate and fast. This dissertation deals with the optimization of iterative algorithms, with parallel computing through an implementation on Graphics Processing Units (GPUs) to make the 3D reconstruction faster using Compute Unified Device Architecture (CUDA). Iterative algorithms have shown to produce the highest quality DBT images, but since they are computationally intensive, their clinical use is currently rejected. These algorithms have the potential to reduce patient dose in DBT scans. A method of integrating CUDA in Interactive Data Language (IDL) is proposed in order to accelerate the DBT image reconstructions. This method has never been attempted before for DBT. In this work the system matrix calculation, the most computationally expensive part of iterative algorithms, is accelerated. A speedup of 1.6 is achieved proving the fact that GPUs can accelerate the IDL implementation.

Otimização da anisotropia fronto-calosa e ressonância magnética [1,5T] no idoso normal e com risco cérebro-vascular

RESUMO: Os mais recentes métodos de neuro imagem tal como a Ressonância Magnética (RM) permitiram obter imagens detalhadas do cérebro humano in vivo. Essas imagens revelam, muitas vezes, achados imprevistos face ao padrão normal, com elevada propensão para os indivíduos idosos e franca coexistência com fatores de risco vascular, como característica dum processo de envelhecimento normal. Embora na última década tenham surgido várias publicações sobre este assunto, ele continua ainda pouco explicado. Um pouco por todo o mundo têm emergido os programas de prevenção da doença e promoção da saúde desenvolvidos pela Saúde Pública suportadas sobretudo pelo avanço das tecnologias médicas que resultaram, entre outros impactos, num crescimento da população idosa. Estima-se, em 2030, uma composição demográfica com 20% de indivíduos acima dos 65 anos. Neste contexto, a doença microvascular cerebral é a causa mais frequente de comprometimento cognitivo vascular no idoso sendo as características senescentes na imagem por RM do tipo lesões isquémicas da Substância Branca (Leukoaraiosis) e enfartes lacunares (Status lacunar), atrofia cerebral, gliose e acumulação excessiva de ferro nos núcleos da base. Esta tese, considerando a linha de investigação de que deriva – Ciências da Vida - especialização em Medicina Clinica - Biotecnologia, reúne e reflete sobre três vertentes ligadas à RM e interdependentes em relação a uma problemática comum. A primeira trata da caracterização da Tecnologia por Ressonância Magnética existente em Portugal, a qual inclui uma avaliação exploratória da aplicação da técnica de Difusão Anisotrópica nos estudos cerebrais. As dimensões analíticas estudadas foram a Tecnológica, Sociodemográfica e Económica. Na recolha de dados recorreu-se a várias fontes de informação e a uma metodologia exploratória faseada, validada pela triangulação dos resultados. A sua análise obedeceu a critérios de estratificação e agrupamento segundo as mesmas dimensões analíticas. Otimização da anisotropia fronto-calosa e [RM 1,5T] no idoso normal e com risco cérebro-vascular A segunda descreve o estudo anátomo – radiológico que recaiu sobre parâmetros de quantificação assente na temática do cérebro do idoso em cadáver segundo uma metodologia experimental aplicada às métricas da difusão por RM. Na terceira, e última, é apresentado o estudo técnico - radiológico para avaliação e otimização da imagem ponderada em difusão em estudos clínicos associados ao cérebro do idoso “The Usual Brain Aging” ou Envelhecimento Cerebral Normal, com base metodológica assente nos critérios e indicadores estabelecidos pelo Estudo de Imagem de Roterdão (Rotterdam Scan Study - RSS). Como principais resultados obteve-se que não existem em Portugal estruturas para avaliação dos equipamentos pesados ou Agência de Avaliação das Tecnologias da Saúde para desenvolver o importante papel da produção de estudos comparativos entre os equipamentos disponíveis no mercado, a relação preço-qualidade e a sua afetação às necessidades clínico-epidemiológicas. Constatou-se que a implementação de equipamentos de RM está fortemente assente em critérios económicos carecendo de recomendações e diretivas para o uso racionalizado destas tecnologias. Quanto a dados quantitativos concluímos que a maioria dos equipamentos está instalada em instituições privadas (80,2%); a intensidade de campo magnético mais frequente é [1,5T] com 119 equipamentos; os equipamentos estão instalados