Meeting Banner
Abstract #2304

Improvement of Colorectal Liver Metastases Detection Sensitivity and Specificity by Hemodynamic Response Imaging Combined with a Machine Learning Approach

Yifat Edrei1,2, Moti Freiman3, Eitan Gross4, Nathalie Corchia1, Leo Joskowicz3, Rinat Abramovitch1,2

1The Goldyne Savad Inst. for Gene Therapy, Hadassah Hebrew University Medical Center, Jerusalem, Israel; 2MRI/MRS lab HBRC, Hadassah Hebrew University Medical Center, Jerusalem, Israel; 3School of Engineering and Computer Science, The Hebrew University, Jerusalem, Israel; 4Pediatric Surgery, Hadassah Hebrew University Medical Center, Jerusalem, Israel

Colorectal liver metastases (CRLM) are a major cause of death of colorectal-arcinoma patients. Recently, we demonstrated the feasibility of Hemodynamic Response Imaging (HRI), an fMRI method combined with hypercapnia and hyperoxia, for monitoring liver perfusion. In CRLM animal model, we compared the HRI sensitivity to the standard DCE-MRI perfusion-imaging. Furthermore, we developed software, based on a machine-learning approach, for the interactive classification of suspected-CRLM. We concluded that HRI has a higher sensitivity to subtle changes in liver blood-flow induced by CRLM than DCE-MRI, and that the machine-learning approach can provide a useful assistance to early and accurate detection of CRLM.