Visual Analytics Interface for Digital Pathology in Hospitals


The UMC Utrecht Department of Pathology is one of the frontrunners of digital pathology in Europe. The department has published numerous articles on the topics of validation and implementation of Digital Pathology solutions, and has active research projects in image processing and machine learning for anatomic pathology applications.

Since February 2016, the department has implemented a fully digital pathology workflow using a Picture Archiving and Communication System (PACS) delivered by Sectra. The full digitization of the workflow has enabled digitally availability of the cases before the glass slides are distributed, which gives pathologists an opportunity to perform digital case review.

The full digital workflow offers a unique possibility to apply image processing and machine learning algorithms to specific cases before the digital review, presenting the pathologists with the results from the analysis as soon as they open the case for review. This upfront case analysis by machine learning algorithms can offer pathologists meaningfully support for their daily diagnostic tasks, saving them time and increasing the quality of diagnosis substantially.


Type master project
Place external
Supervisors Michel Westenberg, Mitko Veta


More information: link
Relevant paper: link


With this project, we aim to build an interface between our machine learning algorithms and the PACS system. The implemented solution will have to:

a)   Retrieve information from the Laboratory Management System (LMS), that keeps track of which slides have been scanned and which type of tissue they contain.

b)  Use that information to select the appropriate machine learning algorithm.

c)  Retrieve the images from the PACS system using the Sectra PACS API and apply the algorithm to the images.

d)  Present the results, the original images, and the information from the LMS in a meaningful and intuitive way to the pathologist via the image viewer and possibly a new visualization component.

The master student working on this project will have to implement this interface. Once the system is implemented in practice, the student will perform a study about the effects of the use of the results from the machine learning algorithms on the work of the pathologists.


  • The student must be comfortable using Python and have coding experience.
  • Communication for this project will be primarily in English.

Stretch goals

The student can perform an analysis to determine the optimal way of presenting results of the machine learning algorithms to the viewer. Which type of annotations would increase viewability and readability of the results when they are viewed by an expert?