maioritariamente nos distritos de Lisboa (55 unidades) e do Porto (39 unidades); o rácio médio de equipamentos por habitante em Portugal é de 1 para 65 195 habitantes; a amplitude de gradientes com maior expressão na amostra é 30-39mT/m; a maioria dos equipamentos foi instalada no intervalo temporal [2009-2012] com 59 equipamentos; apenas 6 instituições desenvolvem investigação clinica e a maioria das bobinas para estudos de crânio são do tipo Array. O estudo de otimização da técnica da difusão revelou, quanto à avaliação dos valoresb, que os mais baixos (b=500 s/mm2 e b=1000 s/mm2), apresentam maior IS e SNR sendo esta uma boa medida referente à qualidade de imagem, no entanto, os valores-b mais elevados (b=2000 s/mm2) apresentam maior CNR e CR, face aos anteriores, o Otimização da anisotropia fronto-calosa e [RM 1,5T] no idoso normal e com risco cérebro-vascular que apesar de proporcionar inferior detalhe anatómico e consequentemente inferior qualidade de imagem, num encéfalo normal, pode auxiliar na interpretação e apresentar vantagens na identificação de lesões microvasculares sempre que persistirem dúvidas em relação ao diagnóstico diferencial de doença microvascular do tipo status lacunar ou Hiperintensidades da Substância Branca. As alterações deste parâmetro são particularmente refletidas nas diferenças da avaliação da qualidade de imagem na região fronto-calos Concluímos da avaliação quantitativa da concentração média de ferro (26Fe) em todas as faixas etárias que os núcleos da base que apresentam maior concentração são, por ordem decrescente: Substância Nigra, Globus Pallidus, Putamen, Tálamo, Núcleo Rubro e Núcleo Caudado; que existe uma predominância na concentração de ferro (26Fe) no hemisfério esquerdo e que os indivíduos do género masculino apresentam mais ferro (26Fe) que os do género feminino nas faixas etárias [30-40[, [40-50[ e [50-60[. Como principal conclusão do estudo da concentração média de ferro em relação à idade destacamos que a concentração média de ferro (26Fe) é superior nos grupos etários superiores, logo aumenta com a idade, sobretudo na Substância Nigra e no Núcleo Lenticular. No estudo técnico-radiológico encontrámos evidências do aumento da difusibilidade da água na substância branca subcortical dos sujeitos idosos comparativamente aos mais jovens. Uma relação idêntica foi avaliada nos tálamos. O aumento relacionado com a idade parece ser predominantemente observado em doentes com mais de 65 anos de idade o que pode refletir alterações estruturais ligeiras associados ao envelhecimento normal. Os resultados indicam que a análise quantitativa das imagens ponderadas em difusão fornece informações, sobre a estrutura do cérebro, as quai s não estão disponíveis apenas por inspeção visual, tanto nas imagens ponderadas em difusão como em outras sequências de aplicação clínica de rotina. Para fazer face às desvantagens dos sistemas de quantificação das HSB os quais são dispendiosos, complexos, requererem tecnologia e formação específicas, recomendamos que a aplicação automatizada GUIAL, desenvolvida ao longo do nosso trabalho é de utilização elementar e prática para que seja introduzida nos sistemas de aquisição de imagem por RM com o fim de integrar o processamento de imagem nos indivíduos portadores de fatores de risco vascular. A avaliação do ADC, nesta amostra, permitiu concluir que a variação desta variável é explicada estatisticamente pela existência da condição clínica de status lacunar em ambos os hemisférios ou por outras palavras, o status lacunar influencia o valor de ADC. Embora uma pequena percentagem da variação do ADC seja explicada pelo género, o ADC nos homens foi superior ao das mulheres o que não nos surpreende, pois são também os homens onde a frequência de doença microvascular se revelou mais expressiva. Os valores do ADC, globalmente, entre os hemisférios cerebrais não mostraram alterações exceto na SB entre os idosos e não idosos. A SB da região frontal mostrou valores diminuídos na anisotropia e isotropia face às restantes áreas anatómicas. Os estudos indicam que na idade avançada existe uma maior predisposição para suscetibilidades de estrutura com status de desconexão. A classificação das HSB foi superior em indivíduos mais velhos e com status lacunar, e em menor quantidade (inferior rating de classificação) nos indivíduos idosos sem status lacunar. As alterações volumétricas foram mais frequentes no homem do que na mulher, presumivelmente devido à associação