Thesis – Soft Matter Lab

Jesus Pineda defended his PhD thesis on November 11th, 2025. Congrats!

Jesus Pineda defended his PhD thesis on November 11th, 2025. Congrats!
The defense took place in SB-H7 lecture hall, Institutionen för fysik, Johanneberg Campus, Göteborg, at 9:00.

Title: Inductive Biases for Efficient Deep Learning in Microscopy

Abstract: Deep learning has become an indispensable tool for the analysis of microscopy data, yet its integration into routine research remains uneven. Several factors contribute to this gap, including the limited availability of well-annotated datasets and the high computational demands of modern architectures. Microscopy introduces further challenges, as it spans diverse modalities and scales, from proteins to tissues, producing heterogeneous data that defy standardization. Generating reliable annotations also requires expertise and time, while unequal access to high-performance computing further widens the divide between well-resourced institutions and smaller laboratories.

This dissertation argues that the prevailing paradigm of scaling models with ever-larger datasets and computational resources yields diminishing returns for microscopy. Instead, it explores the role of inductive biases as a foundation for building models that are more data-efficient, computationally accessible, and scientifically meaningful. Inductive biases are structural assumptions embedded in model design that guide learning toward patterns aligned with the underlying problem. The first part of this work examines their central role in the advancement of modern deep learning and the diverse ways they shape model behavior.

This potential is demonstrated through three case studies. First, MAGIK employs graph neural networks to analyze biological dynamics in time-lapse microscopy, uncovering local and global properties with high precision, even when trained on limited data. Next, MIRO leverages recurrent graph neural networks to process single-molecule localization datasets, improving the efficiency and reliability of clustering for variable biological structures and scales while retaining strong generalization with minimal supervision. Finally, GAUDI introduces a representation learning framework for characterizing biological systems, providing a physically meaningful representation space for interpretable and transferable analysis.

The findings presented here demonstrate that the integration of inductive biases provides a cohesive strategy to extend the reach of deep learning in the life sciences, enhancing accessibility and ensuring scientific utility under resource constraints.

Thesis: https://gupea.ub.gu.se/items/672c7946-51d6-4773-ad8c-35a3eed41499

Supervisor: Giovanni Volpe
Co-Supervisor: Carlo Manzo
Examiner: Raimund Feifel
Opponent: Anna Kreshuk
Committee: Juliette Griffié, Daniel sage, Daniel Persson
Alternate board member: Jonas Enger

Martin Selin defended his PhD thesis on October 8th, 2025. Congrats!

Cover of the PhD thesis. *(Image by M. Selin.)*

Martin Selin defended his PhD thesis on October 8th, 2025. Congrats!
The defense took place in PJ, Institutionen för fysik, Origovägen 6b, Göteborg, at 13:00.

Title: Advanced and Autonomous Applications of Optical Tweezers

Abstract: Optical tweezers have become a central tool, using lasers to manipulate and probe objects with exceptional precision enabling single-molecule, single-cell, and single-particle studies. However, this precision comes at the cost of throughput.

By developing a fully autonomous system we can adress this limitation of optical tweezers. The system is capable of perfoming multiple different experiments independently and of operating for over 10 hours continously. Using the same system we also investigate particle adsorption into liquid-liquid interfaces revealing never before seen dynamics.

These developments help optical tweezers by bridging the gap between single-molecule, cell or particle studies and ensemble measurements, enabling the application of deep learning for advanced modeling and unlocking the potential of optical tweezers for large, data-driven studies.

Thesis: https://gupea.ub.gu.se/handle/2077/87446?show=full

Supervisor: Giovanni Volpe
Examiner: Raimund Feifel
Opponent: Borja Ibarra
Committee: Dag Hanstorp, Timo Betz, Kristine Berg-Sørensen
Alternate board member: Paolo Vinai

Laura Natali defended her PhD thesis on March 28th, 2025. Congrats!

