Objective: Childhood maltreatment (CM) is a significant risk factor for the development and severity of bipolar disorder (BD) with increased risk of suicide attempts (SA). This study evaluated whether a machine learning algorithm could be trained to predict if a patient with BD has a history of CM or previous SA based on brain metabolism measured by positron emission tomography. Methods: Thirty-six euthymic patients diagnosed with BD type I, with and without a history of CM were assessed using the Childhood Trauma Questionnaire. Suicide attempts were assessed through the Mini International Neuropsychiatric Interview (MINI-Plus) and a semi-structured interview. Resting-state positron emission tomography with 18F-fluorodeoxyglucose was conducted, electing only grey matter voxels through the Statistical Parametric Mapping toolbox. Imaging analysis was performed using a supervised machine learning approach following Gaussian Process Classification. Results: Patients were divided into 18 participants with a history of CM and 18 participants without it, along with 18 individuals with previous SA and 18 individuals without such history. The predictions for CM and SA were not significant (accuracy = 41.67%; p = 0.879). Conclusion: Further investigation is needed to improve the accuracy of machine learning, as its predictive qualities could potentially be highly useful in determining histories and possible outcomes of high-risk psychiatric patients.

Resumo Objetivo: Comparar protocolos automatizados e manuais de ressonância magnética para estimar a concentração hepática de ferro em 1,5 T. Materiais e Métodos: Foi realizada ressonância magnética hepática em 53 pacientes com suspeita de sobrecarga de ferro hepática e 21 controles, seguida da estimativa cega da concentração hepática de ferro por dois examinadores usando mapas automáticos T2* e T1, assim como o manual T2* e o método signal-intensity-ratio. O desempenho foi medido usando curvas ROC e a correlação interobservador e intraobservador usando o coeficiente de correlação intraclasse bidirecional. Resultados: O desempenho da curva ROC separando pacientes e controles mostrou áreas sob a curva de 0,912 para o mapa automático T2*, 0,934 para o método signal-intensity-ratio, 0,908 para manual T2* e 0,80 para mapa T1 (este difere significativamente dos outros três métodos). Houve boa correlação interobservador e intraobservador (coeficiente de correlação intraclasse entre 0,938 e 0,998; p < 0,05). Correlações envolvendo o mapa T1, embora ainda significativas, foram menores. Conclusão: Em 1,5 T, o mapa T2* representa uma nova ferramenta rápida e promissora para avaliar o diagnóstico de sobrecarga de ferro hepática, enquanto o mapa T1 mostrou menor precisão. O desempenho do mapa T1 foi menor que o dos métodos T2*.

Abstract Objective: To compare automated and manual magnetic resonance imaging protocols for estimating liver iron concentrations at 1.5 T. Materials and Methods: Magnetic resonance imaging examination of the liver was performed in 53 patients with clinically suspected hepatic iron overload and in 21 control subjects. Liver iron concentrations were then estimated by two examiners who were blinded to the groups. The examiners employed automated T2* and T1 mapping, as well as manual T2* and signal-intensity-ratio method. We analyzed accuracy by using ROC curves. Interobserver and intraobserver agreement were analyzed by calculating two-way intraclass correlation coefficients. Results: The area under the ROC curve (to discriminate between patients and controls) was 0.912 for automated T2* mapping, 0.934 for the signal-intensity-ratio method, 0.908 for manual T2*, and 0.80 for T1 mapping, the last method differing significantly from the other three. The level of interobserver and intraobserver agreement was good (intraclass correlation coefficient, 0.938-0.998; p < 0.05). Correlations involving T1 mapping, although still significant, were lower. Conclusion: At 1.5 T, T2* mapping is a rapid tool that shows promise for the diagnosis of liver iron overload, whereas T1 mapping shows less accuracy. The performance of T1 mapping is poorer than is that of T2* methods.

