Microwave imaging for breast cancer detection Sara Salvador Summary Epidemiology of breast cancer The early diagnosis: State of the art Microwave imaging: The signal processing algorithm 2D/3D models and results Future work why it is so important outlook basic principles advantages 1
Epidemiology of breast cancer Breast tumour is nowadays the most common malignancy in women of the western world and it is the second leading cause of cancer mortality, following lung cancer. 1 000 000 cases 400 000 deaths 30 000 cases 11 000 deaths Why is early diagnosis so important? The probability of surviving decreases very rapidly as the tumour develops. Early-stage diagnosis is the principal instrument to face this pathology. 5-years survival rate (%) tumor stage at the moment of the diagnosis Source: american cancer society ( www.cancer.org ) 2
Outlook over early diagnosis The most common and efficient diagnostic method is mammography, which has remarkably reduced the mortality rate of breast cancer, BUT: it suffers from a high false negative (10-30%) and false positive (10%) rate; it is often unable to distinguish between malignant and benignant lesions; its efficiency decreases when it is applied to radiographically dense tissues; it is uncomfortable and painful for patients; it is based on the use of ionizing radiations. State of the art Microwave imaging for breast cancer detection was proposed for the first time in 1998 by an international group of researchers [1]. It is a pretty new topic which keeps evolving. What has been done up to now can be resumed as [2]: numerical analysis in 2D and 3D models experimental analysis on simple phantoms [1] S.C.Hagness, A.Taflove, J.E.Bridges, Two-dimensional FDTD analysis of a pulsed microwave confocal system for breast cancer detection: fixed focus and antenna array sensors, IEEE Trans. Biomed. Eng. 1998;45(12):1470-1479 [2] E.C.Fear, Microwave imaging of the breast, Technology in cancer and research treatment 2005; 4(1):69-82 3
Basic principles of µwave imaging (1) 55 50 45 40 35 30 25 20 15 10 Normal and tumoral breast tissues have a very high contrast (1:2-1:10) in electromagnetic properties (ε and σ) in the microwave frequency range (300MHz-30GHz). Relative permittivity according to Debye s model Contrast 5:1 5 0 1 2 3 4 5 6 7 8 9 f (GHz) 10 σ (S/m) f (GHz) normal tissue tumour Electric conducibility according to Debye s model Note that biological tissues are dispersive mediums, i.e. their dielectric properties vary with the frequency. 10 9 8 7 6 5 4 3 2 Contrast 12:1 1 Contrast 5:1 0 0 1 2 3 4 5 6 7 8 9 10 Basic principles of µwave imaging (2) Waves propagating through a medium are partially or totally reflected every time they encounter a discontinuity in dielectric properties. By properly analyzing the reflections that a wave undergoes while propagating in the structure under exam, it is possible to locate any scattering element. excitation t transmitting/receiving antenna material 1 transmitting/receiving antenna material 2 ds material 1 signal collected by the antenna t signal collected by the antenna dt t ds dt = 2 v 4
Basic principles of µwave imaging (3) An antenna, located in proximity of the breast, transmits a wave and collects the reflected signal; the antenna is moved to different positions and in each location a new acquisition is performed. dt 1 A 2 dt 2 A 2 ds 1 ds 2 A 1 dt 3 A 3 ds i dti v = 2 prop A 1 ds 3 A 3 Basic principles of µwave imaging (4) The examination apparatus can be represented as follows: excitation signal received signal The patient lies in prone position with her breast extending through a hole in the examination table. The antenna is moved to different positions around the breast. 5
Advantages of microwave imaging semplicity (mobile antenna + network analyzer + PC); velocity (data are collected and their elaboration is performed in post-processing); comfort for the patient (no breast compression); it is based on the use of non-ionizing radiations (no radiological risk and low cost). The post processing algorithm The signals must be elaborated in order to reduce the disturbing elements (skin reflection, noise, clutter due to the dishomogeneity of tissues ) and highlight the response of the tumour. At each point of the domain (breast) is assigned a value of intensity, obtained by summing the informations collected in the different positions of the antenna. S n A n d P,An P τ P, An 2 = signal collected in the n-th position of the antenna x τ P,An d P, An t v N antennas I( P) = S n ( τ P, An) n= 1 4 6
Numerical simulations on a 2D model tumour (ε s =50,ε =4,σ=0,7 S/m, τ = 6,4 ps) source point coupling medium ( = 10) ~ ε ε s ε σ s ε + j 1+ jωτ ωε = 0 skin layer (2 mm) ( =34, σ = 0,4 S/m) mammary tissue (ε s =10,ε =7,σ= 0,15 S/m, τ = 6,4 ps). Dishomogeneity of ε ±10%. 7,8 cm 10 cm (absorbing condition on the domain s boundary) domain s boundary 2D results 7
3D simulations the antenna (1) Beside the breast structure, the 3D model includes the transmitting/receiving antenna. The project of the antenna was one of the main features of our work. The antenna requirements are the following: ultrawideband (1-8 GHz) good directivity small dimension We choosed the Vivaldi antenna. 3D simulations the antenna (2) Substrate =10 5.7x5.3 cm Substrate =10 7.04x7.8 cm 8
Numerical simulations on a 3D model The model is far more realistic than the 2D one: it includes breast tissue, glands, skin and the nipple. skin =31, σ=4 S/m nipple =45, σ=5 S/m gland =12, σ=0.45 S/m gland =14, σ=0.42 S/m normal tissue =10, σ=0.4 S/m gland =13, σ=0.48 S/m 3D results (1) Planar map, 1cm tumour Planar map, 6mm tumour planar scanning Planar map, 1cm tumour 9
3D results (2) sagittal scanning planar scanning Planar map, 1cm tumour Sagittal map, 1cm tumour Future work Improvements of the post-processing algorithm; Construction of the antenna; Construction of breast simulant phantoms and measurements; Measurements on real patients. 10
Thank you for your attention sara.salvador@delen.polito.it 11