Cover of the PhD thesis. *(Image by L. Natali.)*

Laura Natali defended her PhD thesis on March 28th, 2025. Congrats!
The defense took place in PJ, Institutionen för fysik, Origovägen 6b, Göteborg, at 10:00.

Title: Neural Networks for Complex Systems: From Epidemic Modeling to Swarm Robotics

Abstract: Deep learning models, inspired by the structure of the brain, were first developed in the last century. These models are trained to recognize patterns in large amounts of data. Recently, deep learning has made a big impact, both in research and in everyday applications, like healthcare, image recognition, and language translation.

However, despite their advancements, these models still fall short of the abilities found in biological brains, which are adaptable, energy-efficient, and have evolved over millions of years. In contrast, artificial models are specialized and struggle to adapt to new information.

To help address this gap, we have developed a robotic experiment that combines the programmability of artificial neural networks with some of the physical constraints seen in biological systems.

Thesis: https://hdl.handle.net/2077/84676

Supervisor: Giovanni Volpe
Examiner: Bernhard Mehlig
Opponent: Hamid Kellay
Committee: Maria Guix Noguera, Juliane Simmchen, Michael Felsberg
Alternate board member: Paolo Vinai

Harshith Bachimanchi defended his PhD thesis on March 26, 2025. Congrats!

From left: Anupam Sengupta (opponent), Harshith Bachimanchi, Giovanni Volpe (supervisor). *(Photo by A. Argun.)*

Harshith Bachimanchi defended his PhD thesis on March 26th, 2025. Congrats!
The defense took place in PJ, Institutionen för fysik, Origovägen 6b, Göteborg, at 13:00.

Title: Deep Learning Enhanced Optical Methods for Single-Plankton Studies

Abstract: Among Earth’s earliest life forms, cyanobacteria reshaped the planet by oxygenating the atmosphere during the Great Oxidation Event 2.4 billion years ago. This process, which led to ozone formation and UV protection, paved the way for more complex photosynthetic organisms—phytoplankton, the eukaryotic descendants of cyanobacteria. Today, phytoplankton drive the global carbon cycle, producing 50–80% of Earth’s oxygen and fueling the marine food web. Microzooplankton consume nearly two-thirds of the organic carbon generated, yet despite their ecological significance, tracking biomass flow at the single-cell level remains a major challenge.

This thesis presents novel methodologies that integrate advanced optical techniques, deep learning, and simulated datasets to analyze microplankton dynamics with unprecedented resolution.

A key contribution is a deep-learning-enhanced holographic microscopy approach that quantifies microplankton biomass at the single-cell level while simultaneously capturing their three-dimensional swimming behavior. This method overcomes computational bottlenecks in traditional holography, enabling high-throughput analysis across diverse species and size ranges. Expanding on this, I demonstrate its application in mixed-species experiments to examine feeding interactions between phytoplankton and microzooplankton, capturing biomass transfer and behavioral shifts during predation.

Beyond direct imaging, this thesis leverages synthetic data to advance microscopy-based research. Neural networks trained on simulated microscopy datasets are used to detect, segment, and classify plankton species while reconstructing motion dynamics. To showcase the versatility of this approach, I present its application in a non-biological setting—detecting bubble-propelled artificial micromotors within complex experimental backgrounds. In addition to object detection, these methods also enable motion characterization of microscopic entities. To demonstrate this, I introduce synthetic microscopy videos that model microscopic organisms undergoing various anomalous diffusion behaviors. This framework is then used to develop a method that extracts motion characteristics without explicit trajectory linking, broadening its applications beyond plankton ecology.

Finally, I investigate how zooplankton—key players in the marine food web—respond to ocean wave-induced light patterns using an LED matrix. The results suggest that zooplankton use steady light sources, such as celestial objects, to ascend more rapidly during favorable low-turbulent conditions, offering new insights into their migratory strategies. Collectively, this thesis bridges marine ecology, microscopy, artificial intelligence, and biophysics to provide new tools for exploring the unseen dynamics that shape our planet.

