Training EPFL students with world-class expertise in imaging.

Master Projects in Imaging

Interdisciplinary Master projects (semester projects and Master theses) in imaging are available for EPFL students.

To propose imaging-related Master projects, please contact [email protected]

Available Projects

3D super-resolution optical fluctuation imaging for molecular parameter estimation 

Master thesis, SV / STI 

The LBEN at EPFL is offering a master project related to super-resolution imaging. Several techniques have been developed in order to overcome the diffraction limit in fluorescence microscopy. They rely on exploiting a priori knowledge about the quantum mechanical properties of fluorophores.

For super-resolution optical fluctuation imaging (SOFI), higher order statistics (cumulants) of a time series of blinking fluorescence emitters are computed. SOFI analysis yields essentially background free images and is ideally suited for fast, 3D imaging using a multi-plane microscope. Balanced (b)SOFI deals with the nonlinear image contrast and enables up to fivefold improved spatial resolution in 2D. It allows extraction of molecular parameter maps such as the density and state lifetime of molecules. Molecular counting (or density estimation) based on SOFI is robust against imaging artifacts, which makes SOFI a unique tool in the super-resolution microscopy research field.

We recently used bSOFI to investigate clustering of membrane proteins in T cells with an aim to understand their response In this project, you will explore the extension of bSOFI parameter estimation to 3D imaging. Based on an existing 2D simulation framework, different multi-plane configurations and fluorophore properties can be simulated (brightness/ signal-to-noise, blinking parameters, photo-bleaching) and evaluated using SOFI processing.

You will perform super-resolution imaging of cells and/or DNA origami/protein calibration standards on our custom multi-plane microscopes. We aim to demonstrate multi-plane bSOFI parameter estimation for the first time. Depending on your skills and interests, the in silico or experimental aspects of the project can be emphasized.

Multi-plane high-order SOFI with a self-blinking dye : COS-7 cells immunostained microtubules with AbberiorFlip565. a) Pseudo widefield image, 2nd & 4th order SOFI; b) and c) close-up. <2kWcm-2 532nm. Scale bar 5μm.


  • A. Descloux et al. Combined Multi-Plane Phase Retrieval and Super-Resolution Optical Fluctuation Imaging for 4D Cell Microscopy. Nature Photonics (2018).
  • Geissbuehler, S. et al. Live-cell multiplane three-dimensional super-resolution optical fluctuation imaging. Nature Communications 5, 1–7 2014
  • Jungmann, R. et al. Multiplexed 3D cellular super-resolution imaging with DNA-PAINT and Exchange-PAINT. Nat Methods 11, 313– 318 (2014).

Type: Master thesis

Time: Fall 2019


Recommended requisites: Optics, microscopy, data analysis (Matlab), signal processing, cell culture, labelling of cells and preparation for super-resolution microscopy

Simulating realistic synthetic data sets for the development of a self-driving microscope

Master semester project or Master thesis, SB / STI 

Super-resolution microscopy offers a unique insight at nanoscale into cells processes and architecture. Although it provides a major gain in spatial resolution, super-resolution microscopy remains relatively inaccessible to biology laboratories due to the high level of expertise required to optimize image acquisition.

The goal of this project is to build an interactive software (“Flight Simulator”) for generating simulated super-resolution datasets. It should also enable on-the-fly modification of the acquisition parameters in an attempt to mimic, for the first time, what would actually happen on the microscope when an expert does the acquisition. Such an interface is a requirement for the development, validation and comparison of future automated, machine learning driven, routines dedicated to super-resolution microscopy.

Type: Master semester project or Master thesis

Time: Fall 2019


Recommended requisites: Basic coding skills, image processing knowledge

3D parameter free resolution estimation

Master semester project, SV / STI 

The laboratory of nanoscale biology (LBEN) at EPFL is offering a semester (master) project related to the resolution estimation of 3D imaging datasets.

As he was studying the ultimate limit of optical microscope in terms of resolution (i.e. the smallest resolvable distance between two ideal point sources), Ernst Abbe understood in 1983 that the fundamental limit was related to the wave nature of light and that it was impossible to resolve any structure smaller than half the wavelength of the light used to illuminate the sample (λ/2). However, Ernst Abbe could not foresee the astonishing improvement of optical sensor, light sources and most importantly fluorescent labels. About 20 years ago, it was realized that that by being able to switch molecules on or off (stochastically or via stimulated emission), it was possible to localize molecules with a resolution far beyond the Abbe limit.

