Click here to


Are you sure ?

Yes, do it No, cancel

Determining Necessity of Online Adaptive Replanning Based On Image Similarity for MRI-Guided Radiation Therapy

S Lim*, Y Zhang , E Ahunbay , E Paulson , X Chen , X Li , Medical College of Wisconsin, Milwaukee, WI


(Sunday, 7/14/2019) 5:00 PM - 6:00 PM

Room: Stars at Night Ballroom 2-3

Purpose: Online adaptive replanning (OLAR), enabled with the recently-introduced MR-Linac, can substantially improve radiation therapy. It is labor intensive, time consuming, and unneeded for minimal interfractional deformations. Unnecessary effort is avoided if need for OLAR is determined a priori. This work aims to develop a method to rapidly determine the need for OLAR by analyzing image similarity between the reference and fractional MRIs.

Methods: Longitudinal fractional (weekly or daily) MRIs acquired using a MR-Linac (Unity, Elekta) for representative patients with pancreatic or liver cancers were analyzed. Fractional MRIs were pre-processed (e.g., bias correction, normalization) to remove image variations. They were classified as replanning or repositioning sets based on both commonly-used dose-volume criteria and comparison to reference plan. For each case, contours of the planning target volume (PTV), organs-at-risk (OARs) and a 20-mm expansion ring from PTV, created off-line on the reference MRI, were populated onto fractional MRIs. A series of image texture features in the ring were extracted from both reference and fractional MRIs. Features that describe differences between the two image sets were selected to define an image similarity index (ISI). An ISI threshold that indicates if OLAR is necessary was determined based on the patient dataset. The obtained ISI threshold was tested using different fractional MRIs.

Results: Preliminary results show that an ISI value close to 0 indicated minimal anatomical changes in PTV and/or OARs inside the ring, thus, OLAR is not necessary. For the replanning MRI set tested, the average ISI was 0.37 ± 0.13, while it was 0.12 ± 0.05 for the repositioning set.

Conclusion: The proposed image similarity index can be used to determine if online adaptive replanning is necessary without time-consuming segmentation, thus, can be implemented in MRI-guided OLAR to avoid unnecessary OLAR effort.


Not Applicable / None Entered.


Not Applicable / None Entered.

Contact Email