Sophie Fischer
Interests
I'm in my third year of a DPhil in Computer Science, and my research is in evaluation metrics in semantic segmentation for medical imaging:
- Semantic Segmentation - This is a specific task within the field of Computer Vision, where the goal is to identify which pixels make up a given object.
- Medical Imaging - I mainly work with medical imaging data, across a range of imaging modalities (e.g. MRI, CT, X-ray).
- Evaluation Metrics - When a model produces a segmenation, evaluation metrics are used to quantitatively measure how "correct" it is.
There are dozens of different metrics in use, and the choice of evaluation metric is critical in understanding the differences between segmentations of the same image. I'm particularly interested in understanding how evaluation metrics used within the Computer Vision community relate to the clinical measurements used in practice, which are often the reason the segmentation is required.
Within semantic segmentation for medical imaging I am also interested in:
- Loss functions
- Ground truth uncertainty
- Computational geometry
Selected Publications
-
Hairy ground truth enhancement for semantic segmentation
Sophie Fischer and Irina Voiculescu
In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) Workshops. June, 2024.
Details about Hairy ground truth enhancement for semantic segmentation | BibTeX data for Hairy ground truth enhancement for semantic segmentation | Link to Hairy ground truth enhancement for semantic segmentation