4D Treatment (Planning) Workshop
at GSI Darmstadt
Theorieseminarraum -
Information

Title:	Dr.
Family Name:	Riboldi
First Name:	Marco
Gender:	male
E-Mail:	marco.riboldipolimi.it
Profession:	Engineer
Affiliation for Badge:	Politecnico di Milano
Institute:
Address (Street):	P.za Leonardo da Vinci 32
Postal Code, City:	20133 Milano
Country:	Italy
Phone:
Fax:
Arrival Date:	December 8
Departure Date:	December 10
Hotel Information:	NONE
Tour:	No
Author(s):	Riboldi M(1), Gianoli C(1), Torshabi AE(2), Pella A(1), Baroni G(1,2)
Institute(s):	(1) Bioengineering Department, Politecnico di Milano, Milano (IT) (2) Bioengineering Unit, Fondazione CNAO, Pavia (IT)
Title:	Breathing phase detection and motion monitoring for 4D treatment planning and delivery
Abstract ( 300 words ):	Four-dimensional (4D) treatment planning and delivery require accurate breathing phase identification. Conventional techniques are based on a single external surrogate to provide a signal that correlates to breathing motion. This typically results in motion artifacts at the treatment planning stage, along with a reduced confidence in real-time tumor position estimation during treatment. In this contribution we present a 4DCT image sorting strategy based on the detection of multiple external surrogates and data mining techniques. The k-means clustering method is applied to classify CT images acquired over multiple breathing cycles in different respiratory phases. A complimentary analysis of different approaches to infer tumor motion based on multiple external surrogates over a long time period is presented. The implemented correlation models comprise linear/quadratic correlation, artificial neural networks and fuzzy logic. Results show that the availability of redundant motion surrogates may result in enhanced breathing phase identification and reduced 4DCT image artifacts. The implemented correlation models achieve an error reduction up to 10.8% for the fuzzy logic approach with respect to linear/quadratic correlation, measured at the 95% confidence level. In conclusion, multiple surrogates are suggested to optimize breathing phase detection. The combination of multiple surrogates with complex non linear techniques is put forward to enhance tumor motion prediction over long time periods.

TPWorkshop2010 Registration Main Page