Respiratory movement correction through object tracking in perfusion imaging of lung tumours: Proof of principle

Ziele: Therapy assessment in solid tumours is moving away from morphologic criteria (e.g. RECIST) to functional imaging of biological effects of new drugs (e.g. VEGF inhibitors). Respiratory motion is a challenge to imaging of microvascular properties in lung tumours. Methode: We simulated a hemithorax of 50 patients with a round tumour in a software phantom. Respiratory motion was induced as sinusoidal coordinate shift of the tumour. Contrast agent bolus passage through the tumour was modelled as density change over time based on two gamma functions (ground truth). Tracking was performed by object recognition based on resolved equivalence classes. Absolute deviation of mean tumour density over time from ground truth was assessed for statistical significance using T-tests with Bonferroni correction. Ergebnis: No patient showed significant deviation of tumour density time course from ground truth, T ranged from -2.4 to 2.7. Schlussfolgerung: The proposed method might save computing time in comparison to non-linear registration algorithms, however our approach only allows to assess functional properties of the tumour and not arbitrary voxels within an examination.

Assessment of contrast change over time in tumor perfusion is feasible using object tracking in a software model. Translation to animal/human imaging is future research.

Korrespondierender Autor: Tepper M

Klinik für Diagnostische Radiologie, UK-SH Campus Kiel, Kämpenstr. 6, 24106 Kiel

E-Mail: [e-mail]

motion compensation - motion correction - motion - tracking - object recognition