The impact of AI on society is undeniable, especially in machine vision and video understanding tasks. However, applications of AI in biomedicine are hampered by the need for large amounts of training data which can be expensive and difficult to collect (for example patient biopsies, imaging protocols which damage samples or large variability between samples); by high inference latencies, making real-time interventions unfeasible; and by large memory and computation requirements, hindering the development of affordable benchtop devices for AI-assisted diagnostics. Specifically, efficient, automated inference of the dynamical states of biological tissue is an important and underexplored problem, with potential applications in creating digital twins for physiological and pharmaceutical research as well as for providing real-time, AI-driven interventions.

In this PhD position, we explore the intersection between artificial and biological neural networks. We investigate automated regression of the dynamic state variables of biological neurons using state-of-the-art computer vision methods, which have the potential to directly map a wide array of neural tissue images onto biologically realistic neural network models. However, performing this mapping from image space, where information is contained in the relative activity patterns of the neurons, and not in their individual pixel location, requires developing new computer vision models which are rotation, scale and permutation invariant and can ignore absolute spatial information. Such an approach can both reduce the computational cost of the traditional brute-force approaches—which employ costly grid-search methods to fit parameters of a neural tissue model—and reduce the high data requirements of training the machine learning model—by constraining the output space via the use of physiologically plausible mathematical neuron models.  

This project will be jointly supervised by Nergis Tomen from the Computer Vision Lab in the Department of Intelligent Systems (EEMCS faculty), and Daan Brinks from the Brinks lab in Neurophysics in the Department of Imaging Physics (AS faculty). During your PhD, you will develop efficient computer vision algorithms to analyze neural tissue images and videos in order to map them onto neuroscientific models. The associated biomedical challenge is to computationally reconstruct biological neural networks with minimal prior information. You will work closely with other PhD students in the Computer Vision Lab, who study efficient machine learning, and students in the Neurophysics Lab who study biological neural networks and neural simulations. The PhD position is primarily focused on computational development, but could include biomedical experiments or simulations if the student is interested.

Do you have an interest in making AI more efficient? Would you like to use AI to have a positive impact on healthcare? Do you like software design and optimization? Do you have an interest in applying your skills in a project with many engineering challenges? Would you like to broaden your skills in an interdisciplinary environment and interact with computer scientists, machine learning experts, physicists and neuroscientists? Then apply!

This one of four PhD postions within the BIOLAB, for a complete overview check: https://www.tudelft.nl/ai/biolab

In the Biomedical Intervention Optimization lab (BIOlab), experts in computer vision (Tömen), reinforcement learning (Böhmer), neural architecture (Brinks), deep learning and computational physics (Perkó), and biomedical imaging (Gruβmayer, Carroll) join forces to create high-efficiency, real-time, AI-driven feedback and control in biomedical applications.

To qualify for this position you must have:

• completed a relevant MSc degree in artificial intelligence, computer science, computational neuroscience or an applied sciences field related to PhD research
• demonstrated competences in one or more of these categories: reinforcement learning, machine learning, computational neuroscience, or relevant appled sciences
• demonstrated experience in programming in at least one of the following programming languages: Python, C/C++
• a proven record and interest in further developing your modelling, programming and analytical skills
• strong affinity and enthusiasm for state-of-the-art reinforcement learning and AI approaches
• excellent critical and analytical thinking skills
• An affinity with teaching and guiding students
• Proficiency in expressing yourself verbally and in writing in English
• The ability to work in a team and take initiative
• in depth knowledge of mathematical and computational models of spiking neurons and spiking neural networks

In addition:

• strong knowledge of Python and knowledge of or motivation to work with popular deep learning libraries (Tensorflow, PyTorch) is a plus
• background knowledge in computational sciences (computational physics, biology, chemistry) or nonlinear dynamics is a plus
• background or interest in neuroscience is a plus

You will receive a 5-year contract and will be deployed for AI-related education for the usual teaching effort for PhD candidates in the faculty plus an additional 20%. The extra year compared to the usual 4-year contract accommodates the 20% additional AI, Data and Digitalisation education related activities. All team members have many opportunities for self-development. You will be a member of the thriving TU Delft AI Lab community that fosters cross-fertilization between talents with different expertise and disciplines.

Delft University of Technology

Delft University of Technology is built on strong foundations. As creators of the world-famous Dutch waterworks and pioneers in biotech, TU Delft is a top international university combining science, engineering and design. It delivers world class results in education, research and innovation to address challenges in the areas of energy, climate, mobility, health and digital society. For generations, our engineers have proven to be entrepreneurial problem-solvers, both in business and in a social context. At TU Delft we embrace diversity and aim to be as inclusive as possible (see our Code of Conduct). Together, we imagine, invent and create solutions using technology to have a positive impact on a global scale.

Challenge. Change. Impact! 

Biomedical Intervention Optimization lab (BIOlab)

This position is connected to the Biomedical Intervention Optimization lab (BIOlab). BIOLab is a new TU Delft Artificial Intelligence Lab. Artificial Intelligence, Data and Digitalisation are becoming increasingly important when looking for answers to major scientific and societal challenges. In a TU Delft AI Lab, experts in ‘the fundamentals of AI technology’ along with experts in ‘AI challenges’ run a shared lab.

As a PhD, you will work with at least two academic members of staff and three other PhD candidates. In total TU Delft will establish 24 TU Delft AI Labs, where 48 Tenure Trackers and 96 PhD candidates will have the opportunity to push the boundaries of science using AI. Each team is driven by research questions which arise from scientific and societal challenges, and contribute to the development and execution of domain specific education.

Goal of the Biomedical Intervention Optimization lab (BIOlab)

Modern machine learning algorithms have achieved unprecedented accuracy in image and video understanding tasks by purely learning from data. These powerful abilities come at the price of enormous amounts of training data, memory and computational requirements. However, those resources are rarely available to real-time feedback systems in medical intervention and biomedical research.

The lab will focus on improving the efficiency of machine learning algorithms by designing novel artificial neural network architectures, developing new reinforcement learning and generative algorithms, and incorporating biologically-inspired neural network models. These newly developed concepts and algorithms will be applied to a wide range of problems in biomedical applications, such as optimizing tumor irradiation protocols with missing information, and in smart (super-resolution) microscopy to limit irradiation damage to delicate living samples.

The Faculty of Electrical Engineering, Mathematics and Computer Science (EEMCS) brings together three disciplines - electrical engineering, mathematics and computer science. Combined, they reinforce each other and are the driving force behind the technology we use in our daily lives. Technology such as the electricity grid, which our faculty is helping to make future-proof. We are also working on a world in which humans and computers reinforce each other. We are mapping out disease processes using single cell data, and using mathematics to simulate gigantic ash plumes after a volcanic eruption. There is plenty of room here for ground-breaking research. We educate innovative engineers and have excellent labs and facilities that underline our strong international position. In total, more than 1,100 employees and 4,000 students work and study in this innovative environment.

Click here to go to the website of the Faculty of Electrical Engineering, Mathematics and Computer Science.