No description available.

<p>The medical imaging system Sectra PACS from Sectra Imtec contains a 3D mode that can be used for visualising image stacks from e.g. computed tomography. Various structures of human anatomy can be visualised in the 3D mode, but visualisations of the bronchial tree of the lungs rarely become good enough to be useful. The goal of this work was to investigate ways of improving such visualisations.</p><p>Various approaches were studied, evaluated and tested. The fact that most effort was needed for small structures with sizes similar to the resolution of the images made things slightly more complicated. A method classifying neighbourhoods based on local structure emerged as most promising, and was used as foundation for a proposed algorithm. It creates a mask representing the presence of bronchi, allowing the hiding of uninteresting structures in its proximity. The algorithm was then implemented so that it could be tested together with the existing system.</p><p>The method was found to work well and was able to detect the smaller tubes of the bronchial tree and output the desired classification mask. Its usefulness was somewhat reduced by issues relating to speed, and the fact that many computed tomography image stacks lack the necessary resolution for visualising the finer details of the bronchial tree.</p>

No description available.

The wealth of medical-related information available today gives rise to a multidimensional source of knowledge. Research discoveries published in prestigious venues, electronic-health records data, discharge summaries, clinical notes, etc., all represent important medical information that can assist in the medical decision-making process. The challenge that comes with accessing and using such vast and diverse sources of data stands in the ability to distil and extract reliable and relevant information. Computer-based tools that use natural language processing and machine learning techniques have proven to help address such challenges. This current work proposes automatic reliable solutions for solving tasks that can help achieve a personalized-medicine, a medical practice that brings together general medical knowledge and case-specific medical information. Phenotypic medical observations, along with data coming from test results, are not enough when assessing and treating a medical case. Genetic, life-style, background and environmental data also need to be taken into account in the medical decision process. This thesis’s goal is to prove that natural language processing and machine learning techniques represent reliable solutions for solving important medical-related problems. From the numerous research problems that need to be answered when implementing personalized medicine, the scope of this thesis is restricted to four, as follows: 1. Automatic identification of obesity-related diseases by using only textual clinical data; 2. Automatic identification of relevant abstracts of published research to be used for building systematic reviews; 3. Automatic identification of gene functions based on textual data of published medical abstracts; 4. Automatic identification and classification of important medical relations between medical concepts in clinical and technical data. This thesis investigation on finding automatic solutions for achieving a personalized medicine through information identification and extraction focused on individual specific problems that can be later linked in a puzzle-building manner. A diverse representation technique that follows a divide-and-conquer methodological approach shows to be the most reliable solution for building automatic models that solve the above mentioned tasks. The methodologies that I propose are supported by in-depth research experiments and thorough discussions and conclusions.

The medical imaging system Sectra PACS from Sectra Imtec contains a 3D mode that can be used for visualising image stacks from e.g. computed tomography. Various structures of human anatomy can be visualised in the 3D mode, but visualisations of the bronchial tree of the lungs rarely become good enough to be useful. The goal of this work was to investigate ways of improving such visualisations. Various approaches were studied, evaluated and tested. The fact that most effort was needed for small structures with sizes similar to the resolution of the images made things slightly more complicated. A method classifying neighbourhoods based on local structure emerged as most promising, and was used as foundation for a proposed algorithm. It creates a mask representing the presence of bronchi, allowing the hiding of uninteresting structures in its proximity. The algorithm was then implemented so that it could be tested together with the existing system. The method was found to work well and was able to detect the smaller tubes of the bronchial tree and output the desired classification mask. Its usefulness was somewhat reduced by issues relating to speed, and the fact that many computed tomography image stacks lack the necessary resolution for visualising the finer details of the bronchial tree.

Thesis (Ph. D. in Biomedical Informatics)--Vanderbilt University, Dec. 2009. / Title from title screen. Includes bibliographical references.

Background: Despite growing utilization of mobile phones and websites for consumers seeking health care advice, the area is largely understudied. A niche market for these applications is in diabetes care. Since diabetes is a chronic condition requiring daily monitoring it is a good candidate for consumer health informatics and especially interactive websites and mobile phone applications. As the obesity epidemic continues, so too the prevalence of type 2 diabetes continues to rise. This chronic condition can lead to major complications and high medical cost. It is on the rise in countries all over the world, and beginning to impact people at younger ages. Low cost interventions are being explored to mitigate these complications and cost. Objective: To examine the effectiveness of consumer health informatics, such as websites, personal digital assistants, and mobile phone applications that claim to help people with diabetes self-monitor diet and exercise behaviours to lose weight. Methods: A search for relevant literature was conducted using PUBMED, Cochrane, and IEEE Xplore, with the search terms: (mhealth OR mobile health OR phone OR web* OR ehealth OR internet OR ICT) AND diabetes AND (diet* OR exercise OR physical activity). Also, a bibliographic search was done to identify any studies that were missed in the initial search. The search was not limited to any date range, but articles were identified from the time period of September 2000 through April 2012. Only articles in English were included. Studies were included if the program included an interactive logging feature for diet and/or physical activity. Studies that looked at type 1 disbetes were excluded. Results: A total of 10 original studies were found that met the inclusion criteria. Including 2 qualitative design, 1 randomized trial, and 7 randomized control trials. There was a great deal of heterogeneity among the studies. Delivery methods varies, studies including the following are: *  Mobile device only: 3 *  Website only: 6 *  Website plus mobile device: 1 Many different outcome measures were used across the studies including: behavioural, physiological, psychosocial, as well as usability and satisfaction. Overall, adherence and follow up were low. Dietary tracking generally appears not to be as effective as broad goals such as, “each more fruits and vegetables”. Exercise tracking was more effective at increasing physical activity. Message boards and peer support did not show an increase in effectiveness, but personal online coaches and personalized emails showed promising results. Usability and satisfaction was high in those that reported it, but the large number of dropouts are not considered in this. Conclusions: At this time, consumer health informatics does not seem to be an effective solution in facilitating significant behavior change for people who have type 2 diabetes. Future programs should look at ways to increase adherence and usage of the programs because the people who did use the programs daily benefited more than sporadic users. Components that showed promising results are access to a personal online coach, personalized weekly emails, integration with a pedometer that automatically uploads to a tracking program, and broader food related goals. Further testing is necessary to determine if this type of intervention is effective. / published_or_final_version / Public Health / Master / Master of Public Health