Thesis: https://hdl.handle.net/2077/84857

Supervisor: Giovanni Volpe
Examiner: Raimund Feifel
Opponent: Anupam Sengupta
Committee: Elisa Berdalet, Maria Guix Noguera, Josefin Titelman
Alternate board member: Paolo Vinai

Gan Wang defended his PhD thesis on 20 January 2025. Congrats!

Gan Wang defended his PhD thesis on 20 January 2025. Congrats!
The defense took place in PJ, Institutionen för fysik, Origovägen 6b, Göteborg, at 10:00.

Title: Microfabrication technique applications: from passive particle manipulation to active microswimmers, micromachines, and fluidic control

Abstract: Overcoming Brownian motion at the micro- and nanoscale to achieve precise control of objects is crucial for fields such as materials science and biology. Significant progress has been made in trapping and manipulating micro- and nanoscale objects, either by generating gradients through external physical fields or by engineering systems that can harvest energy from their environment for autonomous motion. These techniques rely on the precise application of forces, such as optical and electromagnetic forces, and have found extensive applications across various scientific disciplines. Recent advances in micro- and nanofabrication technologies have greatly enhanced the generation and regulation of these forces, offering new possibilities for manipulating micro- and nanoscale objects.

This thesis applies traditional micro- and nanofabrication techniques, typically used in semiconductor manufacturing, to construct micro- and nanostructures for manipulating forces, primarily critical Casimir forces and optical forces, to achieve precise control over microscale object movement.

I first show the fabrication of periodic micropatterns on a substrate, followed by chemical functionalization to impart hydrophilic and hydrophobic properties. Near the critical temperature of a binary liquid, attractive and repulsive critical Casimir forces are generated between the micropatterns and microparticles. These forces allow the stable trapping of the microparticles on the substrate and the manipulation of their configuration and movement.
Then, my research transitions from passive control to active motion by fabricating metasurfaces capable of modulating optical fields and embedding them within micro-particles (microswimmers). This enables light-momentum exchange under planar laser illumination, resulting in autonomous movement of the microswimmers. By varying the metasurface design as well as the intensity and polarization of the light, complex behaviors can emerge within these microswimmers. Subsequently, My research focused on using these microfabrication techniques to build micromotors integrated on a chip surface. These micromotors couple with other objects through gear structures, creating miniature machines that can execute functional tasks. Finally, by altering the configuration of these machines and the distances between them, I acheived precise, multifunctional control over fluid dynamics, facilitating the transport of micro- and nanoscale objects.

Insights gained from this research suggest innovative manufacturing approaches for scalable manipulation of particles, more intelligent microrobots, and powerful miniaturized on-chip machines, with applications across various fields.

Thesis: https://hdl.handle.net/2077/84048

Supervisor: Giovanni Volpe
Examiner: Dag Hanstorp
Opponent: Peer Fischer
Committee: Heiner Linke, Anna Maciolek, Hao Zeng
Alternate board member: Francesco Ferranti

Benjamin Midtvedt defended his PhD thesis on 9 January 2025. Congrats!

Benjamin Midtvedt, PhD defense. *(Photo by H. P. Thanabalan.)*

Benjamin Midtvedt defended his PhD thesis on 9 January 2025. Congrats!
The defense will take place in PJ, Institutionen för fysik, Origovägen 6b, Göteborg, at 13:00.

