Caractérisation spectrale par tomodensitométrie en double énergie et analyse de texture radiogénomique avec intelligence artificielle pour améliorer l'évaluation et développer des biomarqueurs en imagerie pour les tumeurs de la tête et du cou


Reza Forghani

CIUSSS du Centre-Ouest-de-l'Île-de-Montréal


Domaine : cancer

Programme Chercheurs-boursiers cliniciens - Junior 1

Concours 2018-2019

Partenaire :

Fondation de l'Association des radiologistes du Québec

Medical imaging is at the heart of cancer diagnosis and therefore optimal cancer treatment and surveillance. There is increasing interest in using an advanced form of computed tomography or CAT scan, called dual energy CT, for improving the evaluation of head and neck cancer. There is in addition a wealth of information in medical images, beyond what the human eye can see. Using sophisticated computer algorithms for fine image analysis known as texture or radiomic analysis, combined with artificial intelligence, this information can be used to predict important tumor characteristics. These include prediction of how a tumor may respond to a given treatment, helping provide care that is tailored to the individual patient, or predict certain aspects of a tumor's molecular or genetic profile, without any invasive procedure. This exciting emerging field is referred to as radiomics or radiogenomics.

The first part of the study intends to demonstrate that dual energy CT is the best first test for certain types of head and neck cancer and can reduce the need for multiple scans. Then, I will use artificial intelligence assisted radiomic analysis and high-end tumor molecular profiling to identify unique imaging profiles that correspond to a tumor's genetic profile and better predict important characteristics such as whether a tumor has metastasized to lymph nodes, which could help reduce unnecessary extensive surgeries in many patients compared to what is currently done, in addition to providing other important information on tumor behavior and molecular characteristics. The cutting-edge studies proposed could provide a new horizon in the application and use of diagnostic images and artificial intelligence for head and neck cancer evaluation, providing either information not currently available or only available by invasive procedures such as biopsy or tumor resection and extensive molecular analyses not feasible in many settings.