Aim of research: This research aimed to design and develop a synthetic anthropomorphic breast phantom with cancer mimicking lesions and use this phantom to assess the relationship between lesion visibility and breast thickness in mammography. Due to the risk of cancer induction associated with the use of ionising radiation on breast tissues, experiments on human breast tissue was not practical. Therefore, a synthetic anthropomorphic breast phantom with cancer mimicking lesions was needed to be designed and developed in order to provide a safe platform to evaluate the relationship between lesion visibility and breast thickness in mammography. Method: As part of this research custom Polyvinyl alcohol (PVAL) breast phantoms with embedded PVAL lesions doped with contrast agent were fabricated and utilised. These breast phantoms exhibited mechanical and X-ray properties which were similar to female breast/breast cancer tissues. In order for this research to be useful for human studies, patient safety factors have constrained the extent of this research. These factors include compression force and radiation dose. After acquiring mammograms of phantoms with varying thicknesses, the image quality of the embedded lesions were evaluated both perceptually and mathematically. The two-alternative forced choice (2AFC) perceptual method was used to evaluate image quality of the lesions. For mathematical evaluation the following methods were utilised: line profile analysis, contrast-to noise ratio (CNR), signal-to noise ratio (SNR) and figure of merit (FOM). Results: The results of the visual perception analysis of the mammograms demonstrate that as breast compressed thickness reduces the image quality increases. Additionally, the results display a correlation in the reduction in the level of noise with the reduction in breast thickness. This noise reduction was also demonstrated in the profile plots of the lesions. The line profile analysis, in agreement with visual perception, shows improvement of sharpness of the lesion edge in relation to the reduction of the phantom thickness. The intraclass correlation coefficient (ICC) has shown a great consistency and agreement among the observers for visibility, sharpness, contrast and noise. The ICC results are not as conclusive for the size criterion. Mathematical evaluation results also show a correlation of improvement in the image quality with the reduction in breast thickness. The results show that for the measures CNR, SNR, and FOM, the increase in image quality has a threshold after which the image quality ceases to improve and instead begins to reduce. CNR and FOM dropped when the breast phantom thickness was reduced approximately 40% of its initial thickness. This consistently happened at the point where the filter changed from rhodium (Rh) to molybdenum (Mo). Conclusion: This breast phantom study successfully designed and developed an anthropomorphic compressible breast phantom with cancer mimicking lesions with mechanical and X-ray properties similar to human breast tissue. This study also demonstrates that as breast compressed thickness reduces the visibility of the perceived lesion increases. The radiation dose generally decreases up to the point that the filter changes from rhodium to molybdenum. After this point, the radiation dose increases regardless of the phantom thickness. The results from this thesis are likely to have implications for clinical practice, as they support the need for compression/thickness reduction to enhance lesion visibility.
| Identifer | oai:union.ndltd.org:bl.uk/oai:ethos.bl.uk:668743 |
| Date | January 2015 |
| Creators | Ossati, M. S. |
| Publisher | University of Salford |
| Source Sets | Ethos UK |
| Detected Language | English |
| Type | Electronic Thesis or Dissertation |
| Source | http://usir.salford.ac.uk/36136/ |
Page generated in 0.002 seconds