Title: Annotation-free deep learning for quantitative microscopy

Abstract: Quantitative microscopy is an essential tool for studying and understanding microscopic structures. However, analyzing the large and complex datasets generated by modern microscopes presents significant challenges. Manual analysis is time-intensive and subjective, rendering it impractical for large datasets. While automated algorithms offer faster and more consistent results, they often require careful parameter tuning to achieve acceptable performance, and struggle to interpret the more complex data produced by modern microscopes. As such, there is a pressing need to develop new, scalable analysis methods for quantitative microscopy. In recent years, deep learning has transformed the field of computer vision, achieving superhuman performance in tasks ranging from image classification to object detection. However, this success depends on large, annotated datasets, which are often unavailable in microscopy. As such, to successfully and efficiently apply deep learning to microscopy, new strategies that bypass the dependency on extensive annotations are required. In this dissertation, I aim to lower the barrier for applying deep learning in microscopy by developing methods that do not rely on manual annotations and by providing resources to assist researchers in using deep learning to analyze their own microscopy data. First, I present two cases where training annotations are generated through alternative means that bypass the need for human effort. Second, I introduce a deep learning method that leverages symmetries in both the data and the task structure to train a statistically optimal model for object detection without any annotations. Third, I propose a method based on contrastive learning to estimate nanoparticle sizes in diffraction-limited microscopy images, without requiring annotations or prior knowledge of the optical system. Finally, I deliver a suite of resources that empower researchers in applying deep learning to microscopy. Through these developments, I aim to demonstrate that deep learning is not merely a “black box” tool. Instead, effective deep learning models should be designed with careful consideration of the data, assumptions, task structure, and model architecture, encoding as much prior knowledge as possible. By structuring these interactions with care, we can develop models that are more efficient, interpretable, and generalizable, enabling them to tackle a wider range of microscopy tasks.

Thesis: https://hdl.handle.net/2077/84178

Supervisor: Giovanni Volpe
Examiner: Dag Hanstorp
Opponent: Ivo Sbalzarini
Committee: Susan Cox, Maria Arrate Munoz Barrutia, Ignacio Arganda-Carreras
Alternate board member: Måns Henningson

Ivo Sbalzarini (left) and Benjamin Midtvedt (right). *(Photo by H. P. Thanabalan.)*

Benjamin Midtvedt (left), Giovanni Volpe (right), announcement. *(Photo by H. P. Thanabalan.)*

From left to right: Ignacio Arganda, Arrate Muñoz Barrutia, Susan Cox, Benjamin Midtvedt, Giovanni Volpe, Ivo Sbalzarini. *(Photo by H. P. Thanabalan.)*

Jesús Domínguez defended his PhD thesis on 6 September 2024. Congrats!

The three platforms developed to observe and characterise bacterial collective behaviour in different conditions. *(Image by J. Dominguez.)*

Jesús Manuel Antúnez Domínguez defended his PhD thesis on 6 September 2024. Congrats!
The defense took place in PJ, Institutionen för fysik, Origovägen 6b, Göteborg.

Title: Microscopic approaches for bacterial collective behaviour studies.

Abstract: Bacteria significantly impact our lives, from their beneficial role as probiotics to their involvement in infection environments. Their widespread presence is largely due to their ability to adapt to diverse conditions through collective behavior, which enables the development of complex strategies from the contributions of simple individual entities. However the understanding of these systems is limited by the reach of current study techniques. This work presents the development of three platforms designed to perform microscopic studies and characterise bacterial collective behaviors in situ, profiting the advantages of microfluidics over traditional culture techniques.

The first platform integrates bacterial culture on solid agar directly on the microscope stage, allowing for extended observation periods of up to a week. The agar is housed within an elastomer structure sealed with glass, ensuring environmental isolation while maintaining optical accessibility. This platform was used to document the complex social strategies of Myxococcus xanthus, including motility mechanisms, predation organisation, and fruiting body formation.

The second platform is an automated testing system for quantifying bacterial viability under various conditions. Using microfluidic technology, this platform streamlines and parallelise the process. It adapts the Ames genotoxicity test to a miniaturized version, using microscopy imaging as the readout. This approach reduces experimental turnaround time and minimizes the handling of hazardous substances.

