DICOMplicate
(2026) EEML05 20261Division for Biomedical Engineering
- Abstract
- Breast cancer is the most prevalent form of cancer among women in Sweden, with 9 000 diagnoses during 2023. While the five-year survival rate reaches 92.5% due to national screening programs, transitioning from 2D Digital Mammography to 3D Digital Breast Tomosynthesis (DBT) could further improve early detection. However, the high volumes of data increases the examination times, preventing it from becoming the clinical standard. While AI models offer a potential to speed up the process, their development is limited by both the need of large datasets to ensure unbiased and accurate models, and the need of pseudonymization to follow patient integrity laws. The M-BIG database, containing 450 000 examinations, could be a significant resource... (More)
- Breast cancer is the most prevalent form of cancer among women in Sweden, with 9 000 diagnoses during 2023. While the five-year survival rate reaches 92.5% due to national screening programs, transitioning from 2D Digital Mammography to 3D Digital Breast Tomosynthesis (DBT) could further improve early detection. However, the high volumes of data increases the examination times, preventing it from becoming the clinical standard. While AI models offer a potential to speed up the process, their development is limited by both the need of large datasets to ensure unbiased and accurate models, and the need of pseudonymization to follow patient integrity laws. The M-BIG database, containing 450 000 examinations, could be a significant resource but lacks a scalable tool for data pseudonymization. This project addresses this need by developing DICOMplicate, a graphical user interface built with Streamlit. The tool allows the user to perform SQL queries and export DICOM files where all patient-identifying metadata is automatically removed. By ensuring safe pseudonymization while enabling easy data extraction, DICOMplicate makes it possible for more researchers to access the data. Performance evaluations show that the tool can reliably export 5 000 images in 30 minutes, demonstrating its potential to become a valuable tool in breast cancer research. (Less)
Please use this url to cite or link to this publication:
https://lup.lub.lu.se/student-papers/record/9240470
- author
- Indebetou, Daga LU and Piehl, Virginia LU
- supervisor
-
- Magnus Dustler LU
- Daniel Förnvik LU
- organization
- course
- EEML05 20261
- year
- 2026
- type
- M2 - Bachelor Degree
- subject
- keywords
- DICOM, Pseudonymisering, Bröstcancer, Användargränssnitt, Datadelning
- language
- Swedish
- id
- 9240470
- date added to LUP
- 2026-06-23 12:43:58
- date last changed
- 2026-06-23 12:43:58
@misc{9240470,
abstract = {{Breast cancer is the most prevalent form of cancer among women in Sweden, with 9 000 diagnoses during 2023. While the five-year survival rate reaches 92.5% due to national screening programs, transitioning from 2D Digital Mammography to 3D Digital Breast Tomosynthesis (DBT) could further improve early detection. However, the high volumes of data increases the examination times, preventing it from becoming the clinical standard. While AI models offer a potential to speed up the process, their development is limited by both the need of large datasets to ensure unbiased and accurate models, and the need of pseudonymization to follow patient integrity laws. The M-BIG database, containing 450 000 examinations, could be a significant resource but lacks a scalable tool for data pseudonymization. This project addresses this need by developing DICOMplicate, a graphical user interface built with Streamlit. The tool allows the user to perform SQL queries and export DICOM files where all patient-identifying metadata is automatically removed. By ensuring safe pseudonymization while enabling easy data extraction, DICOMplicate makes it possible for more researchers to access the data. Performance evaluations show that the tool can reliably export 5 000 images in 30 minutes, demonstrating its potential to become a valuable tool in breast cancer research.}},
author = {{Indebetou, Daga and Piehl, Virginia}},
language = {{swe}},
note = {{Student Paper}},
title = {{DICOMplicate}},
year = {{2026}},
}