Quantification and segmentation of breast cancer diagnosis: efficient hardware accelerator approach
The mammography image eccentric area is the breast density percentage measurement. The technical challenge of quantification in radiology leads to misinterpretation in screening. Data feedback from society, institutional, and industry shows that quantification and segmentation frameworks have rap...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | English English English |
| Published: |
2022
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/8430/1/24p%20KHAIRULNIZAM%20OTHMAN.pdf http://eprints.uthm.edu.my/8430/2/KHAIRULNIZAM%20OTHMAN%20COPYRIGHT%20DECLARATION.pdf http://eprints.uthm.edu.my/8430/3/KHAIRULNIZAM%20OTHMAN%20WATERMARK.pdf http://eprints.uthm.edu.my/8430/ |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Institution: | Universiti Tun Hussein Onn Malaysia |
| Language: | English English English |
| Summary: | The mammography image eccentric area is the breast density percentage
measurement. The technical challenge of quantification in radiology leads to
misinterpretation in screening. Data feedback from society, institutional, and industry
shows that quantification and segmentation frameworks have rapidly become the
primary methodologies for structuring and interpreting mammogram digital images.
Segmentation clustering algorithms have setbacks on overlapping clusters, proportion,
and multidimensional scaling to map and leverage the data. In combination,
mammogram quantification creates a long-standing focus area. The algorithm
proposed must reduce complexity and target data points distributed in iterative, and
boost cluster centroid merged into a single updating process to evade the large storage
requirement. The mammogram database's initial test segment is critical for evaluating
performance and determining the Area Under the Curve (AUC) to alias with medical
policy. In addition, a new image clustering algorithm anticipates the need for largescale
serial and parallel processing. There is no solution on the market, and it is
necessary to implement communication protocols between devices. Exploiting and
targeting utilization hardware tasks will further extend the prospect of improvement in
the cluster. Benchmarking their resources and performance is required. Finally, the
medical imperatives cluster was objectively validated using qualitative and
quantitative inspection. The proposed method should overcome the technical
challenges that radiologists face. |
|---|