3D virtual atlas of the uterus.pdf.pdf (5.56 MB)
Download file

3D virtual atlas of the uterus

Download (5.56 MB)


Creating a 3D atlas of an organ’s structure is a highly desirable resource to have as this helps to improve in vivo anatomical understanding. One of the most poorly understood organ systems in the human body is the utero-placental circulation. A major limitation to understanding the utero-placental circulation is the inability to investigate the blood vessels in the pregnant uterus, which change rapidly and are inaccessible to direct measurement during pregnancy. Because of these limitations, the extent of their remodelling/adaptation to pregnancy has largely been described only qualitatively. A historic collection of gravid uterine samples our research group has accessed presents a unique opportunity to directly explore these structures for the first time.

The Boyd and Dixon Collections (Cambridge, UK), contain rare historical collections of preserved and sectioned gravid uterine tissue, treated with a range of histology stains to visualise different parts of the tissue. A slide scanner was used to digitally image specimens containing up to 510 serial sections from samples of 6 – 20 weeks of gestation. These samples were prepared in the 1950s, well before standardised sample preparation and digitisation was available, so are prepared inconveniently and often in poor quality to perform digital analysis. To reconstruct these images into a 3D stack and perform segmentation, an open source custom-written programme was designed specifically to address the novel issues associated with this data. This programme performs significant pre-processing to isolate the samples in their frames. Linear and non-linear registration methods have been developed to ensure the continuity of the structures. Interpolation between missing samples is performed to compensate for tissue damage. Machine Learning methods are used to learn both hand-picked and automatically collected features from the 3D model to segment the entire volume.

This programme has created 3D reconstructions of multiple specimens of gravid human uteri at different stages of gestation. The continuity of the 3D model reveals structures and vessel connectivity that were not previously evident in the 2D sections. Segmentation of specific features of interest is ongoing.

The 3D reconstruction of gravid uterine tissue significantly improves the ability to observe anatomical features. Segmenting individual vessels and features will further enable improve parameterisation of computational models of this circulation, to determine the impact of the dynamic changes in structure on blood flow haemodynamic over the first half of pregnancy.


Jonathan Reshef

Having studied Biomedical Engineering as his undergraduate at the UoA, Jonathan has continued his studies as a masters student with a particular interest in image processing. This work forms the body of his thesis.

Hanna Allerkamp

A post-doc research fellow with the Auckland Bioengineering Institute, Hanna continues to investigate the utero-placental circulation using animal models and understanding how these models relate to human development.

Jo James

Jo is a Senior Research Fellow in the Department of Obstetrics and Gynaecology at the Faculty of Medical and Health Sciences. She co-leads of the Placenta Modelling group which develops in vitro and in vivo silico tools to understand how a healthy placenta forms in early pregnancy, and also leads a research group focussed on the role of placental stem cells in fetal growth restriction.

Alys Clark

Senior Research Fellow in the Auckland Bioengineering Institute, Alys is one of the co-leads of the Placenta Modelling group which develops in vitro and in vivo silico tools to understand how a healthy placenta forms in early pregnancy. She also develops computational models of the lungs and ovaries with the goal of developing efficient and reliable methods for modelling physical processes that occur simultaneously in complex networks of tissue and blood vessels.


Usage metrics