N. Bordes, T. Hugh, M. Doherty, R. Martin and B. Pailthorpe
Resection or ablation of tumours is one treatment available for liver cancer sufferers. This delicate operation consists of removing the tumour(s) and surrounding healthy tissues. The surgery is complicated by the fact that major blood vessels are present in the liver: the surgeon must proceed cautiously. Computer Tomography (CT) scans are used to diagnose the presence of tumours in the liver but also to assess whether the patient is suitable for surgery. The surgeon needs to find the number of tumours, their size and the physical and spatial relationship between the tumours and the main blood vessels. Extracting this information from the CT scan is a time-consuming procedure, which requires manual contouring of the tumour and the main vessels and is complicated by the low contrast in the images. The aim of this project is to create a 3D model of the patient's liver from the CT scans for surgery planning.The data consist of a sequence of CT scan images taken through a torso and saved in DICOM format. Each image consists of a 3mm thick CT scan slice through the torso and the images are sequential, starting at the top of the liver (close to lungs) and moving progressively down through the torso. The contrast in the image is poor, showing very little detail in the liver.
The contrast in each each image was improved by using a series of filters including histogram equalisation and low-pass filtering (see Fig.1).
 |
 |
| Original
CT slice |
Same
slice after filtering |
The 3D reconstruction of the liver was achieved by creating a mask for each slice, consisting of ones in the selected section (the liver) and zeros everywhere else. Originally the mask was created manually however the process was later semi-automated by developing a region growing algorithm. The following is an example of a slice and its corresponding mask:
 |
 |
| Filtered
and cropped CT
slice |
Associated
mask |
 |
| Graphical User Interface
for semi-automatic mask generation |
In order to visualise the liver, tumour and blood vessels, the CT scans and the mask were read as separate volume, and then multiplied to isolate the liver. An isosurface was used to extract the blood vessels and the tumor as shown below.
 |
| 3D
reconstruction of the liver, blood vessels and tumours |
This work was originally started in VisLab at the University
of Sydney and continued at the University of Queensland.
Richard Brickacek and Nathan Ford, 3rd year students at the University of Sydney in 2001 helped with the initial data handling as part of their scientific visualization project.
The Royal North Shore hospital provided the CT scan. We thank the medical technical staff at RNS for their time.
Publications
R. Martin, N. Bordes, T. Hugh, B. Pailthorpe "Semi-Automatic Feature Delineation In Medical Images" Australasian Symposium on Information Visualisation, Christchurch, New Zealand, 2004. Conferences in Research and Practice in Information Technology, Vol. 35, pp 127, N. Churcher and C. Churcher Eds (2004).
R. Martin, "Semi-automatic Segmentation of Liver, Tumours and Blood Vessels from CT Scans" Honours Thesis (2003).
M. Doherty, N. Bordes, T. Hugh, B. Pailthorpe, "3D Visualisation of Tumours and Blood Vessels in Human Liver", Pan-Sydney Area Workshop on Visual Information Processing (VIP2002), Sydney, Australia. Conferences in Research and Practice in Information Technology, Vol. 22. Pp 27, J.S. Jin, P. Eades, D.D. Feng, H. Yan, Eds (2002).