Student for designing clinician-centred AI verification tools in radiotherapy

Are you interested in improving the implementation of AI and clinical practice? Do you want to contribute to research that develops user-friendly tools to optimise radiotherapy workflows and increase the adoption of DL models?

What will you do?

To optimise the implementation of deep learning (DL) models in the clinical radiotherapy workflow, we are developing an automated quality assurance (auto-QA) system that can detect when and where a DL model makes mistakes. A critical aspect of this system is ensuring that clinicians can easily and effectively use it in their daily practice and that their needs are taken into consideration.

• You will investigate clinician preferences and needs for an auto-QA tool in radiotherapy
• You will design and prototype a user-friendly interface or workflow for the auto-QA system
• You will collaborate with clinicians and researchers to ensure the tool is intuitive, efficient, and clinically valuable
• You will evaluate the usability of the tool through feedback sessions or pilot testing
• You will not have to design the DL segmentation model
• There is room for your own input and creativity
• There is interest in turning the results into an academic publication

What is this research about?

At the department of radiotherapy, a radiation treatment plan is made for every individual patient based on imaging scans. The tumour and organs at risk are delineated to optimise dose to the target and minimise dose to healthy surrounding tissue. However, manual delineation is time-consuming, so we have implemented automated delineation with a deep learning (DL) model since 2018. While DL models have high average accuracy, they can make mistakes on individual patients or structures. Therefore, all DL segmentations must be evaluated by clinicians. This requirement reduces clinician trust, limits usability and limits the adoption of DL models.

Therefore, there is a growing interest in methods to automatically assess the quality of DL segmentation and predict the influence of errors on the treatment plan. This information can then be provided alongside the model output to improve clinical implementation of the DL model. Recent research within the UMCG and other hospitals has developed technical solutions for error detection in DL segmentation, but the next challenge is designing tools that clinicians will actually want to use and that present this information in an intuitive way. This requires understanding their workflows, preferences and pain points with DL segmentation.

We are looking for a master’s thesis student who wants to contribute to developing a clinician-centred auto-QA tool that can be integrated into the radiotherapy workflow to increase the adoption of DL models.

What do we ask?

For this project, we are looking for a university master’s or bachelor’s student with a background in human-machine interaction, computational cognitive sciences, behavioural sciences, neuroscience, or a similar field.

• You have an interest in user-centred design and clinical applications
• You have experience with usability testing, UX/UI design, or human factors research
• You are able to independently conduct research and have strong analytical skills
• You enjoy collaborating with multidisciplinary teams (clinicians, AI researchers, etc.)

What do we offer?

• Internship agreement with UMCG
• Good supervision at UMCG
• Scientific working environment (AI in radiotherapy group)
• In consultation, you can partly work from home.