When it comes to pediatric oncology surgery, i.e. surgery of children with cancer, there is a lack of technical innovations specifically designed and developed for this field. One technique is three-dimensional (3D) visualization, which allows quick mapping of the findings in an examination. However, since tumour tissue may have similar density as the surrounding soft tissue, it can be difficult to visualize without using image segmentation.

In this project, testing of the segmentation program SmartPaint was performed on neuroblastoma in pediatric imaging data. A pipeline was created, from image data input to a 3D visualization of the tumour in relation to its surroundings. The images used in this project were provided by the Department of Radiology at the Uppsala University Hospital and consisted of two anonymized T2-weighted MRI (magnetic resonance imaging) scans of a pediatric patient with abdominal neuroblastoma. 

To incorporate SmartPaint in the pipeline, further development was performed with the main modification being the addition of support for loading multiple two-dimensional (2D) DICOM images that together build a 3D volume. Visualizations were created in the software 3D Slicer, with the help of volume rendering, multiplanar reformation (MPR) and a segmentation module called TotalSegmentator.

With this project, it has been shown how straightforward the process of acquiring a 3D visualization of a tumour can be. Feedback on SmartPaint was received from radiologists and implementing this feedback, as well as performing a more quantitative evaluation of the pipeline, remain to be completed in the future.