com a elevada classificação de status lacunar. Um aumento do índice de Evan correspondeu, neste estudo, ao aumento das HSB, à diminuição do volume cerebral total, à expansão ventrículo-sulcal frontal e ao aumento da medida do ângulo caloso. Estes resultados foram agravados pela classificação elevada de status lacunar nos indivíduos que apresentaram indícios de doença de pequenos vasos, com manifesto aumento dos espaços de Virchow-Robin,enfartes lacunares ou HSB. Esses resultados foram mais expressivos no género masculino do que no feminino revelando uma maior vulnerabilidade sobretudo na atrofia frontal nos homens. Por sua vez as dimensões do Corpo Caloso tornaram-se reduzidas devido à compressão dos ventrículos laterais e terceiro ventrículo. Estes indicadores tiveram expressão particularmente nos indivíduos com mais de 65 anos. ---------------------------------------------------------------------------------------------ABSTRACT: The latest neuroimaging methods, such as Magnetic Resonance Imaging (MRI), have enabled detailed images of in vivo human brain. These images reveal often unexpected findings related to the normal pattern, with high predisposition for the elderly people with forthright coexistence with vascular risk factors such as characteristics of a normal aging process. Although it has been, in the last decade, several publications on this subject, it is still little explained. All over the world have emerged disease prevention programs and health promotion developed by the Public Health sector, supported mainly by the advancement of medical technologies that have resulted, among other impacts, in a growing of the elderly population. It is estimated, in 2030, a demographic composition with 20% o people over 65 years. In this context, microvascular disease is the most common cause of cognitive vascular impairment in the elderly and senescent characteristics in the MRI trough ischemic lesions of the white matter (Leukoaraiosis) and lacunar infarcts (lacunar status), cerebral atrophy, gliosis and iron accumulation in the basal ganglia in excess.This thesis, considering the research line that stems - Life Sciences - specialization in Clinical Medicine, Biotechnology, gathers and reflects on three aspects linked to MR, interdependent and related to a common problem. The first deals with the Magnetic Resonance Technology characterization in Portugal, which includes an exploratory evaluation of the implementation of Anisotropic Diffusion technique in brain studies. The analytical dimensions studied were the Technologic, Socio-demographic and Economics. Collecting data was supported by different sources of information and was applied an exploratory methodology whose results were validated by triangulation. The research method was grouped and stratified criteria under the same analytical dimensions. The second describes the anatomical study - which was focused on radiological measurement parameters, based on the brain’s specimen under an experimental methodology applied to MRI diffusion metrics. Radiological evaluation and optimization of the weighted image diffusion in clinical studies were associated with the brain of the elderly "The Usual Brain Aging" with methodological basis based on established criteria and indicators by Rotterdam Scan Study (RSS). The main results obtained reveal the inexistence frameworks in Portugal for evaluation of equipments or Agency of Health Technology to produce studies comparing the equipment available on the market, the value for money and its allocation to clinical and epidemiological needs. It was found that the implementation of MRI equipment is strongly based on economic criteria lacking recommendations and guidelines for the rationalized use of these technologies. As the quantitative data we conclude that most of the scanners are located in private clinical institutions (80,2%); the most frequent magnetic field intensity is [1.5T] with 119 scanners; the scanners are mainly installed in Lisbon (55 units) and Porto (39 units) districts; the average ratio of equipment per capita in Portugal is 1 to 65 195 people; the gradient power with higher expression in the sample is 30-39mT / m;most of the scanners were installed in the years range [2009-2012 years] with 59 equipment; only 6 clinical placements develop clinical research and the most coils for brain studies are of Array type. The optimization study of diffusion technique revealed, as the assessment of the