Description du projet
Tester le dépistage de précision pour le cancer du sein
La mammographie numérique constitue la méthode de référence absolue pour détecter le cancer du sein, mais elle a ses limites: le sous-diagnostic et le surdiagnostic (faux positifs). La tomosynthèse numérique du sein (TNS) a été adoptée progressivement dans le dépistage du cancer. Le projet SIMULTANEOUS DBTMI combine la TNS et l’imagerie mécanique (IM) avec l’intelligence artificielle (IA) pour construire un système prototype avec une qualité d’image optimale. Le prototype sera évalué précliniquement grâce à des essais cliniques virtuels et des fantômes physiques, suivis d’un essai pilote coclinique (combinant des données simulées et cliniques). Le projet vise également à introduire des méthodes d’IA, sous la forme de réseaux d’apprentissage profond, pour découvrir des corrélations inconnues entre caractéristiques et améliorer les performances de la TNS-IM.
Objectif
This MSCA is designed to support Dr. Predrag Bakic in his professional development and reintegration into European research community, after he obtained Ph.D. and spent 12 years as a faculty in USA. The host institution, Lund University (LU), is one of the largest in Sweden and among the most prestigious in Europe. Dr. Bakic and his LU supervisors, Dr. Sophia Zackrisson and Dr. Anders Tingberg, share the research focus in breast imaging, with unique complementary expertise: Dr. Bakic in Virtual Clinical Trials (VCTs) based upon the simulation of breast anatomy and imaging systems, and LU team in Mechanical Imaging (MI) and conducting clinical trials of breast imaging.
Our action is motivated by a persistent challenge of underdiagnosis and false positives in breast cancer healthcare. The four most exciting innovations in breast cancer imaging that have recently emerged include: Digital Breast Tomosynthesis (DBT), MI, VCTs, and artificial intelligence (AI). In this application we will utilize extensive experience of LU and Dr. Bakic to interconnect these innovations efficiently and flexibly, enabling significant benefits. Within the two-year timeline, we will design and build a simultaneous DBT and MI (termed DBTMI) prototype system, and develop image processing and DBT reconstruction to maximize image quality. We will evaluate the prototype, first preclinically by VCTs and physical phantoms, followed by a pilot co-clinical trial (which combines clinical and simulated data). We will also explore introducing modern AI methods, in the form of Deep Learning Networks (DLN) to improve DBTMI performance. DLN has demonstrated ability to discover complex correlations in clinical images, leading to superior detection and classification of clinical findings. Combined complementary experience, carefully designed knowledge-exchange activities, and LU excellent institutional resources, guarantee the success of this application, and Dr. Bakic's successful reintegration.
Champ scientifique
Programme(s)
Régime de financement
MSCA-IF-EF-RI - RI – Reintegration panelCoordinateur
22100 Lund
Suède