Presentation is loading. Please wait.

Presentation is loading. Please wait.

©2005 Surgical Planning Laboratory, ARR Slide 1 Prostate Image Processing Steven Haker, PhD.

Similar presentations


Presentation on theme: "©2005 Surgical Planning Laboratory, ARR Slide 1 Prostate Image Processing Steven Haker, PhD."— Presentation transcript:

1 ©2005 Surgical Planning Laboratory, ARR Slide 1 Prostate Image Processing Steven Haker, PhD

2 ©2005 Surgical Planning Laboratory, ARR Slide 2 The Basic Problem New MR imaging parameters and high field strengths hold promise for increased sensitivity and specificity in cancer detection.

3 ©2005 Surgical Planning Laboratory, ARR Slide 3 The Basic Problem New MR imaging parameters and high field strengths hold promise for increased sensitivity and specificity in cancer detection. Constraints of low-field strength intra-operative scanners and limited time in the operating room limit imaging.

4 ©2005 Surgical Planning Laboratory, ARR Slide 4 The Basic Problem New MR imaging parameters and high field strengths hold promise for increased sensitivity and specificity in cancer detection. Constraints of low-field strength intra-operative scanners and limited time in the operating room limit imaging. Solution: Image registration, which allows all pre-operative imaging to be used for targeted therapy.

5 ©2005 Surgical Planning Laboratory, ARR Slide 5 Challenge: MR-Guided Prostate Therapy High quality 1.5 Tesla imaging is not available in the our 0.5T intra-operative MRT

6 ©2005 Surgical Planning Laboratory, ARR Slide 6 Promise: New MR Acquisitions Promising new imaging techniques are not readily available in the operating room. Haker, ISMRM 2005 Diffusion ImagingT2-Weighted Imaging

7 ©2005 Surgical Planning Laboratory, ARR Slide 7 Promise: New MR Acquisitions MR-Spectroscopy T2-Weighted Imaging MR Spectroscopy yields information on local metabolism

8 ©2005 Surgical Planning Laboratory, ARR Slide 8 Image Fusion for Increased Specificity Summary Statistical Mapping showing most suspicious focus in red. Using image fusion and machine learning techniques, we maximize information from multiple MR imaging techniques into one summary image. Chan, et al., 2003

9 ©2005 Surgical Planning Laboratory, ARR Slide 9 The Need for Registration How can these imaging techniques be used in the operating room to guide therapy? –Limited time –Low field-strength

10 ©2005 Surgical Planning Laboratory, ARR Slide 10 The Need for Registration How can these imaging techniques be used in the operating room to guide therapy? –Limited time –Low field-strength Image registration allows us to align pre- operative imaging with intra-operative imaging. –Targets may be chosen in pre-operative imaging and used to guide therapy. –Pre-operative imaging can be visualized overlaid with intra-operative imaging.

11 ©2005 Surgical Planning Laboratory, ARR Slide 11 Intra-operative 0.5T Pre-operative 1.5T T2 Deformed pre-op T2 FSE Registration – Our Method Registration

12 ©2005 Surgical Planning Laboratory, ARR Slide 12 Surface Matcher – Elastic Model Pre-Op Intra-Op

13 ©2005 Surgical Planning Laboratory, ARR Slide 13 Registration Pre-operative data (left), needs to be registered to intra-operative images (right). We use a bio-mechanical model to warp the central gland (red) and peripheral zone (green).

14 ©2005 Surgical Planning Laboratory, ARR Slide 14 Non-Rigid Registration - Prostate Registration of T2 imaging yields image with greater conspicuity

15 ©2005 Surgical Planning Laboratory, ARR Slide 15 Registration – Pre-Operative Targeting On the left, targets for biopsy chosen using pre-operative 1.5 Tesla MR imaging. On the right, registered positions in intra- operative 0.5 T imaging taken in the operating room. Davatzikos External Collaboration (UPenn)

16 ©2005 Surgical Planning Laboratory, ARR Slide 16 Clinical Application – Prostate Therapy Registration for targeted MR biopsy –Done regularly as part of our MR-guided prostate biopsy program. –Targets from imaging (Spectroscopy, T2-W imaging) –Targets from our collaboration with U Penn (statistical model of likely cancer sites)

17 ©2005 Surgical Planning Laboratory, ARR Slide 17 Clinical Application – Prostate Therapy Registration for targeted MR biopsy –Done regularly as part of our MR-guided prostate biopsy program. –Targets from imaging (Spectroscopy, T2-W imaging) –Targets from our collaboration with U Penn (statistical model of likely cancer sites) Visualization for needle guidance –Image overlay of preop and intraoperative imaging –Visual feedback of needle placement –Visualization of targets

18 ©2005 Surgical Planning Laboratory, ARR Slide 18 3D Slicer The 3D Slicer is the SPL’s workhorse for visualization and image processing.

