Download presentation
Presentation is loading. Please wait.
1
Bongile Mzenda, Alexander Gegov, David Brown
Improving the transparency in fuzzy modelling of radiotherapy margins in cancer treatment Bongile Mzenda, Alexander Gegov, David Brown
2
Overview Margins in radiotherapy Fuzzy networks Methodology Results
Conclusions
3
Margins in radiotherapy
Account for presence of organ motion, patient setup and delineation errors
4
Margins methods Shortcomings of presently used margin
derivations methods: Do not include delineation errors Do not consider dose effects on surrounding critical organs Cannot be adapted to changing patient conditions
5
Fuzzy networks Offer novel methodology to address above shortcomings
Consist of networked rule based systems Deal with process inputs sequentially while taking into account the interactions and the structure of the system
6
Fuzzy networks General structure
7
Methodology Treatment study used to deduce variation in tumour and critical organ dose sensitive parameters (V99% & V60) with errors Fuzzy network model design
8
Methodology Gaussian membership functions used for inputs and output
Linguistic composition of individual rule bases
9
Results Comparison to fuzzy system & Stroom et al statistical method
10
Results Comparison to fuzzy system & van Herk et al statistical method
11
Results Mean absolute error (MAE) analysis Transparency index (TI) TI
Fuzzy network Fuzzy system 1.25 4.00
12
Conclusions Use of fuzzy network resulted in better
correlation of input and output parameters Fuzzy network results lie in between currently used statistical methods Improved transparency from fuzzy network User friendly for clinical users to present their expert knowledge in rule design
Similar presentations
© 2025 SlidePlayer.com. Inc.
All rights reserved.