Jacqueline Nerney Welch, Jeremy A. Johnson, Michael R. Bax, Rana Badr, Ramin Shahidi

1
A Real-time Freehand 3D Ultrasound System for
Image-guided Surgery

• Jacqueline Nerney Welch, Jeremy A. Johnson,
  Michael R. Bax, Rana Badr, Ramin Shahidi

IEEE Ultrasonics Symposium 2000 October 24, 2000
2
Overview

• Design motivations and decisions
• 3D ultrasound
• Freehand scanning
• Optical tracking
• Volume rendering
• Simultaneous acquisition and visualization
• Methods
• Equipment
• Spatial calibration
• Volume construction and maintenance
• Results
• Future Work

3
Ultrasound

• Ultrasound versus other imaging modalities (CT,
  MR, X-ray)
• Least expensive
• No ionizing radiation
• Compatible with existing surgical instruments
• Widely available and commonly used
• Real-time, interactive nature

4
3D Visualization of Ultrasound

• Compared to 2D, 3D provides
• More intuitive and comprehensible images
• More accurate volume estimation
• Shorter scanning times
• Improved sharing of information

2D Ultrasound Image
Volume Rendered 3D US
5
3D from Conventional 2D Ultrasound
2D Images
Volume Construction Engine
Position Data
Volume Rendering Engine
Workstation
US Probe
Tracking Device
6
Optically Tracked Freehand Acquisition

• Freehand versus other scanning techniques
  (mechanical)
• Greatest freedom of movement
• Compact
• Least cumbersome
• Requires probe position measurements
• Optical versus other position tracking methods
  (magnetic, mechanical, speckle decorrelation)
• Insensitive to metallic surgical equipment
• Allows volume localization

7
Interactive Volume Rendering

• Volume rendering versus other visualization
  methods (slice projection, surface rendering)
• Truest to the data set
• Easiest to interpret
• Segmentation not required
• Computationally expensive but feasible with
  current technology

8
Simultaneous Acquisition Visualization
Acquisition
9
Equipment

• Image Guided Technology FlashPoint 5000 optical
  tracking system with 580 mm camera
• Sonosite handheld ultrasound scanner with 5MHz
  linear probe
• SGI 320 Visual Workstation with a single
  processor running Windows NT

10
Image to Volume Mapping
11
Calibration Parameters
kP

• 6 extrinsic parameters
• Rotation (Ri , Rj , Rk)
• Translation (ti , tj , tk)
• 2 intrinsic parameters
• Image scale (si , sj)
• Can be written as

Probe Tracking Device Coordinates
iP
jP
(Ri , Rj , Rk) (ti , tj , tk) (si , sj)
iS, u
jS, v
Slice Coordinates
12
Calibration Phantom
Image of Phantom During Calibration
Ultrasound Phantom (1/16 Acrylic)
13
Calibration Method

• Obtain feature positions
• Align ultrasound probe
• Capture US image and probe position
• Localize features in image
• Calculate calibration parameters
• Scale factor
• Rotation and Translation

14
Volume Construction and Maintenance
Insertion of New Slices
Removal of Old Slices
Overwrite Existing Slices
Interpolate with Nearby Slices
15
Results
16
Results
17
Future Work

• Quantify and improve system performance
• Spatial and temporal accuracy
• Data rates
• Display position and trajectory of surgical
  instruments
• Apply system to clinical situations

18
Acknowledgements

• Dr. Thomas Krummels lab
• DOD Graduate Research Fellowship
• CBYON, Inc.