At that moment a new research field, called super-resolution microscopy, emerged, with new methods such as STimulated Emission Depletion (STED), STochastic Optical Reconstruction Microscopy (STORM) and many more. One aspect of super-resolution microscopy is that the final resolution is now not only a related to the wavelength but is also a function of the photo-physical properties of the fluorescent label and laser illumination power. Therefore, there is a need for a universal tool that would be able to estimate the resolution of any image, super-resolution or not.

Recently we published a new method for image resolution estimation. The method is parameter free and exploits the phase information contained in the Fourier space of the image. Through the calculation of many partial phase correlation of the image with a filtered version, we are able to reliably extract the frequency support of any image, that is the highest frequency with significant contrast with respect to noise.

So far we demonstrated the ability of the method to estimate the global, local and sectorial resolution of various cell samples and we would like to extend the method in the third dimension. This extension presents several problems that the student will have to understand and overcome, such as limited axial sampling, plane to plane coregistration, increased computational complexity, etc…

The task of the student will be to implement the 3D version of the algorithm based on the current implementation (first in Matlab and then Java). The student will also have to investigate several options for processing optimization (code refactoring, GPU implementation, minimizing the number of correlations to be computed) due to the large amount of data to be processed inherent to volumetric imaging.

Illustration of the method where the formula for d(r) is used to compute several decorrelation curves which allows then to estimate the image resolution.

Type: Master semester project

Time: Fall 2019


Recommended requisites: Knowledge in signal and image processing, Matlab, Java and basics of optical imaging

Recommended courses: Image Processing, Signals and Systems I & II, linear algebra, analysis

Development of an image-analysis algorithm for crack detection in walls

Master semester project or Master thesis, ENAC / STI 

Damage to masonry buildings manifests itself in cracks in load-bearing walls. In large-scale physical tests we simulate the response of such walls and record the deformations using digital images. At present, we use the images only to compute the displacement field using a digital image correlation (DIC) technique.

With this project, we aim at developing algorithms for extracting automatically the crack geometry from the image. Combining crack geometry and displacement field, the crack kinematics can be computed. The information on crack geometry, crack connectivity and crack opening should be represented in a graph data structure. This information serves first to improve our understanding of the damage evolution in such walls.

In a second step, we would like to use it as ground truth for training machine learning algorithms for detecting cracks in real buildings. This requires that the images taken in the laboratory are augmented in such a way that they resemble cracks in real buildings with plastered surfaces.

This project has therefore three objectives: (i) to develop image-analysis algorithm for detecting cracks on laboratory images; (ii) to segment the crack map, geometry, connectivity in a graph data structure; (iii) to augment the images taken in the laboratory to resemble cracks in plastered walls of real buildings.

Type: Master semester project or Master thesis

Time: Fall 2019


Recommended requisites: Coding skills; interest in image processing

Deep learning assisted segmentation and mapping of DNA molecules

Master semester project, SV / STI 

The laboratory of nanoscale biology (LBEN) at EPFL is offering a semester (master) project related to the image processing of super-resolved stretched DNA.

DNA analysis methods have evolved tremendously over the past decade. One of the goal of such techniques is to be able to recognize the species of origin. As an alternative to DNA sequencing (i.e. reading the whole DNA sequence), we have developed in our lab a way to map the DNA to its corresponding species while avoiding complicated PCR reactions and DNA sequencing.

The method is based on sequence specific labelling of DNA and subsequent stretching on a glass surface. The stretched DNA is then imaged with a super-resolution microscope resulting in a sort of bar-code image (Figure). The intensity profile of each DNA molecules is extracted and matched against a database of species. [1,2]

In order to study the entire microbiome, we need to analyse thousands of images, extract all the individual DNA molecules and match them to their sequences. This is too much for manual selection, a method is needed to automatically detect the DNA strands and extract their intensity profile.

The task of the student will be to optimize a new approach to DNA segmentation based on machine learning and work on the automatisation of the full pipeline, from raw images to meaningful information. The second task of the student (Master project) will be to train another neural network to classify the segmented DNA. We will provide the student with experimental images, supervision and expertise to develop the algorithm (Python and Matlab). The student will be able to work in a highly interdisciplinary group with backgrounds ranging from polymer physics, image analysis, microscopy to molecular biology.

Typical microscopy image of labeled DNA. Inset: Zoom on DNA molecules with manual segmentation


  1. Deen, J.; Vranken, C.; Leen, V.; Neely, R. K.; Janssen, K. P. F.; Hofkens, J. Angew. Chem. Int. Ed. 2016, doi:10.1002/anie.201608625
  2. Neely, R. K.; Deen, J.; Hofkens, J. Biopolymers 2011, 95, (5), 298-311.

