Postdoc/PhD Positions in Computer Vision/Artificial Intelligence for Healthcare
The operating room is a high-tech environment in which surgical devices generate a lot of data about the underlying surgical activities. Our research group aims at making use of this large amount of multi-modal data coming from both cameras and surgical devices to develop an artificial intelligence system that can assist clinicians and staff in the surgical workflow. A core component of such an AI assistance system is the recognition of the surgical activities performed by clinicians and staff. In the context of the national AI Chair AI4ORSafety, we are looking for strong researchers to develop novel AI methods that can monitor critical safety steps during surgery.
PhD: Federated Learning for Scaling-up Surgical Activity Analysis
Existing recognition approaches are trained in a centralized manner: they require all the data to be stored on the same server. As institutions cannot easily share their data due to privacy concerns, current methods are often trained on limited datasets that are not representative of the surgical variability. Consequently, they do not generalize well to new clinical environments. A promising direction to address this issue is Federated Learning, in which a shared machine learning model is trained by aggregating locally-computed updates.
This PhD position will focus on designing and evaluating novel methods based on deep learning and federated learning to recognize surgical activities from endoscopic videos. One major application will be the automated assessment of critical safety steps during surgery. To carry out this work, the successful candidate will have access to a unique database of videos stemming from several clinical institutions and also have the possibility to interact with highly motivated clinical partners. By developing privacy-preserving computer vision and machine learning methods, he/she will facilitate the deployment of AI in hospitals.
PhD: Self-supervision for Anatomy and Activity Recognition in Endoscopic Videos
Surgical activities can easily be captured in videos, but annotating the videos is challenging and costly. In traditional computer vision, recent work has demonstrated the high potential of self-supervision methods for reducing the amount of annotations required, but little work has been done on clinical data.
This PhD fellowship will focus on designing and evaluating self-supervision methods to exploit large datasets of unlabelled endoscopic videos, with the goal to improve the segmentation and recognition of surgical anatomy and activity. One major clinical application will be the automated assessment of critical safety steps during the surgery. To carry out this work, the successful candidate will have access to a unique database of videos stemming from several clinical institutions and also have the possibility to interact with highly motivated clinicians. By providing news ways to scale up recognition with unlabelled data, he/she will facilitate the deployment of AI in hospitals.
Postdoc: Generalizable and Explainable AI for the Automated Analysis of Surgical Videos
This postdoctoral fellowship will focus on developing novel methods to analyse endoscopic videos, with the objective to use AI to perform the automated assessment of critical safety steps during surgery. The position aims at researching generalizable and explainable machine learning methods for anatomy segmentation/assessment and activity recognition in the presence of scarce annotations. The successful candidate will work on this project with a team of PhD students, engineers and clinicians, help supervise the research, and also contribute novel methods to scale up recognition on endoscopic videos.
To carry out this work, the successful candidate will have access to a unique database of videos stemming from several clinical institutions and also have the possibility to interact with a highly motivated team of clinicians.
Requirements
For PhD students:
- Master in Computer Science or equivalent
- C++/Python programming skills
- Strong knowledge in computer vision and machine learning
- Proficiency in English (oral and written)
- Experience with Deep Learning is a plus
For postdoctoral fellows:
- PhD in Computer Science or equivalent
- C++/Python programming skills
- Strong knowledge and experience in computer vision, machine learning and deep learning
- Proficiency in English (oral and written)
- Experience with clinical applications/endoscopic videos is a plus, but not mandatory
Environment
The position is located in Strasbourg, France. Strasbourg is a lively, green and cosmopolitan city situated in the heart of Europe and is also home to the European parliament. The successful candidate will be hosted within the AI team of the IHU institute at the University Hospital of Strasbourg. He/She will thereby have direct contact with clinicians, industrial partners and also have access to an exceptional international research environment offering state-of-the-art computing resources and unique clinical facilities. All positions are funded by the project AI4ORSafety, one of the 43 French national Chairs in Artificial Intelligence.
Benefits
Cutting-edge research in an interdisciplinary and leading international research environment
Ability to work at the forefront of a rapidly growing field at the intersection of computer science, artificial intelligence and medicine
Development of real-world AI-based solutions for the operating room
To Apply
Starting dates for all position are flexible. The postdoctoral position is renewable up to 3 years, offers a competitive salary commensurate with experience, and may lead to a permanent contract.
For PhD students:
Please send a long CV, motivation letter and academic transcripts to Nicolas Padoy.
For postdoctoral fellows:
Please send a long CV, motivation letter and list of publications to Nicolas Padoy.
Women are encouraged to apply.
Links
ICube laboratory
IHU MixSurg Institute
Laboratory of excellence CAMI
IRCAD Institute