A Mass Conservation Based Optical Flow Method for Cardiac Motion Correction in 3D PET Data

M Dawood1,2, C Brune2, F Büther1, KP Schäfers1

European Institute for Molecular Imaging Department of Computer Science,

University of Münster, Germany

2

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Cardiac Motion and Partial Volume

Coronal slice through non-attenuated PET1 h. p. i., 18FDG, CHD patient

Cardiac motion

Ungated dataLarge blur, low noise(Problem in plaque imaging)

One phaseSmall blur, high noise

3

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

…

Step 1: Gating

ECG signal

Series of images reconstructed from cardiac gated PET acquisition

4

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Mass conservation in cardiac data

Systole Diastole

Mass Conservation

5

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Step 2: Motion Estimation

6

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

All gates deformed to Diastole

Visual result

Cardiac phases

7

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Ungated dataLarge blur, low noise Noise 25

One phaseSmall blur, high noise Noise 36

All phases motion correctedSmall blur, low noise Noise 22

Visual result

8

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Quantitative results on patient data

Data:

14 patients with known CHDca. 4 MBq/Kg body weight 18F-FDGScan time ca. 15 minutes, 1:15 hours post injectionListmode acquisition on Siemens Biograph 16 scanner

Quatification methods:

Correlation of ROI (40x40x40) with target phaseMyocardial thickness. FWHM of Gaussian fit to line profileMean activity in blood pool in LV

9

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Quantitative results 1: Correlation of end-systolic gate with target phase

1 2 3 4 5 6 7 8 9 10 11 12 13 140.700000000000001

0.750000000000001

0.800000000000001

0.850000000000001

0.900000000000001

0.950000000000001

1

BeforeAfter

Patient No

Corr

elati

on c

oeffi

cien

t

10

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Quantitative results 2: Myocardial thickness

1 2 3 4 5 6 7 8 9 10 11 12 13 14 Avg3

3.5

4

4.5

5

5.5

6

Patient Number

Myo

card

ial t

hick

ness

[mm

]

End-systoleEnd-diastoleEnd-systole after MC

11

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Quantitative results 3: Mean activity in blood pool

12

Motion Correction on 3D PET/CT Data with Optical Flow Algorithms

Thank you

To conclude:

A method for cardiac motion and partial volume correction was presented.

The results on patient data show that the motion was corrected precisely.