bvalues, the lower (b = 500 s / mm2 and b = 1000 s / mm 2), promotes an increase in the SI and SNR being this measure related to a higher image quality, however the highest b values (b = 2000 s / mm 2) have a higher CNR (Contrast to Noise-Ratio) and CR (Contrast Ratio), compared to the previous ones. This may provide less anatomical details and, thus, ower image quality, of a normal brain, however can help the interpretation and have advantages in identifying microvascular injuries when doubts persist regarding the differential diagnosis of microvascular disease of lacunar or WMH (White Matter Hyperintensities) status type. Changes on this parameter are Otimização da anisotropia fronto-calosa e [RM 1,5T] no idoso normal e com risco cérebro-vascular particularly reflected in the differences of image quality evaluation in the frontocallosum anatomical area. We conclude from the quantitative assessment of the average concentration of iron (26Fe), in all age groups to the basal ganglia, that the higher concentrations are, in descending order: Nigral Substantia, Globus pallidus, Putamen, Thalamus, Rubio nucleus and Caudate nucleus; that there is a predominance in the concentration of iron (26Fe) in the left hemisphere and that male gender show higher iron (26Fe) level tha females, in the age groups [30-40 [[40-50 [and [50- 60 [. Regarding a main conclusion of the mean concentration study of iron, in terms of age we point out that the average concentration of iron (26Fe) is higher among older groups and increases with age, especially in Nigral Substantia and Lenticular Nucleus. On the technical and radiological study we found evidence of an increased in water /diffusivity in the ubcortical white matter of the elderly compared with younger subjects. A similar relationship was assessed in the Thalamus. The increase agerelated seems to be predominantly observed in patients over 65 years which may reflect minor structural changes associated with normal aging. The results indicate that quantitative analysis of diffusion weighted imaging can provide information about the structure of the brain which is not reached only by visual inspection or standard sequences applied in clinical routine. To address the disadvantages the systems of quantification of WMH which the authors state that are costly, complex, require specific technology and training, we recommend that the automated application GUIAL, developed over our work is basic and practical to use and to be introduced in MR image systems acquisition in order to integrate image processing in patients with vascular risk factors. The evaluation of the ADC showed that its variation is statistically explained by the existence of the medical condition of lacunar status, in both hemispheres, or in other words, the lacunar status influences the ADC value. Although a small percentage of the ADC variation is explained by gender, the ADC in men was higher than women which Otimização da anisotropia fronto-calosa e [RM 1,5T] no idoso normal e com risco cérebro-vascular do not surprise us, since they are also men where the frequency of microvasculardisease has proved more significant. The values of ADC, overall, between the cerebral hemispheres showed no changes but were different in WM among the elderly and non-elderly subjects.The WM's forehead showed decreased values in anisotropy and isotropy face the other anatomical areas. The studies indicate that in old age there is a greater tendency to higher susceptibility to disconnection- status framework. The classification of WMH was higher in elderly people and lacunar status, and fewer (lower classification rating) in the elderly without lacunar status. volumetric changes were more frequent in men than in women, most probably because of its association with high lacunar status rating. An increase of Evan index corresponded, in this study, to an increase in WMH, to a decreased of total brain volume, to a ventricle sulcal frontal and callous angle expansion. These results were wound up by high ranking of lacunar status in subjects who had small vessel disease, clear increase in spaces of Virchow-Robin, lacunar infarctions or WMH. These results were more significant in males than in females revealing vulnerability particularly in the frontal atrophy in men. In turn the size of Corpus Callosum because reduced due to the compression of the lateral and third ventricles. These indicators had expression particularly in individuals over 65 years.

