A mammogram and breast ultrasound-based expert system with image processing features for breast diseases
Loading...
Date
2007
Authors
Ngah, Umi Kalthum
Journal Title
Journal ISSN
Volume Title
Publisher
Abstract
Survival rates for breast cancer patients may be increased when the disease is
detected in its earliest stage through mammography. A thorough assessment during
breast screening would also include clinical, physical examination and ultrasound. The
implementation of mass screening would result in increased caseloads for radiologists
which would incur chances of improper diagnosis. Diagnosticians with the training and
experience to interpret mammographic images and breast ultrasounds are scarce. The
existence of an expert system would facilitate computer aided study and learning and
produce more experts in the area and would also prove to be useful in the training of
radiologists in the early part of their career. The archiving of knowledge gathered in this
area with patient cases would also promote the interpretation of images in a more
consistent manner and may be referred to from time to time. This study focuses on
developing expert systems based on the interpretation of mammographic (MAMMEX)
and ultrasound (SOUNDEX) images that may be used by expert and non-expert
doctors to deduce cases (according to the BI-RADS ‘Breast Imaging Recording and
Data System’) based upon patients’ history, physical and clinical assessment as well
as mammograms and breast ultrasound images. Digital enhancement of mammograms
and breast ultrasound through the existence of image processing routines may help to
accentuate images in the process of analyzing procedures. Image based extension of
the expert systems have also been built. A total of 179 retrospective cases from the
Radiology Department, Hospital Universiti Sains Malaysia were tested, producing an
accuracy, sensitivity and specificity of 97%, 96% and 92% respectively for MAMMEX
and 99%, 98% and 100% for SOUNDEX. The Receiver Operating Characteristics
(ROC) curve analysis produced an Area Under the Curve (AUC) with values of
0.997(±0.003) for MAMMEX and 0.996(±0.004) for SOUNDEX. The Randomized
Complete Block Design (RCBD) and the Two-Way Analysis of Variants (ANOVA)
proved that the results of MAMMEX and SOUNDEX are consistent with the
radiologists’ opinion. Two extensions of image processing algorithms, namely the
Fuzzy-Count Image Processing (FCIP) and the Automated Modified Seed-Based
Region Growing (AMSBRG) techniques are also implemented to facilitate the detection
of microcalcifications
Description
PhD
Keywords
Radiology , Mammogram , Breast diseases