PhD Student / Postdoc Artificial intelligence-based risk and outcome prediction in calcified breast...
Every year, about 24,000 women are referred by the national breast cancer screening program for further testing. A third of these women are referred because of only calcifications seen on the mammogram. Roughly 75% of women referred for calcifications do not have breast cancer. In the remaining 25% of women, breast cancer or a precursor thereof, namely ductal carcinoma in situ (DCIS), might be present. While DCIS is a precursor of breast cancer, current evidence shows that many DCIS diagnosis would remain indolent and would not cause any danger for the woman's life. Despite this, there is currently no way to distinguish between women whose DCIS will or will not develop into invasive breast cancer. Unfortunately, this means that thousands of women undergo hospital visits, or even surgery, radiotherapy, or systemic treatment they do not need. Therefore, most women are referred unnecessarily in retrospect, leading to unrest, uncertainty, and possibly overdiagnosis and overtreatment.
We want to prevent that in the future with the help of artificial intelligence (AI). Using AI, we will address the problem at two levels: at the screening level where we use radiological data to determine which calcifications would not be a sign of breast cancer or DCIS and therefore, do not require a referral; and once referred for suspect DCIS, at the diagnostic level where we will build AI models that integrate the data from radiology, pathology, and molecular analyzes to determine which lesions will remain indolent and therefore do not require treatment. For this, we will need to develop deep learning techniques for detection, classification, and outcome prediction of calcifications and DCIS in multidisciplinary data. Additionally, we will design novel self-supervised contrastive learning techniques and combine these with model interpretability techniques to discover new knowledge about breast cancer. Ultimately, we want to make breast cancer screening even more targeted by optimizing the referral of women with calcifications and the subsequent decision-to-treat.
The job description
As a PhD-candidate, you will be responsible for developing and evaluating state-of-the-art deep learning techniques in multidisciplinary data. Finally, you will validate these algorithms in independent cases to ensure the devised AI-algorithms' applicability in clinical practice.
You are embedded in the ICAI AI for Oncology Lab, a collaboration between the Netherlands Cancer Institute and the Informatics Institute of the University of Amsterdam. The lab's mission is to develop innovations in artificial intelligence for the improvement of diagnosis and therapy of cancer. You will discuss results with our team, publish your work in artificial intelligence / medical journals, and present it at international conferences. As part of the project, you will be collaborating with experts in breast radiology of the Radboud University Medical Center.
The daily supervisor in this project is Dr. J. Teuwen, lab manager of the ICAI AI for Oncology Lab. Other team members are the scientific directors Prof. Dr. C. Sanchez (UvA) and Prof. Dr. J-J. Sonke. The project's clinicians are Prof. Dr. J. Wesseling, breast pathologist (NKI), and Dr. R. Mann, breast radiologist (NKI / Radboudumc).
Vul in waar je vergelijkbare vacatures zoekt en vergeet je e-mailadres niet!
We heten wel StudentJob, maar iedereen is even welkom. Ook als je al wat meer ervaring hebt. Meer weten? Check onze FAQ.