New Developments in ImageGuided Prostate HIFU Surgery

1
New Developments in Image-Guided Prostate HIFU
Surgery

• Kris A. Dines
• XDATA Corporation
• Michael Penna, Rich Pfile
• Indiana University, Purdue University
  Indianapolis
• Roy Carlson, Russ Fedewa, Woshing Chen
• Focus Surgery
• Ralf Seip, Director RD, Focus Surgery
• Narendra T. Sanghvi, President CEO, Focus
  Surgery

This work was partially supported by SBIR NIH
Grant 2R44DK059664-02, awarded to Focus Surgery,
Inc. Indianapolis, IN. Principal Investigator
Narendra T. Sanghvi.
2
Development Areas

• Part 1
• Anatomic-Based Planning

• Part 2
• Lesion Detection

3
Part 1 Anatomic-Based Treatment Planning

• Current methods require zones on 15 slices
• No adjustment of on/off times for anatomy
• Need to recognize capsule many times
• New method based on model of capsule
• Recognize capsule once
• Leads to automatic layout of sites
• 3D visualization of all structures and sites

4
Current Sector-Linear Planning
5
Current Multiple-Slice Planning Screen
6
New Anatomic-Based Method

• Trace capsule on a few slices
• Identify urethra if desired
• Automatically detect rectal wall
• Model structures as mathematical 3d surfaces
• Automatically layout plan
• Edit plan if needed

7
Critical Anatomic Structures

• Rectal Wall protect
• Urethra Possibly protect
• Capsule Treat
• Nerve Bundles Protect
• Not modeled
• Visualized by Doppler
• Seminal Vesicles
• Not modeled

8
Anatomic Model-Based Treatment Planning
User Step 1
Work In-Progress
User Step 2
9
Slices for Tracing
10
Tracing Data for Surface fits
11
(No Transcript)
12
Automatic Refinement of Traces
Movie Automatic Play
Current area of investigation
13
(No Transcript)
14
Trace or Click Points
Rectal Wall automatically detected
15
How Our Approach is Novel

• Use a few sector and linear traces instead of
  many sector-only
• Use mathematical surface models instead of
  connecting slices
• Mathematical models ? Compute lesion placement
  automatically

16
Rectal Wall Model
The rectal wall is modeled by a tube whose
central axis coincides with the probe axis and
whose radius varies as a polynomial (typically
cubic) along its length
Least-squares fit of average radius of trace
sample points
17
Urethra Model
The urethra is modeled as a parametric tube whose
radius, R, is constant (2.5mm), and whose curved
central axis follows a polynomial path that is
typically cubic
Least-squares fit of center-gravity of trace
sample points at each sector slice
18
Capsule Model
Deformable ellipsoid
Least-squares fit of sample points
19
Simulated Capsule Fits
Ideal
15 Sector
5 Sector, 3 Linear
Accuracy 95
Accuracy 85
20
Surface Plots
Human Subject Early Test
21
Surface Fit Capsule, Rectal Wall, UrethraHuman
subject
2 Sector 2 Linear
22
Solid Model Capsule Rectal Wall
Edges detected by gradient in 3D
Capsule
Urethra
Lesion Sites (Treatment Plan)
Rectal Cavity
Transverse slice through 3d model Generated by
filling voxels inside surface models
23
Anatomic Models Embedded in Volume Image
VolumePro Display
24
Automatic Anatomic-Based Treatment Plan
25
3D Anatomic Model-Based Treatment PlanHuman
Subject vKal2
Movie Automatic Play
26
Lesion Placement in Human subject
27
With Urethra
28
Anatomic Model-Based Treatment PlanTissue-Mimicki
ng Prostate Phantom
Movie Automatic Play
29
Lesion Placement in Phantom
30
Image Processing Better Boundaries?

• Connect
• Rotational Kernel Transform
• max sum along all lines through the pixel
• Contrast
• Histogram Equalize using stats in central zone
• Sharpen
• unsharp masking

Current area of investigation
31
Conclusions

• Anatomic-Based Treatment Planning
• Reduces treatment planning time
• Potential to avoid complications
• Gives physician 3D view of plan

Almost Ready for Initial Clinical Testing
32
Part 2 Lesion Detection Control
33
Real-Time Detection of Multiple Lesions during
HIFU Treatments using Imaging (Pulse-Echo)
Ultrasound Hypothesis
Focal Zone
The RF ultrasound echo signal acquired from the
focal zone of the HIFU transducer during
treatment contains information which can be
extracted using signal processing techniques to
detect the HIFU-induced lesion.
Tissue
Confocal Imaging Transducer
HIFU Transducer
Probe Tip
34
Imaging 1. Scanning Typical Backscattered
Ultrasound Signals vs. Treatment Time
Before
T 1s
T 2s
Echo Amplitude
After
T 3s
Distance mm
35
Pulse-Echo Ultrasound RF Data Acquisition during
HIFU Delivery
HIFU ON
HIFU OFF


T0s
T-
T3s
T9s
8
A-Lines acquired during brief (50ms) pauses in
the HIFU ON Time once every 500ms. (6 Total)
A-Lines acquired during the HIFU OFF Time once
every 500ms. (3 Total)
Lesion Overlay Display is updated at these times
once every 500ms.
Pre-Treatment Reference Line
Pre-Lesion Reference Line
36
(No Transcript)
37
(No Transcript)
38
(No Transcript)
39
(No Transcript)
40
(No Transcript)
41
(No Transcript)
42
(No Transcript)
43
(No Transcript)
44
(No Transcript)
45
Conclusions

• Auto-Detect Lesion Formation
• Auto-Control HIFU application time/intensity
• Potential to Reduce Treatment time
• More test data needed to calibrate/validate

46
Future RD

• Automatic Detection capsule boundary
• Image enhancement ? improve boundaries
• Auto-detect lesion formation
• from changes in ultrasound signals
• control HIFU application time/intensity
• Spin-off of technology to breast surgery

47
Questions?
48
Imaging and Treatment Geometry
Sector Scan
Homogeneous Coordinate Transforms Relate Spaces
Linear Scan
Head
screen upper left
screen upper left
Foot