Dosimetry as a Function of Anatomy and Exposure Source

Dosimetry as a Function of Anatomy

and Exposure Source

Andreas Christ, Marie-Christine Gosselin,

Marcel Zefferer, Stefan Benkler, Eugenia Cabot,

Sven Kühn and Niels Kuster

Foundation for Research on Information Technologies in Society, Swiss Federal Institute of Technology, Zürich

• review fundamental dosimetric quantities

• discuss characteristics of sources of RF and ELF electromagnetic fields and their numerical representation

• discuss impact of anatomical properties and incident field conditions on the absorption in the body

• present the latest developments in human body modeling

Dosimetry vs. Incident Exposure

• Human exposure assessment is the discipline of dosimetry. • Incident field exposure assessment is only a proxy of human

exposure since its correlation with induced field values is poor in the best case.

• However, incident field exposure is sufficient to demonstrate

compliance because it is easy to measure and the following can be derived:

X V/m ---> <Y W/kg anywhere in the body X A/m ---> <Z W/kg anywhere in the body

• In general, compliance is only achieved if both values, i.e., the electric and mangetic fields, are below the derived limits.

• Therefore, limits based on power density should not be used in the standards since it is only valid in the special case of plane-wave exposure.

-> in summary: incident exposure quantities have little correlation with dosimetry but are suitable for compliance testing if E- and H-fields are considered.

Meaningful Quantities

• whole-body absorption (to protect from unwanted thermal loads) • locally induced temperature increases (to protect from locally induced

unwanted thermal bioresponses)

• locally induced field values (to protect from possible athermal and non-thermal field effects)

• SAR

-> in summary: standards are rather unclear about the rationale of the selected quantities.

SAR Averaging Volume

• The averaging of the peak spatial SAR (psSAR) in a cubical volume is a historical selection and has NO physical meaning. It is greatly direction dependent and therefore difficult/time consuming to

determine (to our knowledge nobody has ever evaluated the maximum cube values).

• contiguous tissue is a large improvements over the cube (unique definition) but is more conservative and may overestimate the thermal correlations.

-> in summary: the concept contiguous tissue is superior in terms of non-ambiguous definition of the assessment but has not been enforced yet.

Characteristics of EM Field Sources

• far-field like RF sources • near-field RF sources

electric field _{magnetic field}

Far-Field Like RF Sources

0 -10 -15 -20 -25 -30 Normalized Field S trength in dB -5

• measured E- and H-fields in front of a base station antenna at 1840MHz

• geometry or distance to body > approx. λ/2

• directed power flux (wave vector k is real), E and H correlated by Z (depending on dimensions of the

source)

• incident field can be very inhomogeneous unless distance -> infinity

• negligible impact of loading (back scattering) of the

exposed subject on the source

Near-Field RF Sources

• geometry and distance < λ/2

• little or no directional power flux (reactive fields, wave vector is complex or

imaginary)

• complex field distribution, correlation of E and H not obvious

• generally strong interaction between source and

exposed subject

• exposure of the user of a cell phone and a bystander

0 -20 -40 -60 -80 -100

Quasistatic Approximation of ELF Fields

• geometry and distance << λ • no power flux

• no correlation of E and H (generally only either

component present in the source)

• new electro- and magneto-quasi-static solvers (FE-based) for dosimetric applications have been developed

• current density in a child model in front of generic induction hob

0 -10 -20 -30 -40 -50

• Far-field exposure is not necessarily plane wave exposure. The numerical evaluation of far-field and far-field-like exposure should be based on validated models or consider a worst-case set of

incident fields.

• For the simulation of near-field exposure, the source model must correctly reproduce the effect of loading the reactive field with the exposed subject.

• A novel solver for dosimetric ELF simulations has been developed. It can handle anatomical models of more than 10 million cells.

• The whole body SAR is a rather robust quantity. Comparatively simple anatomical models yield consistent results depending on the

ratio of the body cross section and the mass. For the correct

rendering of the absorption in the skin, the resolution should be

better than 2mm.

Whole Body Exposure from a Base Station Antenna

• field distribution in the direction of the main beam in front of a base station antenna (1.9m height, 900MHz, 17dBi directivity) • strong concentration of the fields in the center of the antenna

even at 4m 0.5m 1.0m 2.0m 4.0m Radom Patches Feedpoint Ground

Permissible Antenna Power for Human Models

distance = 0.5m

• Different anatomical body models (male, female, 55kg 101kg, 1.60m

-1.80m) have been complemented by scaled models.

• The maximum permissible power decreases monotonically with the ratio of

the exposed cross section to the mass of the body (CSM).

5 5.5 6 6.5 7 7.5 8 8.5 9 9.5 10 x 10- 3 0 50 100 150 200 250 300 350 400 Power (W)

Cross- section on mass ratio (m2/kg)

450 MHz 900 MHz 1800 MHz 2140 MHz

• The whole body SAR is a rather robust quantity. Comparatively simple anatomical models yield consistent results depending on the ratio of the body cross section and the mass. For the correct

rendering of the absorption in the skin, the resolution should not exceed 2mm.

• For local and far-field like exposure of larger sections of the body,

the maximum psSAR depends on partial body resonances and the

tissue layering. They vary significantly among individuals and depend

on the posture of the body.

Typical Locations of the Peak Spatial Average SAR

Local SAR at 900MHz: Impact of Penis Length

Ratio 10g SAR in penis to wb SAR

Penis length (mm)

original length

original

length=65mm

length=125mm

• increase of the 10g SAR in the penis by

approx. 1.8dB due to partial body resonance in the penis (900MHz)

• generic phone with integrated antenna at 50mm distance in front of abdomen

• tissues: skin (2mm - 3.5mm), fat (7mm - 15mm), muscle

fat muscle skin 50mm generic phone

SAR Distribution in the Abdomen for the Generic Phone

• SAR distribution at 1800 MHz normalized to output power • av. SAR ratio (anat. / hom.) of 1.2dB (1g) and 0.6dB (10g)

• skin thickness: 3.3mm, fat thickness: 9mm (approximate values)

0 -5 -10 -15 -20 -25 Normalized SAR [ dB] anatomical _homogeneous

• The whole body SAR is a rather robust quantity. Comparatively simple anatomical models yield consistent results depending on the ratio of the body cross section and the mass. For the correct

rendering of the absorption in the skin, the resolution should not exceed 2mm.

• For local and far-field like exposure of larger sections of the body, the maximum psSAR depends on partial body resonances and the tissue layering. They vary significantly among individuals and depend on the posture of the body.

• The impact of age dependent changes of tissue properties has

been a long term issue of discussion. Recent data show that the

effects are small. Age dependent changes of the anatomy can have

a large impact on the exposure of particular organs.

Assessment of Age Dependencies on SAR

• 3 anatomical head models: Visible Human (VH), 3 and 7 year old

children (3YC, 7YC)

• generic dual band phone (900MHz and 1800MHz) with internal antenna • touch and tilted positioning [Kainz

et al., 2005, PMB, 50(14): 3423-3445]

• assessment of Peak Spatial Av. SAR for the head without pinna tissue and for brain tissue

• comparison of 10g av. SAR without pinna to SAM phantom (IEEE 1528)

10g Average SAR at 900MHz in the Touch Position

Visible Human 3 year old child 7 year old child

SAR Ratio Anatomical vs. SAM Cole-Cole 10kg 50kg 250kg -10 -15 -20 -5 0 -25 ADD COLORBAR Normalized SAR in dB

10g Average SAR at 1800MHz in the Touch Position

Visible Human 3 year old child 7 year old child

SAR Ratio Anatomical vs. SAM 10kg 50kg 250kg -10 -15 -20 -5 0 -25 ADD COLORBAR

Exposure of the Brain at 1800MHz

• SAR distribution at 1800MHz in brain tissue

• cube location at maximum 1g Peak Spatial Average SAR

Visible Human 3 Year Old Child

-10 -15 -20 0 -5 -25

Peak Spatial Average SAR in the Brain at 1800MHz

• SAR maximum located in cerebellum of children

• current density maximum of the phone at the antenna

• strong increase of SAR in children because SAR

maximum is directly located at current maximum

• no impact on compliance testing but significant for epidemiological studies 0% 50% 100% 150% 200% 250% 300% 350% 400% 450% 500%

Touch Tilted Touch Tilted

1g 10g

A

v

. SAR Ratio Children vs. VH

3 year old child 7 year old child

Conclusions: Dependence of EM Absorption on Anatomy

• The whole body average SAR is a robust quantity which can be correlated with the cross section to mass ratio. The grid resolution should be better than 2mm for convergent results.

• Partial body resonances and standing wave effects in layered tissues depend on individual properties of the body model. They can lead to variations in local SAR of more than 3dB. Generally, a large set of different models and postures is necessary to completely assess such effects. This should be complemented by generic body models. • The effect of age dependent changes of dielectric tissue parameters

on the psSAR is not larger than other intersubject variations (less than ±30%). Additional differences can occur due to different pinna thicknesses.

• Anatomical changes due to growth of the skull can lead to significant differences of the exposure of certain brain regions (above 6dB, depending on phone design).

Tasks for Conservative Human Exposure Assessment

• assessment of the operational conditions resulting in the maximum radiated power

• determination of conservative position/max incident field distribution • determination if the scatterer (human body) may affect the source

resulting in higher exposure

• determination of conservative posture

• determination of conservative tissue composition • determination of conservative tissue parameters

• development of anatomical models of adults and children using novel techniques for segmentation and CAD modeling

• development of a pregnant woman model

• identification of functional brain regions (Talairach Transformation) • poser software to change the posture of the models’ limbs

The Virtual Family

• adult male: 34 years, 1.74m, 70kg

• adult female: 26 years, 1.60m, 58kg

• girl: 11 years, 1.48m, 34kg

• boy: 6 years, 1.07m, 17kg

• 4 more children under development

• using the AMIRA 4.0 software

• surface reconstruction (marching cube), surface smoothing (spring model), reduction of complexity (triangle collapse)

• 84 different tissue types (CAD objects)

• adjacent CAD objects share common surfaces

• export of organs and tissues as watertight CAD parts in SAT format

Construction of CAD Objects

Development of a Pregnant Woman Model

• MR scans of the torso of a

pregnant woman (7th

gesta-tional month)

• resolution 0.94 x 0.94 x

4.4mm3,

• coronal and axial scans avaliable

Exposure Evaluation of Functional Brain Regions

• The Talairach Coordinate System defines a labeling hierarchy and coordinates of functional regions of the brain.

• The coordinate system is defined by the anterior and posterior commissurae and the bounding box of the brain.

• The outline of the brains of the anatomical models are normalized (rotated, scaled, warped) to match the Talairach coordinates.

axial orientation coronal orientation _{saggital orientation}

AC PC AC/PC AC PC S I S I L R L R A P A _P

Talairach Transformation in SEMCAD X

• The Talairach Coordinate System are identified in all anatomical models.

• A dialog in the post processor of SEMCAD X will allow the selection of brain regions on the hierarchical levels defined in by Talairach. • All SAR extraction functions of SEMCAD X will be avaliable for the

selected regions (statistics, averaging, visualization, etc.)

5. Cell Type 4. Tissue Type 3. Gyrus 2. Lobe 1. Hemisphere Left Cerebrum Right Cerebrum Left Cerebellum Right Cerebellum Left Brainstem Right Brainstem Inter-Hemispheric * Anterior Lobe Frontal Lobe Frontal-Temp. Space Limbic Lobe Medulla Midbrain Occipital Lobe Parietal Lobe Pons Posterior Lobe Sub-lobar Temporal Lobe *

Middle Occ. Gyrus

Middle Temp. Gyrus

Nodule Orbital Gyrus Precentral Gyrus Precuneus Pyramis Pyramis of Vermis Sup. Frontal Gyrus Sup. Occipital Gyrus Sup. Parietal Lobule

Sup. Temporal Gyrus

Thalamus CSF Gray Matter White Matter * Brodmann area 37 Brodmann area 38 Caudate Body Caudate Head Caudate Tail Corpus Callosum Dentate Hippocampus Hypothalamus Lat.Dorsal Nucleus Lat. Geniculum Body Lat. Globus Pallidus Lat. Post. Nucleus

Outlook: Poser Software for Anatomical Models

• objective: anatomically correct repositioning of limbs and joints

• bending of skeleton, recalculation of soft tissues under approximate conservation of organ masses

• applicable to all virtual family models and future anatomical CAD models

Summary and Conclusions

• Human exposure assessment is a discipline of dosimetry; the correlation with incident exposure assessments is not straightforward.

• The evaluation of different exposure situations has shown that the induced SAR can vary by more than 3dB dB depending on individual anatomical features.

• For the correlation of the incident fields with human exposure, the coupling mechanism of the particular exposure situations must be understood and the induced fields must be quantified considering the anatomical variability of the exposed group of the population.

• Significant progress has been made with respect to the accuracy of the modeling of the human body. Ongoing developments include the modeling of a pregnant women model, an obese adult model and a novel software to change the posture of the CAD models.