Master projects

Prediction of Paediatric Anticancer Drug Sensitivity from Multi-omics Inputs

Student

Hui Jeong Jung

external page LinkedIn Hui Jeong Jung

Supervisor

Sarah Brüningk, Catherine Jutzeler

Summary

In her masters thesis, Hui Jeong investigates state of the art models for the prediction of drug sensitivity in adult and pediatric cancer cell lines. While adult cancer cell lines have been extensively studied and modeled, paediatric cancer focused models are rare. Recently, a new paediatric cancer cell line multi-omics dataset has been released. This dataset includes DNA sequencing, RNA sequencing, DNA methylation, and pharmacological and genetic CRISPR-Cas9 loss of function screen data. The aim of this project is to characterize and predict paediatric cancer drug sensitivity from these multi-omics inputs and to contrast these against adult tumor drug response.

Temporal Patterns in Spinal Cord Injury Recovery

Student

Maya Louage

external page LinkedIn Maya Louage

Supervisor

Lucie Bourguignon

Summary

In her masters thesis, Maya investigates the relation between time of injury and spinal cord injury (SCI) recovery by employing novel statistical tools in hypothesis testing within the SCI field. In her masters thesis, Maya investigates the relation between time of injury and spinal cord injury (SCI) recovery by applying previously unused statistical tools in the SCI hypothesis testing field. The motivation behind investigating the time of injury hypotheses is multifaceted. Firstly, trauma injuries often exhibit "off-hour effects," linked to treatment complications. Secondly, infections are known to hinder SCI recovery, prompting an examination of potential seasonal effects due to infections' seasonal patterns. Lastly, Maya analyzes the recovery of injuries during the COVID period, considering the surge in infections and potential disruptions in rehabilitation access during this time.

Exploring novel metrics for assessing and predicting recovery after spinal cord injury

Student

Adrian Kaufmann

external page LinkedIn Adrian Kaufmann

Supervisors

Louis Lukas

Summary

In this thesis, Adrian is working on graph-based subphenotypoing and low dimensional representation of patient trajectories. Additionally, his project aims to explore novel metrics for assessing and predicting recovery after spinal cord injury.

Joint Representation Learning for Oncology Application

Student

Tanya Nandan

external page LinkedIn Tanya Nandan

Supervisors

Sarah Brueningk, Samuel Håkansson

Summary

In this thesis, Tanya employs a multi-modal approach, integrating various data modalities such as MRI imaging, RNAseq, and Copy Number Variation (CNV) to offer a more comprehensive understanding of brain tumors. Two recently published unsupervised machine learning algorithms, namely Multi-view Spectral Clustering and Joint Multidimensional Scaling, are being implemented. The objective is to derive meaningful joint representations of image and sequencing data. The overarching goal is to identify specific subcategories of tumors and unveil corresponding biomarkers that play pivotal roles in prognosis and therapy sensitivity.

Automatic Tissue Diffusion Segmentation from Multimodal Lumbar Spine Magnetic Resonance Imaging (MRI)

Student

Thomas Iff

external page LinkedIn Thomas Iff

Supervisor

Maria Monzon

Summary

The thesis of Thomas focuses on automating the multiclas image segmentation of multimodal magnetic resonance images of the spine. Our goal is to develop an algorithm grounded in diffusion models and subsequently assess its performance by comparing it to the current state-of-the-art methods employed for this specific task.

Image-based Predictions of Radiotherapy Response in Paediatric Diffuse Midline Glioma

Student

Julian Ortner

external page LinkedIn Julian Ortner

Supervisor

Sarah Brüningk, Daria Laslo

Summary

The master's thesis of Julian revolves around predicting radiotherapy response in pediatric diffuse midline glioma from non-invasive magnetic resonance imaging acquired before radiotherapy. To this end, different data-driven evaluation methods, both supervised and unsupervised, will be explored.

Prediction of Lumbar Spinal Stenosis from MRI-Images with Focus on Explainability for Clinical Application

Student

Adriana Brenzikofer

external page LinkedIn Adriana Brenzikofer

Supervisor

Andrea Cina, Maria Monzon

Summary

In her masters thesis, Adriana focuses on the prediction of Lumbar Spinal Stenosis (LSS) from MRI Images using Convolutional Neural Networks (CNN’s). Lumbar spinal stenosis, a common degenerative condition affecting the spine, can lead to significant pain and disability, making early and accurate diagnosis critical. Additionally, this work will address the challenge of applying deep learning models to real-world clinical settings with a focus on explainability and uncertainty estimation. This approach will allow clinicians to gain insights into model’s decision-making process and assess the reliability of its predictions.

Assessment of Domain Generalisation Algorithms in Clinical Datasets

Students

Sofie Bolleter, Linus Werren, Tamara Schneider

external page Linkedin Sofie Bolleter

Supervisors

Aspa Vozi, Hugo Madge Leon, Lakmal Meegahapola

Summary

This focus project evaluates the performance of domain generalisation algorithms for sepsis prediction in intensive care units (ICUs). We leverage publicly available ICU time-series datasets (eICU, MIMIC-IV, and HiRID) to develop and analyse predictive models for sepsis onset. A selection of algorithmic approaches are applied to these models and evaluated to assess their generalisation performance across domains (e.g. different countries or institutions). This study aims to provide a comprehensive analysis of these approaches and their effect on sepsis prediction.

Leveraging Eye and Mouse Tracking to Capture Uncertainty in Manual Tumor Segmentation

Student

Nina Baumgartner

external page LinkedIn Nina Baumgartner

Supervisor

Daria Laslo

Summary

Diffuse midline gliomas are aggressive brain tumours, oftentimes with poorly defined boundaries, making both manual and automated segmentation challenging. While uncertainty annotations are increasingly used in AI models, manual segmentations still lack this valuable information. In this thesis, I use eye- and mouse-tracking data from radiologists during tumour outlining to capture segmentation uncertainty passively without adding to their workload. Such uncertainty annotations can enrich downstream applications, including radiomics and radiotherapy planning, by offering a more nuanced view of the tumor's structure.

Using Large Language Models for Sepsis Prediction on in Intensive Care Units

Student

Jan Berner

external page LinkedIn Jan Berner

Supervisor

Lakmal Meegahapola

Summary

This master’s thesis explores the use of large language models (LLMs) for sepsis prediction in an ICU setting. By converting multimodal time-series data into a text-based format, we aim to enable more personalized and early detection of sepsis. We evaluate this novel approach by benchmarking it against traditional methods to assess whether LLMs can improve predictive accuracy.

Using Large Language Models with Multimodal Time Series Data for Critical Event Prediction in Intensive Care Units

Student

Sophia Ehlers

external page LinkedIn Sophia Ehlers

Supervisor

Lakmal Meegahapola

Summary

The master thesis explores the potential of Large Language Models (LLMs) to predict critical events in Intensive Care Units, including mortality, acute kidney injury, and sepsis. The research utilizes multimodal time series data from prominent public datasets (HiRID, MIMIC-IV, and eICU) to develop innovative prediction models. The methodology implements specialized techniques including prompt engineering tailored for structured clinical time series data, medical-specific LLM fine-tuning, and comprehensive cross-dataset validation to explore model performance across diverse clinical settings.