Mammogram images classification based on fuzzy soft set
Early detection of the breast cancer can decrease mortality rates. Screening mammography is considered the most reliable method in early detection of breast cancer. Due to the high volume of mammograms to be read by a physician, the accuracy rate tends to decrease. Thus, automatic digital mammograms...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Published: |
2016
|
| Subjects: | |
| Online Access: | http://eprints.uthm.edu.my/8882/ http://eprints.uthm.edu.my/8882/1/Saima_Anwar_Lashari.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Early detection of the breast cancer can decrease mortality rates. Screening
mammography is considered the most reliable method in early detection of breast
cancer. Due to the high volume of mammograms to be read by a physician, the
accuracy rate tends to decrease. Thus, automatic digital mammograms reading
becomes highly enviable, it is premised that the computer aided diagnosis systems
are required to assist physicians/radiologists to achieve high efficiency and
effectiveness. Meanwhile, recent advances in the field of image processing have
revealed that level of noise highly affect the mammogram images quality and
classification performance of the classifiers. Therefore, this study investigates the
functionality of wavelet de-noising filters for improving images quality. The dataset
taken from Mammographic Image Analysis Society (MIAS). The best PSNR and
MSE values 46.36423dB (hard thresholding) and 1.827967 achieved with Daub3
filter. Whilst, several medical imaging modalities and applications based on data
mining techniques have been proposed and developed. However, fuzzy soft set
theory has been merely experimented for medical images even though the choice of
convenient parameterization makes fuzzy soft set practicable for decision making
applications. Therefore, the viability of fuzzy soft set for classification of
mammograms images has been scrutinized. Experimental results show better
classification performance in the presence/absence of de-noise filter in mammogram
images where the highest classification rate occurs with Daub3 (Level 1) with
accuracy 75.64% (hard threshold), precision 46.11%, recall 84.67%, F-Macro
75.64%, F-Micro 60% and performance of FussCyier without de-noise filter
classification accuracy 66.49%, precision 80.83%, recall 50% and F-Micro 68.18%.
Thus, the results show that proposed approach FussCyier gives high level of
accuracy and reduce the complexity of the classification phase, thus provides an
alternative technique to categorize mammogram images. |
|---|