Abstract
This master thesis provides the basis for an algorithm realizing in-line monitoring of a pharmaceutical tablet coating process. The used measurement method is optical coherence tomography. The Goal is to detect single tablets within images, which are generated from the measured data. Additionally the position of the tablets in relation to the measurement system should be determined.
Methods:
The algorithms are implemented in Matlab (Version 8.2, The MathWorks Inc., Natick, Massachusetts/USA). The data is classified using the methods "logistic regression", "support vector machines" and "hidden markov models" and the position of the tablets is determined by a dedicated "circle fitting" algorithm.
Results:
The classification of the data offers an accuracy of nearly 96%, a successful position determination is achieved for approximatly 75% of the detected tablets.
Conclusion:
Accuracy and speed of the implemented algorithm meet the requirements of an in-line monitoring tool for film coating processes.
Methods:
The algorithms are implemented in Matlab (Version 8.2, The MathWorks Inc., Natick, Massachusetts/USA). The data is classified using the methods "logistic regression", "support vector machines" and "hidden markov models" and the position of the tablets is determined by a dedicated "circle fitting" algorithm.
Results:
The classification of the data offers an accuracy of nearly 96%, a successful position determination is achieved for approximatly 75% of the detected tablets.
Conclusion:
Accuracy and speed of the implemented algorithm meet the requirements of an in-line monitoring tool for film coating processes.
Translated title of the contribution | Optical coherence tomography for classification and position evaluation of tablets in a pharmaceutical coating process |
---|---|
Original language | German |
Qualification | Master of Science |
Awarding Institution |
|
Supervisors/Advisors |
|
Publication status | Published - 2014 |
Keywords
- tablet coating
- optical coherence tomography
- support vector machines
- hidden markov model
- circle fitting