Type: Master semester project

Time: Fall 2019


Recommended requisites: Knowledge in signal and image processing, machine learning and basics of optical imaging

Multimode optical fiber imaging with a deep learning network 

Master semester project, STI 

Multi-mode fibers (MMFs) are gaining widespread interest for optical imaging at the nano/micro-scale levels due to the high space-bandwidth product. However, the full data communication bandwidth potential of MMFs cannot yet be easily exploited due to inter channel data scrambling inside the fiber. Machine learning has been recently applied to image transmission through MMFs. 

The goal of the project is to recover the scrambled images inside the fiber with high fidelities using the state-of-the-art deep neural networks architecture. It has been shown that not only does the neural network retrieve back the original unscrambled images with high fidelities; it is also able to generalize to images from other categories.

Type: Master semester project

Time: Fall 2019


Recommended requisites: Coding skills in Python, machine learning familiarity, Basic Optics (optional)

Cracking human brain activity measured by functional magnetic resonance imaging

Master semester project or Master thesis, SV / STI 

The MIPLAB has developed a novel framework to analyze human brain activity measured by functional magnetic resonance imaging (fMRI), which allows to non-invasively acquire whole-brain snapshots about every second. The hemodynamic response, through neurovascular coupling and changes in blood oxygenation, serves as a proxy for neural activity. The new framework includes advanced signal processing steps such as regularized deconvolution of the hemodynamic response, which results in a deconvolved (“activity-inducing”) signal and its derivative (“innovation”) signal. Clustering of the innovations then determines consistent spatial patterns of transient activity for which the term “innovation-driven co-activation patterns”—iCAPs is coined. The approach is among the most promising to conduct time-resolved analyses of fMRI data and characterize brain dynamics in terms of building blocks that can be spatially and temporally overlapping.

Within this context, we have several student projects available to tailor and apply the framework to different datasets and applications. This includes large imaging initiatives such as the Human Connectome Project where brain fMRI as well as demographic and behavioral data is available from a large number of healthy volunteers (>1200). We also have several ongoing collaborations with clinical groups that involve patient cohorts. In another project, we are looking into spinal cord fMRI, still largely underexplored, for which the same methodology can be applied. One of the key questions is to relate the imaging-derived features to behavioral measures or clinical scores. We are also interested in linking analysis of functional data with anatomical features of the brain; i.e., how is brain activity “using” the anatomical backbone on which it is expressed.

Type: Master semester project or Master thesis

Time: Fall 2019


 Deep learning for angle estimation in cryo-EM

Master thesis, STI 

Single-particle cryo-electron microscopy (cryo-EM) is a Nobel-prized technology that aims to characterize the 3D structure of proteins at the atomic level. The electron microscope first images numerous (~100k) replicates of a protein, positioned at various orientations. Algorithms then reconstruct a high-resolution 3D structure from the acquired images.

The main challenge in cryo-EM reconstruction, compared to traditional tomographic set-ups, is that the angles at which the images were taken are unknown. Another challenge is that the images are extremely noisy and blurred. The sheer amount of images per protein (~100k), as well as the number of imaged proteins (~4k), should however enable a data-driven approach to overcome those challenges.

The goal of this project is to design a neural network to estimate the angular relation between images of a protein. The developed neural network will be trained and tested on simulated and real data.

Type: Master thesis

Time: Fall 2019


Recommended requisites: Experience with Python programming. Experience with (Deep) Machine Learning (with any framework) is desirable. No experience in biology required. Experience in imaging is a plus.

Optical projection tomography to elucidate neurodegeneration

Master semester project, SV / STI 

The laboratory of nanoscale biology (LBEN) at EPFL is offering a semester (master) project related to the study of neurodegeneration in a mouse model of Alzheimer’s disease using Optical Projection Tomography (OPT).

Our research is set in the context of studying Alzheimer’s disease and observing Amyloid-beta plaques and immune cells using the 3D whole-tissue imaging modality known as OPT [1].

The project’s aim is to apply OPT imaging to identifying amyloidosis levels and immune response in the brain and the gastro-intestinal tract of the 5xFAD mouse model.

The task of the student will entail collaborating with microscopists and biologists to optimize the resolution optics of our existing in-house OPT setup while in parallel testing various antibody-mediated staining of tissue sections and whole tissues. All in vivo work will be performed by appropriately licensed staff members, and thus will not be part of this student project. The student will be able to work in a highly interdisciplinary group with backgrounds ranging from polymer physics, image analysis, microscopy to molecular biology.

Example of OPT acquisition of the 5xFAD brain (A and B) and gut (C). in A, the amyloid plaques have not been segmented, whereas in B they have. 


Type: Master semester project

Time: Fall 2019


Recommended requisites: Basics of optical imaging, basic engineering techniques (e.g. soldering), general biology knowledge