The third platform is a microfluidic system designed for the microscopy observation of bacteria within stabilised droplets. This approach enhances throughput and allows for the production of various types of droplets on the same chip. Bacillus subtilis bacteria were encapsulated in these droplets, and their entire biofilm formation life cycle was observed in detail. Parallel to this, custom software was developed specifically for analysing microscopy images to automatically quantify biofilm formation.

Each of these platforms provides a unique perspectives in the study of bacterial collective behavior to offer a comprehensive toolkit for researchers. complementing one another. This work will equip researchers with the tools to address the mysteries of bacterial collective behavior and opens up new possibilities for application and investigation.

Thesis: https://hdl.handle.net/2077/81543

Supervisor: Caroline Beck Adiels
Examiner: Raimund Feifel
Opponent: Jana Jass
Committee: Edith Hammer, Per Augustsson, Johan Bengtsson-Palme
Alternate board member: Mattias Marklund

Jesús presenting in PJ. *(Photo by A. Ciarlo.)*

Alex Lech defended his Master Thesis on 16 May 2024. Congrats!

Rendering of the absorption of optical power by iron-oxide nanocores in a super-paramagnetic particle. *(Image by A. Lech.)*

Alex Lech defended his Master Thesis on 16 May 2024 at 15:45. Congrats!

Title: Simulation of light-absorbing microparticles in an optical landscape

Abstract:
Simulating the dynamics of active particles play a key role in understanding the many behaviours active matter can exhibit. Experimental studies are more costly than simulations in this regard, as there is much work that needs to be performed with setups and observation time. Computer simulations are a powerful and cost-effective alternative to experiments. One topic of study within active matter is light-absorbing microparticles which are commonly made of silica with a light-absorbing metallic compound such as iron oxide or gold. One such microparticle is the Janus particle, a silica particle with a hemispherical coating of gold as the absorbing compound. When illuminated with a laser, the coating absorbs the light and heats up rapidly, generating a temperature gradient which allows the Janus particle to exhibit self-propulsion and clustering with other Janus particles due to thermophoresis and Brownian motion.

In this thesis, I introduce a simulation model which simulates light-absorbing microparticles with a desired distribution of iron oxide in an optical landscape. In particular, I will consider the case of an optical landscape characterized by a periodical sinusoidal intensity profile of a given spatial periodicity.

The results show that for a hemispherical distribution (Janus particle) there is self-propulsion originating at the side of the cap, with super-diffusive characteristics. When the laser periodicity is similar to the particle radius, it becomes confined between two high intensity peaks. A particle with uniform distribution diffuses with Brownian motion, with no self-propulsion. Clustering behaviour arises when multiple particles are in close proximity to each other, as observed in experiments.

The agreement with experimental results opens up for the opportunity to simulate other light-absorbing particles with different distributions of absorbing compounds.

Supervisor: Agnese Callegari
Examiner: Giovanni Volpe
Opponent: John Klint, Niphredil Klint

Place: von Bahr
Time: 16 May, 2023, 15:45

Emiliano Gómez will defend his PhD thesis on 22 May 2024

Emiliano Gómez will defend his PhD thesis on the 22th of May at 10:30. The defense will take place in KA (Chemistry Department, Johanneberg Campus)

Title: Development and Application of a software to analyse networks with multilayer graph theory and deep learning

Abstract:
Network theory gives us the tools necessary to produce a model of our brain, how the brain is wired will give us a new level of insight into its functionality. The brain network, the connectome, is formed by structural links such as synapses or fiber pathways in the brain. This connectome might also be interpreted from a statistical relationship between the flow of information, or activation correlation between brain regions. Mapping these networks can be achieved by using neuroimaging, which allows obtaining information on the brain in vivo. Different neuroimaging modalities will capture different properties of the brain. Statistical analysis is necessary for extracting meaningful insights regarding the network patterns obtained from neuroimages. For this, huge data banks are a byproduct of the need for enough data to be able to tackle medical and biological questions.

