Implementation of pathologies in the Monte Carlo model in chest and breast imaging

(1)

Institutionen för medicin och vård

Avdelningen för radiofysik

Hälsouniversitetet

Implementation of pathologies in

the Monte Carlo model in chest and

breast imaging

Gustav Ullman, Michael Sandborg, Roger Hunt,

David R Dance and Gudrun Alm Carlsson

Department of Medicine and Care

Radio Physics

(2)

Series: Report / Institutionen för radiologi, Universitetet i Linköping; 94

ISRN: LIU-RAD-R-094

Publishing year: 2003

(3)

Report 94

Dec. 2003 ISRN ULI-RAD-R--94--SE

Implementation of pathologies in the Monte Carlo model in chest and breast imaging

G Ullman1_{, M Sandborg}1_{, Roger Hunt}2_{, D R Dance}2_{and G Alm}

Carlsson1

1. Department of Radiation Physics, Linköping University 2. Department of Physics, The Royal Marsden NHS Trust

Full addresses:

1. Department of Radiation Physics, IMV, Faculty of Health Sciences,

Linköping University, SE-581 85 LINKÖPING,

Sweden

2. Joint Department of Physics, The Royal Marsden NHS Trust and Institute of Cancer Research, Fulham Road,

London SW3 6JJ,

(4)

Introduction

The Monte Carlo simulation model of the chest and breast imaging systems including a voxelised model of the patient are used to compute measures of image quality and patient absorbed dose. It is important that the model computes measures of image quality of pathological details that are similar in size, composition and position as real pathological details in typical chest and breast images. Moreover, the other partners of the co-coordinated research project will produce hybrid images with pathological details and have these images assessed by a group of radiologist. The model will then be used to study to what extent variations in clinical image quality can be explained by variations in physical image quality, for example signal-to-noise ratio. This report

summarizes the selection of pathological details to include in the model of chest and breast imaging systems.

Method

Chest pathologies

A radiologist was interviewed and asked to describe the most important

pathological details in chest imaging. Four pathological structure were identified by the radiologist namely: metastasis, calcifications, Kerley B-lines and fibrosis. In Pettersson 1995, the following description is found:

Metastasis

‘Pulmonary metastases may be solitary or multiple. They are often found with a smooth surface. When multiple, the size differs from one lesion to the other. Most metastases appear near the surface of an individual lobe of the lung and can appear in most part of the lungs.’

Calcification

‘The most common anaerobic infection is tuberculosis. The consolidated lung tissue shrinks so that the remaining lesion is a pulmonary nodule, which may become calcified at a later stage. The peripheral pulmonary calcification is a typical finding in primary tuberculosis.’

Kerley B-lines

‘Neither the secondary lobules nor the interlobular septa are seen in radiographs of normal patients. In patients with venous congestion or lymphatic obstruction with accompanying interstitial edema, the septa become visible as thin, one or two centimeter long stripes perpendicular to the pleura, at the base of the lung. These are Kerley B-lines which may be due to venous congestion, lymphatic obstruction. The lines are therefore fluid in the interalveolar septa.’

Fibrosis

‘Fibrosis is an inflammatory reaction in the endothelial cells in the alveoli. As the disease progresses, there is destruction of the alveolar architecture. There is gradual development of fine reticulonodular opacities in the basal lung segments. The reticular opacities become steadily courser and appear as annular shadows which represents cystic cavities in the lung, giving these a typical honeycomb appearance.’

(5)

Notation

The following notation was used to differentiate between the different chest pathologies. The positions are described with a four letter combination given below. Additional letters in parenthesis is sometimes used to clarify critical positions for example behind the spine or heart.

1st_letter M: Metastasis C: Calcification K: Kerley B-lines F: Fibrosis 2nd_letter R: Right lung L: Left lung 3rd_{letter (longitudal)} W: Lower lung C: Central lung U: Upper lung 4th_{letter (lateral)} P: Peripheral lung C: Central lung D: Median lung In parenthesis H: Behind heart S: Behind spine B: Behind rib

For example MLWD(HS) would mean a metastasis in the left, lower median lung, behind heart and spine.

x,y-coordinates

Below the four-letter combination, the exact location is presented in the Zubal voxel phantom with positions defined at the image detector.

Profile

The projected profile is described as a Gaussian distribution. The profile for linear structures such as fibrosis and Kerley B lines is a line with a Gaussian

intersection. The spherical details such as metastasis and calcifications use a Gaussian distribution.

Tissue type

The tissue type is given. First the tissue type used in the Voxman Monte Carlo code is given and then, in parenthesis, the real pathological tissue’s name.

Composition

The composition is given in atomic numbers. (Not in atomic weight). The soft tissue and lung compositions are taken from ICRU (1992).

Background

(6)

10.2183 H, 0.8742 C, 0.2213 N, 4.6814 O, 0.0094 S, 0.0085 Cl Density: 0.2600 g/cm3

Breast pathologies

The most relevant pathological details for mammography are masses and

calcifications. For the simulation of masses in this first stage of development it is only possible to consider well-circumscribed lesions. No attempt will be made to simulate spiculate lesions because of the difficulty of adding them in a realistic way to a normal mammogram. For the purposes of optimisation, spherical masses will be used of three diameters and at three positions in both the fatty and glandular regions of the breast. For the purposes of unification, the

circumscribed lesions will be matched as closely as possible to those used by the Malmö/Göteborg partner for human observer studies, so that the results. The aim is to correlate calculations with the program with the results from the human observer studies. This partner is following the approach of Skiadopolous et al (2003) to add simulated circumscribed lesions to real mammograms. The sizes and contrast of the latter lesions have at the time of writing, yet to be finalized. The smallest calcifications are the most difficult to visualize. The calcification size, 200 microns, has been set to be just above the threshold for visualization. At this size, calcification structure is difficult to visualize and a spherical shape has been chosen for simplicity.

Masses

Size: The masses would be spheres of diameter 3, 5 and 8 mm. Density: 1.058 g/cm3 _{c(from Johns and Yaffe (1987).}

Location: Three locations in both the adipose and glandular regions of the breast. Detailed locations are not given here because of the availability of several breast phantoms.

Chemical composition by weight: H 10.%; C 18.4%; N 3.2%; O 67.7%; Ash (S,P,K,Ca) 0.5%. Assumed to be the same composition as that of the glandular tissue given by Hammerstein et al (1979)

Calcification

Size: Calcification will be 200 micron spheres. Density: 3.3 g/cm3

Location: Three locations in both the adipose and glandular regions of the breast Detailed locations are not given here because of the availability of several breast phantoms.

(7)

Breast composition

The main background breast composition is adipose and glandular tissue but other tissues are also modeled

Adipose tissue

Density: 0.93 g/cm3

Chemical composition by weight: H 11.2%; C 61.9%; N 1.7%; O 25.1%; Ash (S,P,K,Ca) 0.1%. Hammerstein et al (1979)

Glandular tissue

Density: 1.04 g/cm3

Chemical composition by weight: H 10.%; C 18.4%; N 3.2%; O 67.7%; Ash (S,P,K,Ca) 0.5%. Hammerstein et al (1979)

Skin

Density: 1.09 g/cm3

Chemical composition by weight: H 10.0%; C 20.4%; N 4.2%; O 64.5%; Na 0.2%; P 0.1%; S 0.2%; Cl 0.3%; K 0.1%. Adult skin composition from ICRU Report 46 (1992)

Ducts

Density: g/cm3

Density: 1.04 g/cm3

Chemical composition by weight: H 10.%; C 18.4%; N 3.2%; O 67.7%; Ash (S,P,K,Ca) 0.5%. (Glandular composition from Hammerstein et al (1979))

Coopers ligaments

Density: 1.05 g/cm3

Chemical composition by weight: H 10.2%; C 14.3%; N 3.4%; O 71.0%; Na 0.1%; P 0.2%; S 0.3%; Cl 0.1%; K 0.4%. Adult skeletal muscle composition from ICRU Report 46 (1992)

Results

Chest

Metastasis

Position MRCC M01 Position (x,y): (-5, -12) Size: 4 mm spherical.

