ESTRO 2024 - Abstract Book

S3947

Physics - Image acquisition and processing

ESTRO 2024

Using only the breathing patterns as input our best performing model (XGBoost) predicted with an accuracy of 75%, whether the corresponding scan had noticeable artifacts that might lead to repeat imaging.

As a visual example, in Figure 1 the performance of the model is shown for patient A with regular breathing and no considerable artifacts classified as ‘no artifact’ and patient B with more irregular breathing and strong artifacts classified as ‘artifact’.

Conclusion:

The results demonstrate that it’s feasible to approximate the image quality of a scan based on the corresponding breathing pattern. Further tests on more data must be conducted to specifically assess the benefit in a clinical setting and extend the model.

Keywords: 4DCT, Machine Learning, Image Quality

References:

[1] Werner e al., “Comparison of intelligent 4D CT sequence scanning and conventional spiral 4D CT: a first comprehensive phantom study”, 2021, Physics in Medicine & Biology.

[2] Olsen et al., “Effect of novel amplitude/phase binning algorithm on commercial four-dimensional computed tomography quality”, 2008, International Journal of Radiation Oncology Biology Physics.

[2] “Time series feature extraction on basis of scalable hypothesis tests (tsfresh – a python package)”, 2018, Neurocomputing.

2706

Poster Discussion

APTw-CEST MRI for personalized target delineation of glioblastoma: A prospective pilot study.

Patrick L.Y. Tang 1,2,3 , Marion Smits 1,2,4 , Remi A. Nout 3 , Caroline van Rij 1,3 , Cleo Slagter 1,3 , Annemarie T. Swaak-Kragten 1,3 , Esther A.H. Warnert 1,2 , Alejandra Méndez Romero 1,3 1 Erasmus MC Cancer Institute, University Medical Center, Brain Tumor Center, Rotterdam, Netherlands. 2 Erasmus MC, University Medical Center, Department of Radiology & Nuclear medicine, Rotterdam, Netherlands. 3 Erasmus MC Cancer Institute, University Medical Center, Department of Radiotherapy, Rotterdam, Netherlands. 4 Medical Delta, Cancer Diagnostics 3.0, Delft, Netherlands

Purpose/Objective: