Diffusion MRI Tractography for Improved MRI-guided Focused Ultrasound Thalamotomy Targeting for Essential Tremor

Transcription

1 Diffusion MRI Tractography for Improved MRI-guided Focused Ultrasound Thalamotomy Targeting for Essential Tremor Qiyuan Tian 1,2, Max Wintermark 2, Kim Butts Pauly 2, Pejman Ghanouni 2, Michael Zeineh 2, Casey H. Halpern 3, Diane S. Huss 4, W. Jeffrey Elias 5, Jennifer A. McNab 2 Departments of Electrical Engineering 1, Radiology 2 and Neurosurgery 3, Stanford University Departments of Physical Therapy 4 and Neurosurgery 5, University of Virginia

2 Declaration of Financial Interests or Relationships Speaker Name: Qiyuan Tian I have the following financial interest or relationship(s) to disclose with regard to the subject matter of this presentation: Company Names: GE Healthcare, InSightec Type of Relationship: Research Support 2

3 MR-guided focused ultrasound treatment of essential tremor Essential tremor (ET) Common neurological disorder that causes a rhythmic shaking Tremor suppression 1 Lesion ventral intermediate (VIM) nucleus of the thalamus MR-guided focused ultrasound (MRgFUS) Apply a high-intensity focused ultrasound beam to thermally ablate targeted tissue Axial section of thalamus (Schaltenbrand-Wahren Atlas 2 ) [1] Zesiewicz et al. Neurology 2005;64(12): [2] Schaltenbrand and Wahren,

4 Targeting for MRgFUS treatment of essential tremor Challenges Low intrinsic contrast between thalamic nuclei on standard structural MRI Small size Current methods 1 Start from measurements from the r anterior commissure-posterior commissure (AC-PC) line Adjust targeting and confirm in awake patients that hand tremor is suppressed Anatomic measurements on T2-weighted MRI 1 [1] Ghanouni et al. AJR. 2015;205(1):

5 Limitations of current targeting method 5 Does not account for individual anatomical differences Lacks specificity, i.e. the hand representation Homuncular representation in Vim

6 Diffusion tractography to improve targeting Thalamus and cortex are connected by thalamocortical radiations Target the hand representation in thalamus by mapping structural connections between hand-knob motor cortex and thalamus VIM Homuncular organization in cortex 6

7 A retrospective study Pre-treatment tractography data Post-treatment lesion location Thalamus Hand-knob Probabilistic tractography Identify hand representation in thalamus using diffusion tractography Compare post-treatment lesion location with identified hand representation Correlate probabilistic streamline counts in lesion with clinical outcomes A complementary approach to current targeting methods? 7

8 Data University of Virginia 13 subjects 1 Stanford University 7 subjects Diffusion pre-treatment 20 dir. of s/mm mm 3, 1.2mm spacing 7 min Diffusion pre-treatment s/mm mm 3, no spacing 10 min T1-weighted MPRAGE 1 week post-treatment mm 3, 0.45 mm spacing T2-weighted CUBE 1 year post-treatment mm 3, 0.5 mm spacing [1] Elias et al. N Engl J Med 2013;369:

9 Region of interests (ROIs) 9 Drawn by neuroradiologist M.W. Pre-treatment for tractography Post-treatment lesion location Thalamus Hand-knob Lesion

10 Tractography Methods Probabilistic tractography (FSL s probtrackx) Voxel-wise crossing fiber orientation distribution estimation (FSL s bedpostx) Eddy current correction, motion correction with co-registration (FSL s eddy) Thresholded and volume rendered probabilistic tractography Probabilistic streamline counts for thalamic voxels Thalamus Hand-knob Tractography Thalamus % of thalamushand-knob streamlines 3% 0.3% 10

11 Results: tractography Virginia data Stanford data Thalamus Hand-knob Thresholded and volume rendered probabilistic tractography 11

12 Results: probabilistic streamline counts 3% 0.3% Virginia data Schaltenbrand-Wahren Atlas 1 Stanford data Thalamus Normalized and thresholded probabilistic streamline counts [1] Schaltenbrand and Wahren, % 0.1% 12

13 Co-registration of pre- and post-treatment data Pre-treatment probabilistic streamline counts Post-treatment lesion location 3% Registration Registration Virginia data: boundary based registration from diffusion B0 to T1w Stanford data: nonlinear registration from diffusion B0 to T2w Statistics 0.3% Total probabilistic streamline counts within the lesion 13

14 CRST A&B for treated hand CRST total for whole body Clinical rating scale for tremor (CRST) 1 Part A ~ tremor, part B ~ task, part C ~ disability Lower scores indicates less tremor and better outcome CRST A&B for treated hand CRST Total (A&B&C) for whole body Baseline Month 3 Year 1 Baseline Month 3 Year 1 [1] Fahn et al.,

15 Virginia Results: Tractography vs. CRST scores CRST A&B (treated hand) 3 month r 2 = p= year r 2 =0.112 p=0.263 Probabilistic streamline count in lesion r 2 = p=0.545 Probabilistic streamline count in lesion r 2 = p=0.402 CRST Total (A&B&C, whole body) Probabilistic streamline count in lesion Probabilistic streamline count in lesion 15

16 Stanford Results: Tractography vs. CRST scores CRST A&B (treated hand) 3 month 1 year r 2 =0.536 p= r 2 =0.477 p= Probabilistic streamline count in lesion r 2 = p=0.5 Probabilistic streamline count in lesion r 2 = p=0.652 CRST Total (A&B&C, whole body) Probabilistic streamline count in lesion Probabilistic streamline count in lesion 16

17 Summary patients across two institutions Tractography between thalamus and hand-knob was consistent from subject-to-subject and followed expected anatomy Tractography location in the thalamus was consistent with atlas Higher streamline counts in the MRgFUS lesion correlated with better treatment outcome The most significant correlation was found for scores that considered only the treated hand

18 Future work 18 Acquire data on more patients Account for the lesion thoroughness/permanence Conclusion Diffusion tractography based targeting is a potential complementary approach to current MRgFUS targeting methods

19 19 Acknowledgements Thanks for your attention! McNab Lab Christoph Leuze Manuela Vasquez Jennifer McNab Grant Yang Anne Sawyer, Thomas Brosnan, Gwenaëlle Douaud, Saad Jbabdi P41-EB015891, S10-RR026351