University of Twente Student Theses
Validating XAI techniques in medical image diagnosis: A venture towards algorithm transparency in a socio-technical system.
Fürst, J. (2022) Validating XAI techniques in medical image diagnosis: A venture towards algorithm transparency in a socio-technical system.
|
PDF
726kB |
| Abstract: | Transparency and validation of relevant processes are key factors to consider when developing an algorithm that is to be integrated into a socio-technical system, i.e. a system that views the algorithm and the organization it will be employed in with all relevant surrounding features as one holistic instance, which is yet to be designed. In order to solve the problem of maintaining the transparency of algorithms used in medical diagnosis, it is hence crucial to utilize methods that help clinicians understand the technical processes forming the basis of the information provided by the algorithms. Thus, the application of Explainable Artificial Intelligence (XAI) techniques in this regard is not only advisable and incredibly important to maintain the trust- worthiness that is required for the clinicians to adopt the systems in the first place, but is indispensable for the doctor-patient relationship as well. Within this research, four XAI methods were picked for validation through the means of expert interviews with pathologists. The results showed that whilst there is a certain tendency of preferences within the stakeholder pop- ulation, it should be evaluated on a case-to-case basis which method is most applicable. Furthermore, the integration of the methods is a rather sensible issue. It should thus not only be evaluated which technique to integrate, but how to integrate the method to maximize trust and effectiveness through extensive use. If done effectively, these methods can facilitate the clinicians’ workflow significantly. |
| Item Type: | Essay (Bachelor) |
| Faculty: | EEMCS: Electrical Engineering, Mathematics and Computer Science |
| Subject: | 54 computer science |
| Programme: | Computer Science BSc (56964) |
| Link to this item: | https://purl.utwente.nl/essays/91707 |
| Export this item as: | BibTeX EndNote HTML Citation Reference Manager |
Show download statistics for this publication
Show download statistics for this publicationRepository Staff Only: item control page