In this work we address the need for the computation of quantitative global tumoral state indicators from oncological whole-body PET/computed tomography scans. The combination of such indicators with other oncological information such as tumor markers or biopsy results would prove useful in oncological decision-making scenarios.
From an ordering of 100 breast cancer patients on the basis of oncological state through visual analysis by a consensus of nuclear medicine specialists, a set of numerical indicators computed from image analysis of the PET/computed tomography scan is presented, which attempts to summarize a patient’s oncological state in a quantitative manner taking into consideration the total tumor volume, aggressiveness, and spread.
Results obtained by comparative analysis of the proposed indicators with respect to the experts’ evaluation show up to 87% Pearson’s correlation coefficient when providing expert-guided PET metabolic tumor volume segmentation and 64% correlation when using completely automatic image analysis techniques.
Global quantitative tumor information obtained by whole-body PET/CT image analysis can prove useful in clinical nuclear medicine settings and oncological decision-making scenarios. The completely automatic computation of such indicators would improve its impact as time efficiency and specialist independence would be achieved.
aFaculty of Medicine
bComputer Vision Center, Autonomous University of Barcelona
cNuclear Medicine Department, Hospital de Sant Pau
dFaculty of Mathematics, University of Barcelona, Barcelona, Spain
Correspondence to Frederic Sampedro, MSc, Faculty of Medicine, Autonomous University of Barcelona, Barcelona 08193, Spain Tel: +34 699 805 231; fax: +34 934 037 151; e-mail: firstname.lastname@example.org
Received October 21, 2013
Accepted November 25, 2013