19 ©2005 Surgical Planning Laboratory, ARR Slide 19 Real-Time MR for Biopsy Guidance Real-time FGR image (left) taken during biopsy with needle artifact (arrow). T2 image (right) taken before needle placement, and sampled to display the same spatial locations. Note that in the T2 image on the right the peripheral zone (bright area) is clearly visible (arrow).

20 ©2005 Surgical Planning Laboratory, ARR Slide 20 Real-Time Display Imaging taken before needle insertion (T2-W, left) and during needle placement (middle) can be fused to allow for visualization of prostate peripheral zone and needle simultaneously (right).

21 ©2005 Surgical Planning Laboratory, ARR Slide 21 Real-Time Display Imaging taken before needle insertion (T2-W, left) and during needle placement (middle) can be fused to allow for visualization of prostate peripheral zone and needle simultaneously (right).

22 ©2005 Surgical Planning Laboratory, ARR Slide 22 Image Fusion and Visualization Real time intra-operative images and registered pre-operative image can be fused to aid in needle guidance. Images not otherwise available in the operating room can be utilized. M. So (R25 Fellow) RSNA 2004

23 ©2005 Surgical Planning Laboratory, ARR Slide 23 Program Interaction Capturing Intraoperative Deformations: Research Experience At The Surgical Planning Laboratory Simon K. Warfield, Steven J. Haker, Ion-Florin Talos, Corey A. Kemper, Neil Weisenfeld, Andrea U. J. Mewes, Daniel Goldberg- Zimring, Kelly H. Zou, Carl-Fredrik Westin, William M. Wells, Clare M. C. Tempany, Alexandra Golby, Peter M. Black, Ferenc A. Jolesz and Ron Kikinis. Medical Image Analysis 9 (2005) pp 145-162. This paper reviews the experience at the Surgical Planning Laboratory of developing and applying novel registration and image processing algorithms for capturing intraoperative deformations in support of image guided therapy. Applications to brain and prostate therapy and treatment are presented, showing the diverse range of procedures which can benefit from basic research. New methods which incorporate diffusion tensor imaging and bias field correction into the registration process, as well as novel methods for validation of segmentation, are described.

24 ©2005 Surgical Planning Laboratory, ARR Slide 24 Program Interaction Radiation Oncology –Seed counting post brachytherapy –Bladder registration for dosimetry –Pre to Post therapy CT and MR registration (e.g. NVB dosimetry, A. Szot R25 Fellow) –MR to CT registration for radiation therapy planning.

25 ©2005 Surgical Planning Laboratory, ARR Slide 25 Conclusion Registration allows the best of two worlds –Use of high-quality, innovative imaging for targeting and guidance –Use of real-time imaging to guide needle placement

26 ©2005 Surgical Planning Laboratory, ARR Slide 26 Conclusion Registration allows the best of two worlds –Use of high-quality, innovative imaging for targeting and guidance –Use of real-time imaging to guide needle placement Registration is practical for operating room use –Regular part of our MR-guided biopsy procedures –In-bore display system can provide integrated visual feedback to the doctor.

27 ©2005 Surgical Planning Laboratory, ARR Slide 27 Future Work Migration to higher-field strength, closed bore intraoperative systems (U41). Patient motion and needle placement accuracy (integrate with robotics). Continued research into new imaging modalities (Tensor imaging, gadolinium studies, spectroscopy, SSMs, etc.) Extend collaborations with physicists (XRT) and other researchers. External collaborations (Davatzikos, UPenn) Other prostate therapies (Focused Ultrasound).


Download ppt "©2005 Surgical Planning Laboratory, ARR Slide 1 Prostate Image Processing Steven Haker, PhD."

Similar presentations


Ads by Google