Meeting Abstract Exploring Viewing Behavior Data from ... · Predict Correctness of Students...
3
Meeting Abstract Exploring Viewing Behavior Data from Whole Slide Images to Predict Correctness of Students’ Answers during Practical Exams in Oral Pathology Slawomir Walkowski, 1 Mikael Lundin, 2 Janusz Szymas, 3 and Johan Lundin 2 1 Poznan University of Technology, M. Sklodowska-Curie Square 5, 60-965 Poznan, Poland 2 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, P.O. Box 20, 00014 Helsinki, Finland 3 Department of Clinical Pathology, Poznan University of Medical Sciences, Przybyszewski Street 49, 60-355 Poznan, Poland Correspondence should be addressed to Slawomir Walkowski; [email protected] Received 2 September 2014; Accepted 2 September 2014 Copyright © 2014 Slawomir Walkowski et al. is is an open access article distributed under the Creative Commons Attribution License, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited. Background e use of whole slide images (WSIs) allows tracking and recording how a histological slide is viewed. Gathered data about viewing behavior while interpreting WSIs may result in a variety of analyses. When the tracking is done during an exam, we can discover how students view WSIs. Moreover, we may try to correlate their way of viewing slides with correctness of the answers they give. Particularly, we can potentially find out to what extent a specific viewing behavior is likely to result in a correct or incorrect answer from a student. Method To record viewing behavior, we utilized a soſtware-based view path tracking method, which does not require any specialized equipment. It gathers information about subsequently viewed fragments (view fields) of WSIs. e method was used during exams in oral pathology in Poznan University of Medical Sciences in Poznan, Poland, in years 2012-2013. Each dental student was given 50 exam questions with a WSI attached to each of them. e students were informed about and agreed on the tracking. Stored data and further analysis results are anonymous and so far without any impact on the final students’ evaluation and scores. e WSI viewing system used during the exam was WebMicroscope (Fimmic Ltd, Helsinki, Finland) and the view path tracking method is an integrated, but optional, part of it. In total, we collected information about approximately 130,000 view fields coming from about 180 dental students viewing WSIs during the exams. Gathered data was analyzed numerically, with some help from generated visualizations. A set of statistics was calculated per student per question and it included, for example, number of view fields, magnification level, and dispersion of view fields. Statistical methods were used to assess the correlation between calculated metrics and correctness of students’ answers. We also utilized machine learning approaches to check to what extent viewing behavior data can be used to predict a correct or incorrect answer coming from a student. For this purpose, we used gathered and processed data as a labeled set of instances. Results Two exams were successfully conducted with the view path tracking turned on, which resulted in a dataset covering students’ WSI viewing behavior. e aggregated metrics depicted certain viewing patterns. Analysis of the calculated statistics allowed finding some correlations between metrics values and exam answers. When used as features for machine learning, the metrics helped estimate probabilities of answers correctness. Hindawi Publishing Corporation Analytical Cellular Pathology Volume 2014, Article ID 706470, 2 pages http://dx.doi.org/10.1155/2014/706470
Transcript of Meeting Abstract Exploring Viewing Behavior Data from ... · Predict Correctness of Students...
Meeting AbstractExploring Viewing Behavior Data from Whole Slide Images toPredict Correctness of Students’ Answers during Practical Examsin Oral Pathology
Slawomir Walkowski,1 Mikael Lundin,2 Janusz Szymas,3 and Johan Lundin2
1 Poznan University of Technology, M. Sklodowska-Curie Square 5, 60-965 Poznan, Poland2 Institute for Molecular Medicine Finland (FIMM), University of Helsinki, P.O. Box 20, 00014 Helsinki, Finland3Department of Clinical Pathology, Poznan University of Medical Sciences, Przybyszewski Street 49, 60-355 Poznan, Poland
Correspondence should be addressed to Slawomir Walkowski; [email protected]
Received 2 September 2014; Accepted 2 September 2014
Copyright © 2014 Slawomir Walkowski et al. This is an open access article distributed under the Creative Commons AttributionLicense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properlycited.
Background
The use of whole slide images (WSIs) allows tracking andrecording how a histological slide is viewed. Gathered dataabout viewing behavior while interpreting WSIs may resultin a variety of analyses. When the tracking is done during anexam, we can discover how students view WSIs. Moreover,we may try to correlate their way of viewing slides withcorrectness of the answers they give. Particularly, we canpotentially find out to what extent a specific viewing behavioris likely to result in a correct or incorrect answer from astudent.
Method
To record viewing behavior, we utilized a software-based viewpath trackingmethod, which does not require any specializedequipment. It gathers information about subsequently viewedfragments (view fields) ofWSIs.Themethod was used duringexams in oral pathology in Poznan University of MedicalSciences in Poznan, Poland, in years 2012-2013. Each dentalstudent was given 50 exam questions with a WSI attachedto each of them. The students were informed about andagreed on the tracking. Stored data and further analysisresults are anonymous and so far without any impact onthe final students’ evaluation and scores. The WSI viewingsystem used during the exam was WebMicroscope (Fimmic
Ltd, Helsinki, Finland) and the view path tracking method isan integrated, but optional, part of it. In total, we collectedinformation about approximately 130,000 view fields comingfrom about 180 dental students viewing WSIs during theexams. Gathered data was analyzed numerically, with somehelp from generated visualizations. A set of statistics wascalculated per student per question and it included, forexample, number of view fields, magnification level, anddispersion of view fields. Statistical methods were usedto assess the correlation between calculated metrics andcorrectness of students’ answers. We also utilized machinelearning approaches to check towhat extent viewing behaviordata can be used to predict a correct or incorrect answercoming from a student. For this purpose, we used gatheredand processed data as a labeled set of instances.
Results
Two exams were successfully conducted with the view pathtracking turned on, which resulted in a dataset coveringstudents’ WSI viewing behavior. The aggregated metricsdepicted certain viewing patterns. Analysis of the calculatedstatistics allowed finding some correlations between metricsvalues and exam answers. When used as features for machinelearning, the metrics helped estimate probabilities of answerscorrectness.
Hindawi Publishing CorporationAnalytical Cellular PathologyVolume 2014, Article ID 706470, 2 pageshttp://dx.doi.org/10.1155/2014/706470
2 Analytical Cellular Pathology
Conclusion
Software-based view path tracking appears to be a usefulmethod of discovering WSI viewing behavior and investigat-ing decision making process of dental students who take apractical exam in oral pathology. Analysis of collected dataprovides interesting insights into how the slides are viewed,how the viewing patterns correlate with students’ answers,and what the potential of the view path tracking data is whenpredicting correctness of the answers.
Submit your manuscripts athttp://www.hindawi.com
Stem CellsInternational
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
MEDIATORSINFLAMMATION
of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Behavioural Neurology
EndocrinologyInternational Journal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Disease Markers
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
BioMed Research International
OncologyJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Oxidative Medicine and Cellular Longevity
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
PPAR Research
The Scientific World JournalHindawi Publishing Corporation http://www.hindawi.com Volume 2014
Immunology ResearchHindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Journal of
ObesityJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Computational and Mathematical Methods in Medicine
OphthalmologyJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Diabetes ResearchJournal of
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Research and TreatmentAIDS
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Gastroenterology Research and Practice
Hindawi Publishing Corporationhttp://www.hindawi.com Volume 2014
Parkinson’s Disease
Evidence-Based Complementary and Alternative Medicine
Volume 2014Hindawi Publishing Corporationhttp://www.hindawi.com
