ABCViz

AI + Breast Cancer Visualization

Information Visualization IVIS

2024-03-12

Nathalie Lock

nlock@kth.se

Carolina Dexwik

cdexwik@kth.se

Fernando Cossio

fecr@kth.se

Ankit Grover

agrover@kth.se

Annika Süßenbach

assus@kth.se

Blend AI and human knowledge in the detection of breast cancer through a visualization tool aimed at radiologists.

Target 3.d: Strengthen the capacity of all countries, in particular developing countries, for early warning, risk reduction and management of national and global health risks. [1]

[1] https://sdgs.un.org/goals/goal3#targets_and_indicators

Project goal

Users and tasks

• Radiologists
• Manager of hospital�

​

​

• Overview of patient data
• Sort and filter data
• View explanations of the model
• View patient history

�

​

​

The data

Patient names were randomly generated from a list of common swedish names.

Data displayed in public demo was obtained from EA1141 [1]. Data displayed in final presentation was obtained from CSAW-CC [2].

[1] Comstock, C. E., Gatsonis, C., Newstead, G. M., Snyder, B. S., Gareen, I. F., Bergin, J. T., Rahbar, H., Sung, J. S., Jacobs, C., Harvey, J. A., Nicholson, M. H., Ward, R. C., Holt, J., Prather, A., Miller, K. D., Schnall, M. D., & Kuhl, C. K. (2023). Abbreviated Breast MRI and Digital Tomosynthesis Mammography in Screening Women With Dense Breasts (EA1141) (Version 1) [dataset]. The Cancer Imaging Archive. https://doi.org/10.7937/2BAS-HR33

[2] Strand, F. (2022). CSAW-CC (mammography) – a dataset for AI research to improve screening, diagnostics and prognostics of breast cancer (Version 1) [Data set]. Karolinska Institutet. https://doi.org/10.5878/45vm-t798

Tools & libraries

Evaluations with radiologists

​

• “Find it quite easy to understand”
• Should be able to sort by all scores, not only overall

​

​

​

FEEDBACK

​

• Larger images, “it would be nice to see the highlighted area”
• Scores not clear
• Can’t click on images??

​

​

​

• Overview: “very useful, when you have so many patients in the list its hard to navigate through it. This would help certainly.”

Challenges

• Making a useful interface for the radiologists VS. making cool visualizations for the course
• Find mammograms that can be shown publicly

​

​

�

​

• Learning D3.js
• Svelte/JS
• Preparing heatmaps

�

​

Lessons learned

• D3.js
• Svelte/HTML/CSS/JS

​

​

​

• Organization
• Dividing tasks
• Communication
• Develop and Test�

​

ABCViz Web App

https://abc-viz.vercel.app/

Thank you for listening!

Questions?

​

Nathalie Lock nlock@kth.se

Carolina Dexwik cdexwik@kth.se

Ankit Grover agrover@kth.se

Annika Süßenbach assus@kth.se

Fernando Cossio fecr@kth.se

​

Mario Romero Vega {marior@kth.se}

​

​

ABCViz

2024-03-12