Flagship 4: Patient-specific virtual organs for surgical planning

Principal Investigators

Professor Peter Hunter, University of Auckland
Professor John Windsor, University of Auckland

Associate Investigators

Dr Harvey Ho, University of Auckland
Dr Paul Harris, Callaghan Innovation

Summary

This Flagship will develop a clinical workflow for pre-surgical analysis and planning for the treatment of tumours and other pathologies in soft tissue organs, using patient-specific image-based models.

The software-based workflow will allow uploading of pre-operative patient imaging of soft tissue organs - such as the liver, prostate, thyroid or pancreas - and will deliver a patient-specific model of the organ’s 3D surface geometry and embedded models of, for example, the intrahepatic vascular and biliary anatomy and their relationship to treatment targets (such as a tumour). The model will also facilitate the registration and fusion of patient-specific data from multiple image modalities. The pre-operative planning service will include facilities to compute blood flow in the vascular system, given suitable boundary flows measured with Doppler ultrasound, and, when needed, optimisation of thermal ablation strategies for treating a tumour.

The initial milestones, with an indication of current progress, are:

1. 3D liver model for surgical planning

Several models needed for the virtual liver are being developed: (i) a 3D anatomical model that captures the geometry of the liver surfaces and can be morphed into a patient-specific image dataset, (ii) a macro-circulation model on which the fluid flow equations are solved, (iii) a micro-circulation model that will enable the study of tissue regeneration post-surgery.

The figure below shows (a) the 8 anatomically defined Couinaud segments, (b) bounding boxes, based on these segments, that (c) can be used to morph the generic vascular model to fit patient specific fiducial markers.

A composite image of three figures of different models of human kidney shown side-by-side, with left-to-right arrows between them. The figures are labelled (a), (b) and (c) respectively.
A blood flow model has been developed that allows rapid computation of pressures and flows in the vascular system of the liver, given suitable boundary conditions, to predict the outcome of surgical interventions e.g. left or right lobectomy (see below). This will be available through the GUI described under the next milestone.
A composite image that consists of two drawings of the heart with blood vessels, and a multi-line graph that has time (in seconds) on the x-axis and flow rate (in mL/min.)

2. Clinical image pipeline

A graphical user interface (GUI) is being developed to import patient image data from a PACS and to import the 3D liver model (Fig.(a) below). Methods for aligning the 8 liver segments with fiducial markers from the vascular tree are being developed, as shown below. Note that the user can move the nodes of the morphing mesh in (a) to manually align the model segments with patient images, or (b) use computed metrics (green lines) for fiducial points to automatically find the nodal parameters of the morphing mesh that best fit the patient anatomy.

A composite image of three pictures, presented side-by-side, labelled with (a), (b) and (c) respectively. (a) is a screen capture of a program with a render of the human liver in it. (b) and (c) are different renders of the liver.

The next stage of this development is to use functional measures of segment perfusion (based on the flow modelling described above) to provide the fiducial markers for aligning the generic model with the patient model.

3. MicroCT imaging of liver, prostate, thyroid & pancreas

We are investigating the use of our new MedTech CoRE microCT and the new Otago MARS scanner for obtaining structural information about the liver and other soft organs. Initial experiments have been performed. An illustration of various images of a vascular tumour in liver is shown below (A: Fluorescence microscopy; B: MicroCT; C: 3D image rendering; D: Photomicroscopic image. Ref?).
A composite image of four scan images, labelled A, B, C and D respectively.

4. Hardware for liver surgery

3D Oculus Rift glasses have been purchased and are being linked into the framework above. An ultrasound probe is also being developed in a separately funded Seed project led by Dr Paul Harris.