Exploring TM image texture and its relationships with biomass estimation in Rondônia, Brazilian Amazon

Many texture measures have been developed and used for improving land-cover classification accuracy, but rarely has research examined the role of textures in improving the performance of aboveground biomass estimations. The relationship between texture and biomass is poorly understood. This paper used Landsat Thematic Mapper (TM) data to explore relationships between TM image textures and aboveground biomass in Rondônia, Brazilian Amazon. Eight grey level co-occurrence matrix (GLCM) based texture measures (i.e., mean, variance, homogeneity, contrast, dissimilarity, entropy, second moment, and correlation), associated with seven different window sizes (5x5, 7x7, 9x9, 11x11, 15x15, 19x19, and 25x25), and five TM bands (TM 2, 3, 4, 5, and 7) were analyzed. Pearson's correlation coefficient was used to analyze texture and biomass relationships. This research indicates that most textures are weakly correlated with successional vegetation biomass, but some textures are significantly correlated with mature forest biomass. In contrast, TM spectral signatures are significantly correlated with successional vegetation biomass, but weakly correlated with mature forest biomass. Our findings imply that textures may be critical in improving mature forest biomass estimation, but relatively less important for successional vegetation biomass estimation.

Automatic Mallampati Classification Using Active Appearance Models

Difficult tracheal intubation assessment is an important research topic in anesthesia as failed intubations are important causes of mortality in anesthetic practice. The modified Mallampati score is widely used, alone or in conjunction with other criteria, to predict the difficulty of intubation. This work presents an automatic method to assess the modified Mallampati score from an image of a patient with the mouth wide open. For this purpose we propose an active appearance models (AAM) based method and use linear support vector machines (SVM) to select a subset of relevant features obtained using the AAM. This feature selection step proves to be essential as it improves drastically the performance of classification, which is obtained using SVM with RBF kernel and majority voting. We test our method on images of 100 patients undergoing elective surgery and achieve 97.9% accuracy in the leave-one-out crossvalidation test and provide a key element to an automatic difficult intubation assessment system.

Modified 'dumbbell' technique: a simple and intuitive method to position balloon-expandable stent valves.

Intraoperative cardiac imaging plays a key role during transcatheter aortic valve replacement. In recent years, new techniques and new tools for improved image quality and virtual navigation have been proposed, in order to simplify and standardize stent valve positioning and implantation. But routine performance of the new techniques may require major economic investments or specific knowledge and skills and, for this reason, they may not be accessible to the majority of cardiac centres involved in transcatheter valve replacement projects. Additionally, they still require injections of contrast medium to obtain computed images. Therefore, we have developed and describe here a very simple and intuitive method of positioning balloon-expandable stent valves, which represents the evolution of the 'dumbbell' technique for echocardiography-guided transcatheter valve replacement without angiography. This method, based on the partial inflation of the balloon catheter during positioning, traps the crimped valve in the aortic valve orifice and, consequently, very near to the ideal landing zone. It does not require specific echocardiographic knowledge; it does not require angiographies that increase the risk of postoperative kidney failure in elderly patients, and it can be also performed in centres not equipped with a hybrid operating room.

II Workshop on the use of Computed Tomography (CT) in pig carcass classification. Other CT applications: live animals and meat technology, 16-17th April 2009, Monells (Girona), Spain

Es va realitzar el II Workshop en Tomografia Computeritzada (TC) a Monells. El primer dia es va dedicar íntegrament a la utilització del TC en temes de classificació de canals porcines, i el segon dia es va obrir a altres aplicacions del TC, ja sigui en animals vius o en diferents aspectes de qualitat de la carn o els productes carnis. Al workshop hi van assistir 45 persones de 12 països de la UE. The II workshop on the use of Computed Tomography (CT) in pig carcass classification. Other CT applications: live animals and meat technology was held in Monells. The first day it was dedicated to the use of CT in pig carcass classification. The segond day it was open to otehr CT applications, in live animals or in meat and meat products quality. There were 45 assistants of 12 EU countries.

Inexperienced sonographers can successfully visualize and assess a three-dimensional image of the fetal face using a standardized ultrasound protocol

Introduction: A standardized three-dimensional ultrasonographic (3DUS) protocol is described that allows fetal face reconstruction. Ability to identify cleft lip with 3DUS using this protocol was assessed by operators with minimal 3DUS experience. Material and Methods: 260 stored volumes of fetal face were analyzed using a standardized protocol by operators with different levels of competence in 3DUS. The outcomes studied were: (1) the performance of post-processing 3D face volumes for the detection of facial clefts; (2) the ability of a resident with minimal 3DUS experience to reconstruct the acquired facial volumes, and (3) the time needed to reconstruct each plane to allow proper diagnosis of a cleft. Results: The three orthogonal planes of the fetal face (axial, sagittal and coronal) were adequately reconstructed with similar performance when acquired by a maternal-fetal medicine specialist or by residents with minimal experience (72 vs. 76%, p = 0.629). The learning curve for manipulation of 3DUS volumes of the fetal face corresponds to 30 cases and is independent of the operator's level of experience. Discussion: The learning curve for the standardized protocol we describe is short, even for inexperienced sonographers. This technique might decrease the length of anatomy ultrasounds and improve the ability to visualize fetal face anomalies.

High-resolution selective three-dimensional magnetic resonance coronary angiography with navigator-echo technique: segment-by-segment evaluation of coronary artery stenosis.

