Abstract
Radiologists fulfill a critical role in our healthcare system, but their workload has increased substantially over time. Although algorithmic tools have been proposed to support the diagnostic process, the workload is not efficiently decreased in this manner. However, another possibility is to decrease workload in a different area. The main topic of this thesis is concerned with investigating how simulation training can be realized to aid in the image interpretation skills training of the radiology resident. To realize simulated training it is necessary to know (1) how we can create realistic artificial medical images, subsequently (2) How we can control their variety and (3) how we can adjust their difficulty.
Firstly, it is shown that artificial medical images can blend in with original ones. For this purpose a GAN model is used to create 2-dimensional artificial medical images. The created artificial images are assessed both quantitatively and qualitatively in terms of their realism. Secondly, to better control the variety of the artificial medical images a diffusion model is used to guide both coarse- and fine-features. The results show that the model was able to adjust fine-feature characteristics of the pathology type according to the feedback of the independent classifier. Thirdly, a method is presented to describe the detection difficulty of an (artificial) medical image using quantitative pathology and image characteristics. Results show that it is possible to describe almost two thirds of the variation in difficulty using these quantitative characteristics and as such describe images as having lower or higher detection difficulty.
Finally, the responsible implementation of the medical image simulator to assist in image interpretation skills is investigated. Combining the results of this thesis resulted in a prototype of a 'medical image simulator'. This simulator can take over part of the workload of the supervising radiologists, by providing a means for independent repetitive practice for the resident. The realistic artificial medical images can be varied in terms of their content and their difficulty. This can enable a personalized experience that can enhance training of image interpretation skills and make it more efficient.
Firstly, it is shown that artificial medical images can blend in with original ones. For this purpose a GAN model is used to create 2-dimensional artificial medical images. The created artificial images are assessed both quantitatively and qualitatively in terms of their realism. Secondly, to better control the variety of the artificial medical images a diffusion model is used to guide both coarse- and fine-features. The results show that the model was able to adjust fine-feature characteristics of the pathology type according to the feedback of the independent classifier. Thirdly, a method is presented to describe the detection difficulty of an (artificial) medical image using quantitative pathology and image characteristics. Results show that it is possible to describe almost two thirds of the variation in difficulty using these quantitative characteristics and as such describe images as having lower or higher detection difficulty.
Finally, the responsible implementation of the medical image simulator to assist in image interpretation skills is investigated. Combining the results of this thesis resulted in a prototype of a 'medical image simulator'. This simulator can take over part of the workload of the supervising radiologists, by providing a means for independent repetitive practice for the resident. The realistic artificial medical images can be varied in terms of their content and their difficulty. This can enable a personalized experience that can enhance training of image interpretation skills and make it more efficient.
| Original language | English |
|---|---|
| Qualification | Doctor of Philosophy |
| Awarding Institution |
|
| Supervisors/Advisors |
|
| Award date | 5 Jun 2024 |
| Place of Publication | Enschede |
| Publisher | |
| Print ISBNs | 978-90-365-6084-9 |
| Electronic ISBNs | 978-90-365-6085-6 |
| DOIs | |
| Publication status | Published - Jun 2024 |