Hi everyone,
I wanted to share a novel, non-clinical archaeological application of 3D Slicer (v5.8.1) from my Bachelor’s thesis in Archaeology. A big thank you to the 3D Slicer developer community for providing such an adaptable tool! I’d love to hear thoughts, critiques, or ideas!
My study focused on the non-destructive 3D reconstruction and virtual excavation of intact, block-lifted (en bloc) unurned Iron Age cremation burials (c. 800–400 BC) from the site of Haps-Laarakker. While clinical CT has occasionally been tested on ceramic cremation urns, applying a full open-source 3D workflow to unurned, soil-embedded cremation blocks has, to our knowledge, not ever been documented before.
Destructive physical excavation permanently destroys the fine spatial relationships of fragmented calcined bone, grave cuts, and organic container voids. Commercial imaging suites often create financial barriers for field archaeology units and cultural heritage labs. I committed to building an entirely open-source, fully reproducible workflow using 3D Slicer to ensure other researchers can replicate my methodology without proprietary software.
Key Technical Challenges & Slicer Workflow
The dataset was acquired on a Siemens SOMATOM Definition dual-source scanner (helical scan, reconstructed as DICOM image stacks across axial, coronal, and sagittal planes).
-
Scan Orientation Matters: I tested both horizontal and vertical scanning of the PVC block-lift tubes. Horizontal positioning produced severe streak artifacts, blurred boundaries, and artificial fragment fusion. Scanning the core vertically retained sharp density boundaries, which proved critical for segmenting sub-centimeter bone fragments.
-
Severe Density Overlap: Unlike fresh clinical bone, archaeological calcined bone, pottery shards, iron-rich sand, and gravel inclusions at Haps exhibit heavily overlapping radiodensities (attenuation values). Automated thresholding alone was insufficient:
-
I initiated segmentation using the Threshold effect combined with an island filter (Remove small islands < 120 voxels, calibrated to exclude the local sand grain size of <2000 μm).
-
I utilized Split islands to segments (resulting in over 800 individual candidate segments per core) followed by systematic multi-slice validation across all orthogonal views to remove noise and matrix false-positives.
-
-
Negative Space & Perishable Containers: Beyond bone, I used threshold masking and the Scissors/Cut tool to model negative space (voids and root structures). This allowed me to detect decayed organic structures (Type C cremation containers) and grave cut boundaries that were otherwise completely invisible during subsequent physical laboratory excavation.
-
Spatial Mapping with Markups/Glyphs: By matching numbered osteological fragments from laboratory excavation photos to our Slicer segments, we generated color-coded anatomical glyph models.
Archaeological Takeaway
In one burial (cremation 6436), Slicer’s spatial mapping proved that 24 of 29 identified skeletal fragments followed a deliberate anatomical sequence (cranium in the west, lower limbs in the east), with grave pottery positioned deliberately around rather than within the bone cluster. Slicer allowed us to prove intentional burial rites in an assemblage that would have and has looked like a random heap of fragments in conventional and laboratory excavation alike.
Paper & Zenodo DOI
For anyone interested in my resulting thesis:
Zenodo DOI: https://doi.org/10.5281/zenodo.22917511
Figure 1: Initial intensity thresholding in 3D Slicer. Raw density-based segmentation capturing calcined bone alongside surrounding sediment matrix and streak artifacts prior to voxel filtering. Figure 2: Segmented bone cluster following manual validation. 3D volumetric reconstruction after connected-component filtering (>120 voxels) and multi-slice manual clean-up, preserving the in situ spatial relationships of the cremation deposit. Figure 3: Spatial distribution of skeletal elements using color-coded glyphs. Integration of physical anthropological identifications with the 3D model, highlighting the deliberate anatomical ordering (cranium west to lower limbs east) in cremation 6436. Figure 4: Negative space segmentation revealing structural voids. Volumetric visualization of soil voids and diverted root paths, providing non-destructive evidence for a decayed organic container and grave cut outline (Type C cremation).