The aim of this research was to invetigate the clinical decision-making processes used in podiatry and hence to investigate if a computer expert system could be used to aid the process of clinical decision-making. This was achieved through a sequence of four empirical studies. The initial study used card sorts to investigate seven expert podiatrists’ perceptions of and attitudes toward diagnostic aids, and in particular how podiatrists viewed expert systems. The results showed that expert systems are perceived as different in kind from other diagnostic aids such as X-rays or blood tests. The second study was conducted using one expert and one novice podiatrist and used a task analysis to investigate the types of tasks and skills undertaken by a podiatrist during the diagnosis of a patient in different clinical environments. The results indicate that the work is highly schematised and involves routine tasks such as nail care and callus reduction. In clinic, podiarists perform many tasks quickly. There was little difference between the number of tasks per minute undertaken in a general clinic and the number of tasks in a specialist diabetes clinic. Considering the speed of diagnosis, it is postulated that both expert and novice podiatrists’ use of schemata, pattern matching, and tacit and implicit knowledge dominates their diagnostic activity during consultations. The third study focused on how clinical reasoning and decision-making occur during consultations with a patient. Think-aloud protocols were used to investigate the differences in the clinical reasoning process between five expert and nine novice podiatrists. The speed of diagnosis and general lack of causal assertions suggest that use of schemata and tacit knowledge dominate the diagnosis process for both experts and novices. In a general setting, the novices produced four common clinical reasoning themes. These indicate that pattern recognition is a common method of diagnosis. However, there was an increase in the number of clinical reasoning themes used by experts in a specialist setting, indicating novice—expert differences. The fourth study used laddering interviews on a mixture of twelve NHS and private podiatrists to investigate why podiatrists used certain clinical reasoning themes. A hierarchical value map was derived, showing that, at an initial response level to the laddering questions, certain values were important: the palpation of the foot, building a picture of the foot condition, and being able to use clinical reasoning frequently and immediately. The emphases on palpation and immediacy of reasoning suggest that an expert system is unlikely to serve podiatrists’ needs in clinics. This research has provided a new understanding of the clinical reasoning processes used in podiatry. A podiatrist has a very busy timeline when diagnosing a patient and predominantly uses (and values) tacit knowledge, implicit learning, and compiled skills during consultations. There is little evidence for the need or desire for an expert system in clinical podiatry practice. However, if such an expert system were to be created, then: (a) it would have to be fast and non-intrusive so it can fit into a very busy consultation timeline, (b) it would need a knowledge base that could account for diagnosis of foot and leg conditions based on pattern recognition, and (c) it might be most valuable in the form of a decision support system for professional development that included the full range of expert diagnostic themes

published_or_final_version / Community Medicine / Master / Master of Philosophy

The wealth of medical-related information available today gives rise to a multidimensional source of knowledge. Research discoveries published in prestigious venues, electronic-health records data, discharge summaries, clinical notes, etc., all represent important medical information that can assist in the medical decision-making process. The challenge that comes with accessing and using such vast and diverse sources of data stands in the ability to distil and extract reliable and relevant information. Computer-based tools that use natural language processing and machine learning techniques have proven to help address such challenges. This current work proposes automatic reliable solutions for solving tasks that can help achieve a personalized-medicine, a medical practice that brings together general medical knowledge and case-specific medical information. Phenotypic medical observations, along with data coming from test results, are not enough when assessing and treating a medical case. Genetic, life-style, background and environmental data also need to be taken into account in the medical decision process. This thesis’s goal is to prove that natural language processing and machine learning techniques represent reliable solutions for solving important medical-related problems. From the numerous research problems that need to be answered when implementing personalized medicine, the scope of this thesis is restricted to four, as follows: 1. Automatic identification of obesity-related diseases by using only textual clinical data; 2. Automatic identification of relevant abstracts of published research to be used for building systematic reviews; 3. Automatic identification of gene functions based on textual data of published medical abstracts; 4. Automatic identification and classification of important medical relations between medical concepts in clinical and technical data. This thesis investigation on finding automatic solutions for achieving a personalized medicine through information identification and extraction focused on individual specific problems that can be later linked in a puzzle-building manner. A diverse representation technique that follows a divide-and-conquer methodological approach shows to be the most reliable solution for building automatic models that solve the above mentioned tasks. The methodologies that I propose are supported by in-depth research experiments and thorough discussions and conclusions.