In this work, we present a software “Brain Analysis using Graph Theory 2” (BRAPH 2) (Paper I), which addresses the need for a toolbox designed for both complex graph theory and deep learning analyses of different imaging modalities. With BRAPH 2, we offer the neuroimaging community a tool that is open-source, flexible, and intuitive. BRAPH 2, at its core, comes with multi-graph capabilities. For Paper II, we employed the power of multiplex and multigraph capabilities of BRAPH 2 to analyze sex differences in brain connectivity for an aging healthy population. Finally, for Paper III, BRAPH 2 has been adapted to two new graph measures (global memory capacity, and nodal memory capacity), which obtain a prediction of memory capacity using Reservoir Computing and relate this new measure to biological and cognitive characteristics of the cohort.

Supervisor: Giovanni Volpe
Examiner: Raimund Feifel
Opponent: Saikat Chatterjee, KTH, Stockholm
Committee: Marija Cvijovic, Alireza Salami, Wojciech Chachólski
Alternate board member: Mats Granath

Laura Pérez García defended her PhD thesis on 12 October 2023. Congrats!

A dielectric particle under the influence of the gradient and scattering force. *(Image by L. Pérez García.)*

Laura Pérez García defended her PhD thesis on the 12th of October at 13:15. Congrats!
The defense took place in Faraday, Institutionen för fysik, Origovägen 6b, Göteborg.

Title: Advanced methods for the calibration of optical tweezers

Abstract: Optical tweezers have enabled the manipulation of micron-sized particles with great accuracy since their invention by Arthur Ashkin and colleagues in the 1980s. This technique has had an impact in multiple areas, including biology, physics, nanotechnology, spectroscopy, soft matter and nanothermodynamics.
To perform experiments requiring quantitative transduction
of forces with optical tweezers, the optical tweezers need to be calibrated; that is their stiffness needs to be determined. In this thesis, I present the results that I have obtained for the calibration of optical tweezers using probabilistic approaches.
The goal of these approaches is to use the available data most efficiently and even be able to have an estimation of the error associated with the calibration. This is of the utmost importance when one has limited data, as is often the case with systems out of equilibrium, low signal-to-noise ratios, and systems in which the conditions change with time quite fast. This thesis is divided into two problems. The first problem I had was the unavailability of a comprehensive method to measure force fields in extended, non-conservative, and unstable equilibrium points. For this problem I used Bayesian inference in the form of a maximum likelihood estimator, which allowed me to characterize the force field even in conditions previously not possible to tackle. This parameter-free method called FORMA proved to be more precise, accurate, faster, and less data-intensive than the previous conventional method, i.e. equipartition, MSD, ACF, and PSF. Not only that, but it allowed me to characterize the force field generated by Laguerre-Gaussian beams with different orbital/spin angular momentum, a double-well potential, and a speckle pattern.
The second problem I tackled was the error in the estimators due to
limited bandwidth and finite integration time. For this, we developed the joint probability density function of observing the particle at a given set of positions and times. We derived generalized formulas for the calibration methods; these new formulas successfully correct for the overestimation of the stiffness and the underestimation of the diffusion coefficient caused by a finite integration time; it also accounts for the limited sampling frequency and the trajectory length.
In general, this thesis shows the potential of having a probabilistic and inference approach to the problem of deducing the set of parameters that characterize the Langevin equation of motion of a particle from a time series of its position. The solution to this problem has applications not only to the calibration of optical tweezers but also to microrheology, the behavior of single molecules inside a cell, and animal migration.

Thesis: https://hdl.handle.net/2077/78214

Supervisor: Giovanni Volpe
Examiner: Mattias Goksör
Opponent: Balpreet Singh Ahluwalia
Committee: Thomas Huser, Juliane Simmchen, Kirstine Berg-Sørensen
Alternate board member: Mattias Marklund