Tissue type: Soft tissue (cancer)

Composition: 10.4167 H, 2.1314 C, 3.7626 O, 0.0094 S, 0.1928 N, 0.0056 Cl Density: 1.03 g/cm3

Position MRWP M02 Position (x,y): (-10, -6) Size: 4 mm spherical.

(8)

Position MLWD(HS) M03 Position (x,y): (0.3, -7) Size: 10 mm spherical.

Position MLCD(S) M04 Position (x,y): (0, -15) Size: 10 mm spherical.

Position MLWD(H) M05 Position (x,y): (2, -7.5) Size: 10 mm spherical.

Position MLWP M06 Position (x,y): (8.5, -6) Size: 4 mm spherical.

Position MLUP M07 Position (x,y): (6, -17) Size: 4 mm spherical.

Position MRCM M08 Position (x,y): (-3, -14) Size: 4 mm spherical.

Position MRWM(H) M09 Position (x,y): (-2.7, -6) Size: 10 mm spherical.

Position MLCM M10

Position (x,y): (1.3, -14) Size: 4 mm spherical.

(9)

Density: 1.03 g/cm3 Position MRUC M11 Position (x,y): (-5, -22) Size: 4 mm spherical.

Position MLUC M12 Position (x,y): (3, -22) Size: 4 mm spherical.

Fig. 1. The rings shows the positions of the metastasis details. The metastasis details are distributed in the lung apices, in the lung periphery, in the central lung and behind the spine and heart.

Calcifications

Position CRUC C01 Position (x,y): (-5, -22) Size: 0.5 mm spherical. Tissue type: Calcification. Composition: 1H 13O 3P 5Cl Density: 3.30 g/cm3

Position CLUC C02 Position (x,y): (3, -22) Size: 0.5 mm spherical. Tissue type: Calcification. Composition: 1H 13O 3P 5Cl Density: 3.30 g/cm3

(10)

Fig. 2. The rings shows the positions of the calcification details. The calcifications are located in the upper left and right lung apices.

(11)

Kerley B lines

Position KRWP K01 Position (x,y): (-10, -6)

Size: thickness: 1 mm, area: 1 mm x 1 cm. Tissue type: Water (serum)

Composition: 2H 1O Density: 1.00 g/cm3 Position KRCP K02 Position (x,y): (8.5, -6)

Composition: 2H 1O Density: 1.00 g/cm3 Position KLWP K03

Position (x,y): (-9.5, -10)

Composition: 2H 1O Density: 1.00 g/cm3 Position KLCP K04 Position (x,y): (8, -10)

Composition: 2H 1O Density: 1.00 g/cm3

Fig. 3. The rings shows the positions of the Kerley B lines, positioned in the lower periphery parts of the lungs.

(12)

Fibrosis

Position FLWP F01 Position (x,y): (-10,-6)

Size: thickness: 1 mm, area: 1 mm x 1 cm. Tissue type: Scar tissue.

Position FRWP F02 Position (x,y): (8.5,-6)

Position FRUP F03

Position (x,y): (-7,-19.5)

Position FLUP F04

Position (x,y): (5.5,-19.5)

Fig. 4. The rings shows the positions of the fibrosis details located in the periphery parts of the left and right lungs.

(13)

Breast

Figure 5. Geometric model of the breast (57% glandularity).

Figure 6. Simulated image of the breast above.

Acknowledgements

We would like to thank Associate Professor Sven-Göran Fransson for sharing his knowledge about chest pathology with us. This work was supported by a grant from the European Union 5th_{framework program FIGM-CT-2000-0036.}

(14)

References

ICRU 1992. Photon, electron, proton and neutron interaction data for body tissues. International commission on radiation units, ICRP Report 46 (Bethesda, 1992)

Hammerstein, Miller, White, Masterson, Woodard, Laughlin, Absorbed dose in mammography, Radiology, 130: 485-491 (1970)

Johns PC and Yaffe M J (1987) X_ray characterization of normal and neoplastic breast tissues Phy Med Biol 32: 675-695

Pettersson (ed.). A global textbook of radiology. The NICER centennial book 1995 (The NICER Institute, Oslo 1995) ISBN 82-990882-3-2.

Skiadopolous S, Costaridou L et al. 2003. Simulating the mammographic appearance of circumscribed lesions. Eur. Radiol 13:1137-1147