Cancer remains one of the leading causes of death worldwide. In traditional cancer diagnosis, which relies heavily on 2D imaging, a biopsy is taken from the area of interest, after which the tumour is sliced into thin sections and stained to visualize the cells. However, diagnosing cancer in 2D has significant limitations, such as difficulty in assessing the full extent of the disease and the potential for missing key details. To address these challenges, we propose a proof-of-concept for a 3D reconstruction of consecutive tissue slices.

The pipeline for achieving this proof-of-concept involves three key stages: image registration, image segmentation, and visualization. Image registration is essential for aligning the consecutive slices into a cohesive z-stack. Image segmentation assigns each pixel a value of 0 or 1, indicating whether it belongs to the tissue of interest or not. Finally, the z-stack of tumour slices is visualized using a meshing method, where each pixel is represented as a cube to create volumetric data.

This workflow demonstrates the feasibility of a 3D reconstruction of cancer cells. However, further optimization is required to enable full reconstruction of image sets without the need for rescaling. Additionally, the procedure should be tested on various types of tumours to assess its overall effectiveness.