Differences

This shows you the differences between two versions of the page.

Link to this comparison view

offers:diplomthemen [2018/05/16 11:00]
petra [Laufende Bachelorarbeiten]
offers:diplomthemen [2018/09/12 10:10] (current)
petra [Laufende Bachelorarbeiten]
Line 12: Line 12:
  
 ====== Masterthemen in Magdeburg ====== ====== Masterthemen in Magdeburg ======
 +
 |< 100% 120px - - >| |< 100% 120px - - >|
 ^ Detektion von Aneurysmen mit Deep Learning^^  ^ Detektion von Aneurysmen mit Deep Learning^^ 
Line 97: Line 98:
  
 |< 100% 120px - - >| |< 100% 120px - - >|
-^Bachelor's/ Master's Thesis: Classification and Visualization of Plantar Pressure and Temperature Time-Series in Patients with Diabetic Foot Syndrome^^  +Visual Analytics of Intracranial Aneurysm Classification & Similarity Matching^^ 
-|  {{:offers:diplomthemen:ba_diabetic_foot_syndrome.png? 120}}| Complications afflicted with the diabetic foot syndrome have a substantial impact on the patient’s life qualityThe combination of vascular constrictions and nerve damages (neuropathyleads to a disturbed pressure and pain perceptionAs a consequencehigh local pressure strains remain unnoticed which lead to an impeded regional blood flow in the short term and tissue destructions and ulcerations in the long term. <html><br /></html> In collaboration with the university clinic Magdeburg, study with group of dia-betic patients with severe polyneuropathy and a group of healthy controls was conducted where pressure and temperature was collected with means of an “intelligent” shoe insole. The insole is equipped with multiple sensors which gather pressure and temperature signals and transfer them via Bluetooth to dedicated smartphone app for further analysis. One of the goals is the timely detection of emerging foot ulcerations which is characterized by a significant temperature in-crease of affected regions. <html><br /></html> The goal of the project is the classification of pressure and temperature time series to distinguish between neuropathic diabetics and healthy controls. For examplewe want to study whether disease-specific patterns in the run of the time series can be detected. Predictive features and time-series segments (motifsshould be illustrated by suitable visualizations which can be directly used by the clinical partners. | +|  {{:offers:diplomthemen:ba_va_for_aneurysm_rupture.png? 120}} | Intracranial aneurysms are pathologic dilations of the intracranial vessel wall. They bear the risk of rupture and thus subarachnoidal hemorrhages with often fatal consequences for the patient. Since treatment may cause severe complications as well, substantial research was carried out to characterize the patient-specific rupture risk based on various morphological and hemodynamic parametersClinicians often adapt their treatment decisions by analyzing similar pathologies and conditions with respect to their treatment outcome. For this purpose, we provide a reference database (serving as our training data set).  The goal is to identify the most similar reference cases for new aneurysmThe similarity comprises various factors, e.g. location of the aneurysm, size or previously extracted shape parametres.  <html> <br /> </html> Questions: 1) Given an aneurysm of interest, which are the k most similar aneurysms from the training set (i.e. the reference database)?  2) How much more similar is an aneurysm of interest to its most/second-most/etc. similar aneurysm from the training set in comparison with the average similarity towards an arbitrary aneurysm? 3) How does similarity change when the value of feature F is altered to x? <html> <br /> </html> Similarity calculation is dependent on the feature space. There should be two options for selecting an appropriate feature space: a) Supervised feature selection using target variable (rupture status, course of treatment, ...). Example: Correlation-based feature selection, b) Expert input. The medical expert (radiologist) selects set of relevant features based on his knowledge. <html> <br /> </html> The proposed Visual Analytics system should contain the following components: i) a G U I for comparing aneurysms based on the most important parameters and similarity as described aboveii) input panels for similarity calculations, iii) a radar chart like visualization for juxtaposing 2 or more aneurysms w.r.t. a set of features, ivfurther components upon consultation. |
 |**Prerequisites:**| Experience with R (preferred), Python or MATLAB; working knowledge of data mining | |**Prerequisites:**| Experience with R (preferred), Python or MATLAB; working knowledge of data mining |
-|**Further Information:**|{{:offers:diplomthemen:ba_diabetic_foot_syndrome.pdf|Flyer (in German and English)}}| +|**Contact:** | [[uli.niemann@ovgu.de|Uli Niemann, M.Sc.]], [[sylvia@isg.cs.uni-magdeburg.de|DrSylvia Saalfeld]] 
-|**Contact:** | [[uli.niemann@isg.cs.uni-magdeburg.de|Uli Niemann, M.Sc.]]|+|**Further infos:** | The topic is available as Bachelor's/Master's Thesis, but it can be converted into a student assistant job (//HiWi//). |
  
