Implementation of Active Microwave Imaging Methods to Improve Early Signs of Breast Cancer Detection
Summary: Results, Errors, Discussion and Conclusion
III. RESULTS
The corn syrup medium used has a dielectric permittivity of 18.7 and conductivity of 0.64 S/m at 3000 MHz. Dielectric properties of the breast tissue sample are measured using cavity perturbation techniques and are then compared with that of corn syrup at 3000 MHz to check the compatibility of the corn syrup and breast tissues. The measured dielectric parameters of breast tissues match with the literature data in [1] and [30]. Adequate resolution is achieved as the dielectric permittivity of corn syrup matches with that of the normal breast tissue as seen in Table 2.
 Figure 1 |
Table 1. Dielectric parameters of breast tissue samples and corn syrup at 3000 MHz [13].
The reconstructed 2D tomographic images for the samples 1-4 are shown in Figures 7-10. The dielectric contrast of the samples is clearly distinguishable from the images as well as from the permittivity profiles. Samples 1-3 are having approximately circular cross sections without any cover, where sample 4 is covered in a thin conical polythene paper. This is done to check whether the shape of the sample is reconstructed properly.
As the dielectric permittivity of the coupling medium and normal breast tissue samples are in good match, the scattered tumor inclusions are clearly visible. A resolution of 2 mm is achieved during this reconstruction with the use of corn syrup as the coupling medium. Comparisons of the obtained permittivity values of the breast samples from Figure 7-10, with that measured using cavity perturbation techniques report in Table 2 shows respectable agreement. Also, when we compare the dielectric values and conductivity values of the extrapolation of researches shown in Figure 1, they show good agreement.
 Figure 2 |
Figure 1. Sample 1 (a) 2D microwave tomographic image (b) Dielectric permittivity profile [13].
 Figure 3 |
Figure 2. Sample 2 (a) 2D microwave tomographic image (b) Dielectric permittivity profile [13].
 Figure 4 |
Figure 3. Sample 3 (a) 2D microwave tomographic image (b) Dielectric permittivity profile [13].
 Figure 5 |
Figure 4. Sample 4 (a) 2D microwave tomographic image (b) Dielectric permittivity profile [13].
The time domain response of the breast tissues in samples 1-4 are shown in Figures 10-13. In the figures, the first and last peaks in the encircled region represent reflections from the corn syrup – normal tissue interface. As the dielectric is less here, the reflections are less compared to the other reflections produced by cancerous tissues.
 Figure 6 |
Figure 5. Time domain response of sample 1 – single cancerous tissue of approximate radius 0.5 cm inserted in normal tissue of approximate radius of 1 cm [13].
 Figure 7 |
Figure 6. Time domain response of sample 2 – four cancerous tissue of approximate radius 0.25 cm inserted in normal tissue of approximate radius of 1 cm [13].
Figure 7. Time domain response of sample 3 – four cancerous tissue of approximate radius 0.25 cm inserted in normal tissue of approximate radius of 1 cm [13].
 Figure 8 |
Figure 8. Time domain response of sample 4 – scattered tumorous inclusions of approximate radii 0.1 cm inserted in normal tissue of approximate radius of 1 cm [13].
Other tall peaks correspond to the reflections from the tumor inclusions. In Figure 10, the experimental and FDTD results show good agreement as there is only a single tumor inclusion in sample 1. The experimental and FTDT results do not agree well with Figures 11-13 due to the presence of multiple inclusions. Reflections from nearby contrast points to overlap and get represented as a single point. The time-shift-and-add algorithm applied to the experimental data makes the reflected singals from tumors located opposite to each other to overlap. Even though the exact tumor locations are difficult to find, the regions of dielectric contrast are still satisfactorily detected using this time domain confocal microwave imaging method [13].
Approximate tumor locations with respect to the tallest peak in the figures are calculated from the equation for velocity propagation. The velocity of propagation depends on the dielectric permittivity of the medium given by:
where d is the distance, t is the time taken for propagation,
ε r ε r size 12{ε rSub { size 8{r} } } {}
 Figure 9 |
Table 2. Approximate tumor locations with respect to the first tumorous inclusion [13].
IV. SOURCES OF ERRORS
Special care was taken to eliminate all possible types of errors that could be present. System 2 utilizes a HP 8510 network analyzer that has an accuracy of 0.001 dB for power measurement, 1 Hz for frequency measurement, and 0.01 ns for time domain measurements [31].
The main sources of experimental error are possibly from [13]:
- Edge reflections from the antenna: The FDTD computed end-reflections that are observed at the feed point of the bowtie antenna relative to the excited pulse is -24 dB. For the CMT a time gating of 9.07 ns is provided from the network analyzer to remove these reflections.
- Reflections from the sample holder: The tissue samples are supported on a low loss PVC pipe having loss tangent (tanδ) of 0.002. Hence the reflections here are negligible.
- Medium reflections: Since the tumor is studied using a low contrast matching coupling medium, reflections are minimized and good resolution of the image can be ensured.
- Validity of distorted Born approximation to linearize the inverse scattering problem: This method is adopted to reduce the computational complexity. This matter still needs to be further investigated with strong scatters and fast forward iterative solver.
- Convergence: To ensure that global convergence is achieved, the experiment performed 5 iterations and the same profile as with the 4th iteration was obtained.
V. DISCUSSIONS AND CONCLUSIONS
Current ongoing measurements of electrical properties of normal breast tissue and typical abnormalities will provide a solid foundation for future designs of microwave systems based on emerging technologies. There are still gigantic amounts of research and development that can be done to reach higher expectations of performance. Some examples include better antenna designs, better coupling systems and mediums, better data acquisition systems, hardware design, signal processing methods, and other improvements. Also, with the hopes of progress in numerical methods and increase computational power with decreased computational cost, these methods can only improve.
Altogether, based on the information presented in this investigation, we explored the potential usage of active microwave imaging as an imaging modality for early detection of breast cancer. The main techniques in active microwave imaging involve two main methods: 2D microwave tomographic imaging and confocal microwave imaging (CMI). In vitro studies, we find that using the microwave tomographic imaging method can satisfactorily image normal and malignant tissues by showing a clear distinction in terms of dielectric permittivity. In using the CMI method the location of the tumor could be satisfactorily detected as the strength of the reflected signals in the time domain varies with the dielectric contrast. So the main goal of CMI was to find the locations of dielectric contrast and this was achieved. Overall, there are good reasons to believe that active microwave imaging can be considered for early breast cancer detection.
More about this content: Metadata | Version History | Cite This Content
This work is licensed by david thonglyvong. See the Creative Commons License about permission to reuse this material.
Last edited by david thonglyvong on Dec 16, 2007 1:58 pm US/Central.