2010 Winter Project Week Spine Segmentation Module in Slicer3

• 

  Our 3DSlicer module provides an automated segmentation of the spine in view of the treatment of disc herniation. (video)

• 

  This 3D rendering shows the nerve bundles surrounded by the cerebro-spinal fluid as they traverse the vertebrae.

• 

  Back pain is caused by a disc that protrudes into the vertebral foramen and pinches the nerve bundles as they exit the spine.

• 

  Our 3DSlicer module for automated segmentation was presented at the NA-MIC Project Week in Salt Lake City, Jan 2010.

Key Investigators

• Sylvain Jaume (MIT)
• Martin Löpprich (University of Heidelberg)
• Ron Kikinis, Steve Pieper (BWH)
• Polina Golland (MIT), Ehud Schmidt (BWH)

Objective

We are developing a Slicer module to segment the region within the thecal sac in MRI images of the spine. Our objective is to provide a segmentation and visualization tool to improve the treatment of disc herniation. The structures of interests are the cerebro-spinal fluid (CSF), the discs, the vertebrae and the spinal nerves.

Approach, Plan

Our plan for the project week is to integrate our code into Slicer 3.5. Our code analyzes the intensity profile of different regions in the MRI and automatically defines the optimum region for the CSF.

Progress

The algorithm has been implemented in Slicer 3.5 as an extension module. This volumetric pattern recognition algorithm is fully implemented as a Slicer module using ITK and VTK. The module has been tested on data sets acquired at Brigham and Women's Hospital.

Illustrations

The cause of back pain is the pressure of the herniated disc onto the spinal nerves. The herniated disc pushes the canal for the cerebro-spinal fluid (CSF). The resulting pressure on the nerve bundles can cause severe discomfort and even permanent disability if not treated.

The segmentation the spine overlaid on the MRI image can help the surgeon during a minimally invasive intervention on the herniated disc.

This Slicer rendering shows an herniated disc that protrudes into the vertebral foramen and reduces the opening for the nerve bundles and the cerebro-spinal fluid (CSF).

The CSF has a double role: it acts as a lubricant between the vertebrae and as a mechanical insulation to protect the nerves from outside shocks. The main challenge is to perform the segmentation in a fully automated way.

Surface meshes and volumetric meshes of our automatically created model of the CSF. The paths where the nerves exit the spine appear as two roots descending on each side of the spinal column.

Results of our automated nerve segmentation algorithm. The 3D rendering shows the nerves exiting the spinal cord and the ganglia.

Results of the SpineSegmentation module in Slicer 3.5

Results of the SpineSegmentation module in Slicer 3.5

Double cylinder used for our model-based.

A complex model of a lumbar vertebrae has been created to drive our shape-based segmentation algorithm in a robust and accurate recognition

Intensity profile for the cross-section of a nerve. The MRI image was acquired using the wideband steady-state free-precession (WB-SSFP) protocol.

Organization of the source code for the SpineSegmentation module. This module is available as an extension module for Slicer 3.5.

References

• Segmentation using Slicer 3.5, EMSegment module