PURPOSE: To investigate the feasibility of high-resolution selective three-dimensional (3D) magnetic resonance coronary angiography (MRCA) in the evaluation of coronary artery stenoses. MATERIALS AND METHODS: In 12 patients with coronary artery stenoses, MRCA of the coronary artery groups, including the coronary segments with stenoses of 50% or greater based on conventional x-ray coronary angiography (CAG), was performed with double-oblique imaging planes by orienting the 3D slab along the major axis of each right coronary artery-left circumflex artery (RCA-LCX) group and each left main trunk-left anterior descending artery (LMT-LAD) group. Ten RCA-LCX and five LMT-LAD MR angiograms were obtained, and the results were compared with those of conventional x-ray angiography. RESULTS: Among 70 coronary artery segments expected to be covered, a total of 49 (70%) segments were fully demonstrated in diagnostic quality. The identification of segmental location of stenoses showed as high an accuracy as 96%. The retrospective analysis for stenosis of 50% or greater yielded the sensitivity, specificity, and accuracy of 80%, 85%, and 84%, respectively. CONCLUSION: Selective 3D MRCA has the potential for segment-by-segment evaluation of major portions of the right and left coronary arteries with high accuracy.

IEF pattern classification-derived criteria for the identification of epoetin-delta in urine.

Epoetin-delta (Dynepo Shire Pharmaceuticals, Basing stoke, UK) is a synthetic form of erythropoietin (EPO) whose resemblance with endogenous EPO makes it hard to identify using the classical identification criteria. Urine samples collected from six healthy volunteers treated with epoetin-delta injections and from a control population were immuno-purified and analyzed with the usual IEF method. On the basis of the EPO profiles integration, a linear multivariate model was computed for discriminant analysis. For each sample, a pattern classification algorithm returned a bands distribution and intensity score (bands intensity score) saying how representative this sample is of one of the two classes, positive or negative. Effort profiles were also integrated in the model. The method yielded a good sensitivity versus specificity relation and was used to determine the detection window of the molecule following multiple injections. The bands intensity score, which can be generalized to epoetin-alpha and epoetin-beta, is proposed as an alternative criterion and a supplementary evidence for the identification of EPO abuse.

Cluster kernels for semisupervised classification of VHR urban images

In this paper, we present and apply a semisupervised support vector machine based on cluster kernels for the problem of very high resolution image classification. In the proposed setting, a base kernel working with labeled samples only is deformed by a likelihood kernel encoding similarities between unlabeled examples. The resulting kernel is used to train a standard support vector machine (SVM) classifier. Experiments carried out on very high resolution (VHR) multispectral and hyperspectral images using very few labeled examples show the relevancy of the method in the context of urban image classification. Its simplicity and the small number of parameters involved make it versatile and workable by unexperimented users.

Scene Classification Using a Hybrid Generative/Discriminative Approach

We investigate whether dimensionality reduction using a latent generative model is beneficial for the task of weakly supervised scene classification. In detail, we are given a set of labeled images of scenes (for example, coast, forest, city, river, etc.), and our objective is to classify a new image into one of these categories. Our approach consists of first discovering latent ";topics"; using probabilistic Latent Semantic Analysis (pLSA), a generative model from the statistical text literature here applied to a bag of visual words representation for each image, and subsequently, training a multiway classifier on the topic distribution vector for each image. We compare this approach to that of representing each image by a bag of visual words vector directly and training a multiway classifier on these vectors. To this end, we introduce a novel vocabulary using dense color SIFT descriptors and then investigate the classification performance under changes in the size of the visual vocabulary, the number of latent topics learned, and the type of discriminative classifier used (k-nearest neighbor or SVM). We achieve superior classification performance to recent publications that have used a bag of visual word representation, in all cases, using the authors' own data sets and testing protocols. We also investigate the gain in adding spatial information. We show applications to image retrieval with relevance feedback and to scene classification in videos

Automatic classification of breast density

A recent trend in digital mammography is computer-aided diagnosis systems, which are computerised tools designed to assist radiologists. Most of these systems are used for the automatic detection of abnormalities. However, recent studies have shown that their sensitivity is significantly decreased as the density of the breast increases. This dependence is method specific. In this paper we propose a new approach to the classification of mammographic images according to their breast parenchymal density. Our classification uses information extracted from segmentation results and is based on the underlying breast tissue texture. Classification performance was based on a large set of digitised mammograms. Evaluation involves different classifiers and uses a leave-one-out methodology. Results demonstrate the feasibility of estimating breast density using image processing and analysis techniques