Quantitative Analysis of Static Ventilation Hyperpolarized 3He MR Images

Ajna BorogovacBoston University - College of Engineering

Harvard Medical School - Radiology Department - Brigham and Women’s Hospital

Objectives

• Determine mathematical relationship between intensity of a HP 3He MR image pixel and amount of 3He in the corresponding object voxel

• Determine trachea ventilation

• Develop means of creating specific ventilation profiles of healthy and diseased lungs

• Investigate sensitivity of the ventilation profiles to defect magnitude and size.

Background• Pulmonary Ventilation Disorders

– Asthma• Afflicts 18 million Americans• Causes of airway obstruction:

1.) Bronchospasm 2.) Inflammation of airway lining3.) Sticky mucus secretions

Destroyed Alveoli

Mucus

Inflammation

Collapsed Airway – COPD

• Fourth Leading Cause of Death in U.S.• Causes of airway obstruction:

1.) Destruction and collapse of smaller airways2.) Alveolar wall loss 3.) Thickening of inflamed airways 4.) Sticky mucus secretions

• Pulmonary Imaging Modalities

Background

– Magnetic Resonance Imaging (MRI)

– Computed Tomography (CT)– Positron Emission Tomography (PET)

• Pulmonary Imaging Modalities

Background

– Magnetic Resonance Imaging (MRI)

• Pulmonary Imaging Modalities

Background

– Magnetic Resonance Imaging (MRI)

Magnetic

Field

• Pulmonary Imaging Modalities

Background

– Magnetic Resonance Imaging (MRI)

RF Pulse

Magnetic

Field

• Pulmonary Imaging Modalities

Background

– Magnetic Resonance Imaging (MRI)

RF (MR SIGNAL)

Magnetic

Field

– Magnetic Resonance Imaging• Water based - can’t image lungs

Homogenous signal: healthy ventilation

Heterogenous signal: ventilation defect

Background

– Hyperpolarized 3He MR Imaging• 3He based - enables ventilation studies• Previous Studies: Qualitative analysis of signal distribution

Our Interest

• Development of Quantitative Analysis Methods– Possibility of developing more accurate diagnostic

tools for measurement of ventilation.• Test efficacy of various treatments • Map progress of the ailment by tracking a patient’s

ventilation distribution over time.

MethodsCollect HP 3He MR Images

Pixel Intensity vs. 3He Amount

Healthy Ventilation Profile

Healthy Ventilation Profile with Simulated Defect

Patient Ventilation Profile

MethodsCollect HP 3He MR Images

Pixel Intensity vs. 3He Amount

Healthy Ventilation Profile

Healthy Ventilation Profile with Simulated Defect

Patient Ventilation Profile

MethodsCollect HP 3He MR Images

Pixel Intensity vs. 3He Amount

Healthy Ventilation Profile

Healthy Ventilation Profile with Simulated Defect

Patient Ventilation Profile

• A RMSE-minimizing mathematical fit between pixel intensities and small area increments across tube diameter was found.

MethodsCollect HP 3He MR Images

Pixel Intensity vs. 3He Amount

Healthy Ventilation Profile

Healthy Ventilation Profile with Simulated Defect

Patient Ventilation Profile

MethodsCollect HP 3He MR Images

Pixel Intensity vs. 3He Amount

Healthy Ventilation Profile

Healthy Ventilation Profile with Simulated Defect

Patient Ventilation Profile

• Simulated defects of various radii and strengths across the healthy ventilation HP 3He MR image slices.

• Compared the resulting specific ventilation profiles with the healthy ventilation profile obtained previously.

a.) Homogenous Defect b.) Parabolic Defect:

MethodsCollect HP 3He MR Images

Pixel Intensity vs. 3He Amount

Healthy Ventilation Profile

Healthy Ventilation Profile with Simulated Defect

Patient Ventilation Profile

• The specific ventilation profile for one mild asthmatic was created with the same algorithm as used for healthy lungs.

– One modification: lung boundary has to be user defined where lung edge is affected by a ventilation defect.

Resultant pixels over which ventilation is calculated

Lung boundary prescription

Ventilation pixels located using threshold filtering

Results• Linear relationship is the best mathematical fit between image pixel

intensity and amount of 3He in a corresponding image voxel.

* Representative data for 1.5875 cm diameter tube

Results• Healthy specific ventilation profiles were created.

- Local specific ventilation in central axial locations of lung is steady: fluctuating by no more than 15% from the local mean.

0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Specific Ventilation

Left Lung

Right Lung

Axial Lung Length

0

.2

.4

.6

.8

1

0

.2

.4

.6

.8

1

Spe

cifi

c V

entil

atio

n

Results• Specific ventilation profiles obtained using our methods are not sensitive

enough to detect defects that are too small or too weak.– The overall effect of any defect on specific axial ventilation profile

has at least 15% uncertainty associated with it.

10

Results• HP 3He MRI scan of a patient lung showed small defects along the axial center of the left lung.• The specific ventilation profile of the patient was found to be not sensitive enough to locate these

defects.

0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

0 .1 .2 .3 .4 .5 .6 .7 .8 .9 1

Specific Ventilation

Left Lung

Right Lung

Axial Lung Length

0

.2

.4

.6

.8

1

0

.2

.4

.6

.8

1

Spe

cifi

c V

entil

atio

n

Conclusions• There exists a linear relationship between

intensity of an image pixel and the amount of 3He in a corresponding object voxel.

• Ventilation profile of healthy lung is steady in central axial locations, fluctuating by no more than 15% from the local mean.

• The specific ventilation profiles obtained using our methods are not sensitive enough to detect ventilation defects of too small a size or magnitude.

Acknowledgments• Mitchell Albert, Dr.

• Yang Tzeng Sheng