Resumo Introdução: a prevalência de fungemia hospitalar tem aumentado em todo o mundo e a mortalidade por essa afecção é elevada. Objetivo: avaliar a evolução, na última década, da prevalência e do perfil dos agentes fúngicos isolados em hemoculturas realizadas em um hospital universitário terciário. Método: foram analisados retrospectivamente todos os resultados de hemocultura processados no Hospital das Clínicas da Universidade Federal de Minas Gerais (HC-UFMG), entre os períodos de 2001-2003 e de 2011-2013. Para cada triênio foram registrados o número de hemoculturas coletadas, o percentual de positividade geral e o percentual de fungemia. Também foram catalogadas todas as espécies fúngicas identificadas. Todas as amostras sanguíneas foram incubadas no sistema de automação BacT/ALERT® (bioMérieux). Resultados: entre 2001-2003, foram avaliadas 34.822 amostras, sendo 5.510 (15,8%) positivas. Entre 2011-2013, o número de hemoculturas processadas aumentou para 55.052 amostras, sendo 4.873 (8,9%) positivas. Observou-se um aumento do número de culturas positivas para fungos no período analisado (2001-2003: 4,16%; 2011-2013: 5,95%; p<0,001). Dentre os agentes, as candidemias foram predominantes, principalmente por espécies de Candida não albicans (2001-2003: 57,64%; 2011-2013: 65,17%; p<0,05). Houve também aumento da fungemia por outros gêneros (2001-2003: 2,62%; 2011-2013: 4,48%; p<0,01). Conclusão: houve aumento da prevalência de fungemia na última década no HC-UFMG. Embora as candidemias tenham sido responsáveis pela maioria dos casos, houve aumento de fungemias causadas por outras espécies.

Summary Introduction: The prevalence of nosocomial fungemia has increased worldwide, and mortality caused by this disease is high. Objective: To assess progress in the last decade, and the prevalence and profile of fungal agents isolated in blood cultures performed in a tertiary university hospital. Method: All the results of blood cultures processed at Hospital das Clínicas, Universidade Federal de Minas Gerais (HC-UFMG), in the time intervals 2001-2003 and 2011-2013 were analyzed retrospectively. For each three-year period, the number of collected blood cultures, the overall positivity rate and the percentage of fungemia were recorded. In addition, all identified fungal species were cataloged. All blood samples were incubated in the BacT/ALERT® (bioMérieux) automation system. Results: In 2001-2003, 34,822 samples were evaluated, with 5,510 (15.8%) positive results. In 2011-2013, the number of blood cultures processed increased to 55,052 samples, with 4,873 (8.9%) positive results. There was an increase in the number of positive cultures for fungi in the analyzed period (2001-2003: 4.16%; 2011-2013: 5.95%; p<0.001). Among the agents, candidemias were predominant, especially those caused by non-albicans Candida species (2001-2003: 57.64%; 2011-2013: 65.17%; p<0.05). There was also an increase in fungemia caused by other genera (2001-2003: 2.62%; 2011-2013: 4.48%; p<0.01). Conclusion: There was an increase in the prevalence of fungemia in the last decade at HC-UFMG. Although candidemias have been responsible for most of the cases, there has been an increase in fungemias caused by other species.

O ensino médico, de modo geral, envolve a associação entre teoria e prática, o que é de particular importância no ensino da área de diagnóstico por imagem. A metodologia do ensino baseado em problemas, associada ao ensino a distância, permite alcançar um público amplo e estimular o raciocínio clínico, o espírito crítico e o autodidatismo, características essenciais ao profissional médico. Nesse âmbito, surgiu o projeto "Imagem da Semana", que consiste em uma página no site da Faculdade de Medicina aberta ao público, na qual semanalmente é publicado um novo caso clínico, associado a uma imagem e a uma pergunta objetiva. O visitante é encorajado a responder à pergunta corretamente para ter acesso à discussão do caso e aprender mais sobre o diagnóstico e os diagnósticos diferenciais da doença abordada. O projeto tem sido muito bem recebido pela comunidade acadêmica, servindo também para a educação continuada de médicos generalistas e especialistas, além de divulgar o conhecimento produzido na universidade. Acreditamos que a democratização do conhecimento é um passo importante rumo à melhoria do ensino médico.

It is known that the teaching of medicine involves associating theory and practice. This is of particular importance in the teachings of diagnosis by image. The methodology of problem based learning allows for the reaching of a wide range of people and stimulates clinical and critical thinking and self teaching, essential characteristics of every physician. Having this in mind, we started the project "Image of the Week". The project consists of an internet page in the medical school's website where every week a case and an image is published, together with an objective question. The visitor is encouraged to answer the question correctly to gain access to the case discussion and be able to learn more about the diagnosis and differential diagnosis of the presented disease. The project has been very successful in the academic community and has also been used for continued education by general practitioners and medical specialists, besides sharing the knowledge produced in the University. We believe that the democratization of knowledge is a big step towards the improvement of medical teaching.