 <html><br></html> <html><br></html>
  
 |< 100% 120px - - >| |< 100% 120px - - >|
-^Bachelor's/ Master's Thesis: Interactive Subgroup Discovery in Cohort Study Data^^  +^Bachelor's/ Master's Thesis: Classification and Visualization of Plantar Pressure and Temperature Time-Series in Patients with Diabetic Foot Syndrome^^  
-|  {{:offers:diplomthemen:ba_interactive_subgroup_discovery.png? 120}}| Subgroup Discovery Algorithms aim to find coherent, easy-to-interpret rules concerning target variable and a quality criterionFor instance, in a medical application rules in the form of //(glucose 7.0 mu/l AND sex = women) --Hepatic Steatosis = TRUE/may be found. These rules describe subpopulations whose distribution with respect to an outcome considerably deviates from the whole population. Howeverthe process of finding an actually interesting subset of rules is hampered by ahigh redundancy towards instance coverage, b) tedious parameter tuning and c) necessary manual post-filtering step of the results and means a considerable effort for the data analystTherefore, it is necessary to involve the application expert into the subgroup discovery process. The target of the project is to implement an interactive Subgroup Discovery algorithm which incorporates user feedback during candidate generation within a beam search to increase the quality of the returned set of rules with respect to the above mentioned problems.| +|  {{:offers:diplomthemen:ba_diabetic_foot_syndrome.png? 120}}| Complications afflicted with the diabetic foot syndrome have substantial impact on the patient’s life quality. The combination of vascular constrictions and nerve damages (neuropathy) leads to disturbed pressure and pain perceptionAs a consequencehigh local pressure strains remain unnoticed which lead to an impeded regional blood flow in the short term and tissue destructions and ulcerations in the long term. <html><br /></html> In collaboration with the university clinic Magdeburg, a study with group of dia-betic patients with severe polyneuropathy and a group of healthy controls was conducted where pressure and temperature was collected with means of an “intelligent” shoe insole. The insole is equipped with multiple sensors which gather pressure and temperature signals and transfer them via Bluetooth to dedicated smartphone app for further analysisOne of the goals is the timely detection of emerging foot ulcerations which is characterized by a significant temperature in-crease of affected regions<html><br /></html> The goal of the project is the classification of pressure and temperature time series to distinguish between neuropathic diabetics and healthy controls. For example, we want to study whether disease-specific patterns in the run of the time series can be detected. Predictive features and time-series segments (motifs) should be illustrated by suitable visualizations which can be directly used by the clinical partners. | 
 |**Prerequisites:**| Experience with R (preferred), Python or MATLAB; working knowledge of data mining | |**Prerequisites:**| Experience with R (preferred), Python or MATLAB; working knowledge of data mining |
-|**Further Information:**|{{:offers:diplomthemen:ba_interactive_subgroup_discovery.pdf|Flyer (in German and English)}}|+|**Further Information:**|{{:offers:diplomthemen:ba_diabetic_foot_syndrome.pdf|Flyer (in German and English)}}|
 |**Contact:** | [[uli.niemann@isg.cs.uni-magdeburg.de|Uli Niemann, M.Sc.]]| |**Contact:** | [[uli.niemann@isg.cs.uni-magdeburg.de|Uli Niemann, M.Sc.]]|
  
Line 167: Line 168:
 ====== Laufende Masterarbeiten ====== ====== Laufende Masterarbeiten ======
 ^Name, Vorname^Titel der Arbeit ^Institut/Firma^Betreuer^ ^Name, Vorname^Titel der Arbeit ^Institut/Firma^Betreuer^
-|Sabsch, Tim|Supporting Epidemiological Reasoning from Machine Learning Models for Cohort Study Data|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim| 
-|Dietze, Denis|Entwicklung von Algorithmen zur optischen Montageprüfung variantenreicher Baugruppen|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> | 
 |Uderhardt, Ulrike|Modellierung der Populationsdynamik von rezeptiven Feldern der Fingerspitze im primären somatosensorischen Kortex mit Hilfe von 7 Tesla fMRT|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> | |Uderhardt, Ulrike|Modellierung der Populationsdynamik von rezeptiven Feldern der Fingerspitze im primären somatosensorischen Kortex mit Hilfe von 7 Tesla fMRT|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> |
 +|Vuong, Claudia| Unterstützung von Sicherheitsunterweisungen durch Gamification und Virtual Reality|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> |
 +|Martinke, Hannes|Qualitative Visual Analysis of Blood Vessel Morphology|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> |
 +|Rotärmel, Eduard|Augmented Reality Benutzerschnittstellen für Pilotassistenz|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> |
 +|Stecklina, Marianne|Recognizing Entities in Scanned Business Documents using Deep Learning|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html><br></html> |
 ====== Laufende Bachelorarbeiten ====== ====== Laufende Bachelorarbeiten ======
 ^Name, Vorname^Titel der Arbeit ^Institut/Firma^Betreuer^ ^Name, Vorname^Titel der Arbeit ^Institut/Firma^Betreuer^
Line 175: Line 178:
 |Knaus, Marina|Generierung und Evaluierung eines anwendergerechten User-Interfaces für einen produktspezifischen Konfigurator in einer VR-Umgebung| Universität Magdeburg |<html></html> Bernhard Preim<html>&nbsp;</html><html><br></html> | |Knaus, Marina|Generierung und Evaluierung eines anwendergerechten User-Interfaces für einen produktspezifischen Konfigurator in einer VR-Umgebung| Universität Magdeburg |<html></html> Bernhard Preim<html>&nbsp;</html><html><br></html> |
 |Bloemer, André|Anwendungsmöglichkeiten und Potenziale von Augmented Reality im Bereich Kulturerbe| Universität Magdeburg |<html></html> Bernhard Preim<html>&nbsp;</html><html><br></html> | |Bloemer, André|Anwendungsmöglichkeiten und Potenziale von Augmented Reality im Bereich Kulturerbe| Universität Magdeburg |<html></html> Bernhard Preim<html>&nbsp;</html><html><br></html> |
 +|Allgaier, Maren|Algorithmus zur Segmentierung peripherer Bronchien in Volumenbilddaten|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html>Sylvia Saalfeld<br></html> |
 +|Seeska, Tom|Computergestützte Visualisierung und morpholgische Analyse von Microgliazellen aus konfokalmikroskopischen Daten|Universität Magdeburg|<html></html> Bernhard<html>&nbsp;</html>Preim <html>Ulrich Kalinke<br></html> |
 +