Design sample

A university lab website design sample by Dicolab. MEDAI Lab is fictional — its university, people, projects, papers and posts are not real, and the slices are hand-drawn diagrams, not patient images. Design © Dicolab. Copying or reusing it without permission may lead to legal action. More design samples (Korean)

KOEN

Research

Organ segmentation

Delineating liver, spleen, kidneys and aorta slice by slice in abdominal CT.

Segmentation example on a hand-drawn slice
Example — a hand-drawn diagram. A real site would show figures the lab holds the rights to.

The question

Even for the same liver, boundaries are drawn a little differently depending on the scanner, the protocol and the annotator. We study segmentation models that draw boundaries consistently with few labels and across scanners. We are particularly interested in looking at neighbouring slices together, so that contours do not jump abruptly from one slice to the next.

Methods, data and evaluation

Organ segmentation: methods, data and evaluation
StructuresLiver · spleen · kidneys · aorta and its main branches
DataOur own mathematical phantom; de-identified data from partner institutions (after ethics review)
MethodsLabel-efficient learning, slice-consistency constraints, boundary-weighted training
EvaluationOverlap and boundary-distance metrics, slice-to-slice jitter
ReleasedSynthetic abdominal phantom slices · annotation rules

Selected papers

All 9 papers on organ segmentation
  1. 21
    “Few-label abdominal organ segmentation with slice-consistency priors”

    N. Kim+, D. Choi+, H. Do*

    Journal of Medical Image Computing, Vol. 14, No. 3, 2026.

    Int’l journalOrgan segmentationLINK full textPDF file
  2. 18
    “Evaluating slice-to-slice consistency of abdominal CT organ segmentation” (in Korean)

    N. Kim, Y. Seo, H. Do*

    Journal of the Korean Society of Medical Imaging AI, Vol. 7, No. 2, 2026.

    Domestic journalOrgan segmentationUncertainty & reading supportLINK full textPDF file
  3. 15
    “Scanner-robust kidney segmentation via intensity re-mapping”

    J. Han, N. Kim, H. Do*

    Proc. Asian Conference on Medical Imaging, pp. 45–54, 2025.

    Int’l conferenceOrgan segmentationLINK full text

People

  • Hajin DoPrincipal investigator
  • Narae KimLabel-efficient organ segmentation
  • Daon Choi3D vessel segmentation
  • Jiwoo HanSegmentation across scanners
  • Garam YoonAnnotation tool

Other topics