Simulation of breast cancer imaging using magnetic induction tomography
In order to reduce the physical trauma caused by breast compressions, exposure to radiations and the high price of diagnostic tests, a new cost effective magnetic induction tomography (MIT) system is proposed to identify and locate tumors among the heterogeneous breast tissues. This technique ope...
Saved in:
| Format: | Thesis |
|---|---|
| Language: | English |
| Subjects: | |
| Online Access: | http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/78038/1/Page%201-24.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/78038/2/Full%20text.pdf http://dspace.unimap.edu.my:80/xmlui/bitstream/123456789/78038/3/Gowry.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | In order to reduce the physical trauma caused by breast compressions, exposure to
radiations and the high price of diagnostic tests, a new cost effective magnetic induction
tomography (MIT) system is proposed to identify and locate tumors among the
heterogeneous breast tissues. This technique operates in a non-invasive and contactless
manner with the breasts. The numerical simulation imaging system consists of 16 sensor
coils with 1 coil acting as the transmitter and the rest as receivers at a single time period,
leading to a total of 240 receiver readings. The receiver readings and 240 generated
sensitivity matrices were then used to reconstruct the images of the breast using linear back
projection (LBP) algorithm after a careful comparison has been made on the algorithm with
newton one-step error reconstruction (NOSER) and truncated singular value decomposition
(TSVD) algorithms. The reconstructed images were assessed in terms of three essential
error metrics which are the resolution (RES), magnification (MAG), and the position error
(PE). The average errors are 0.004728, 13.7793, and 45.1929 for the RES, MAG and PE
metrics respectively. Nonetheless, the average error metric values for the images of tumors
located deepest, at the origin (0,0), show better results in terms of PE, that is -2.5356. A
strong correlation between the MIT sensor readings and the size of simulated breast tumor
was also observed from the adjusted R square value which is 0.998, indicating that the data
fitted are very close to the regression line. The obtained results verify that the proposed
MIT design and image reconstruction algorithm provide a promising alternative for breast
cancer imaging although further studies are required to validate the simulation MIT data. |
|---|