EARLY STADIUM DETECTION OF MAMMAE CARCINOMA ON MAMMOGRAPHY IMAGING MODALITY USING K-MEANS CLUSTERING AND MORPHOLOGICAL OPERATIONS SEGMENTATION COMBINATION
Carcinoma mammae a disease that causes cells in the breast to change and grow to form malignant tumors. The gold standard for the diagnosis of carcinoma mammae stage is histopathology results. Mammography examination is an alternative to the diagnosis of carcinoma mammae. As well as the previous met...
Saved in:
Published in | Journal of medical imaging and radiation sciences Vol. 53; no. 4; pp. S52 - S53 |
---|---|
Main Authors | , , |
Format | Journal Article |
Language | English |
Published |
Elsevier Inc
01.12.2022
|
Subjects | |
Online Access | Get full text |
Cover
Loading…
Summary: | Carcinoma mammae a disease that causes cells in the breast to change and grow to form malignant tumors. The gold standard for the diagnosis of carcinoma mammae stage is histopathology results. Mammography examination is an alternative to the diagnosis of carcinoma mammae. As well as the previous method, called Thresholding, which has limitations, it is not enough to separate noise, in addition to the limited Region technique that produces poor contrast and relatively long time limitation, K-means Clustering and Morphological operations become an accurate diagnosis solution through the introduction of algorithms on digital images, this machine learning has the ability to represent features to make predictions so that they have the same expertise results with radiology specialists and even histopathology results. This study aims to use K-means Clustering and Morphological operations which is able to detecting early stages of carcinoma mammae and has similar results Histopathology.
This study used Quasi-experimental research with Post-test Only Control Group Design. Building Machine learning through the Matlab R2021a program. The sample used is 164 mammogram images. Data analysis using the validity test and Wilcoxon statistical test.
The result showed that from 164 samples, the machine learning model performance was good in detecting the stage of carcinoma mammae on mammogram images with an accuracy value of 97.35%, sensitivity of 85.74%, specificity of 96.92%, PPV of 88.04% and the NPV value of 96.91% and there are similarities with Histopathology results.
There are similarities between the results of mammogram image readings in detecting early stages of carcinoma mammae between K-means Clustering and Morphological operations with Histopathology results with a p-value of 0.782, which means that when machine learning is applied to the population, then machine learning provides high accuracy rates in predicting. |
---|---|
ISSN: | 1939-8654 1876-7982 |
DOI: | 10.1016/j.jmir.2022.10.172 |