Breast Cancer Diagnosis Using Neuro-CBR Approach
Breast cancer has become the number one cause of cancer deaths amongst women. Once a breast cancer is detected, it can be classified a benign (not cancerous tissue) or malignant (cancerous tissue). However, it is very difficult to distinguish benign from one that is malignant due to its variability...
Saved in:
| Main Author: | |
|---|---|
| Format: | Thesis |
| Language: | eng eng |
| Published: |
2005
|
| Subjects: | |
| Online Access: | https://etd.uum.edu.my/1304/1/NORLIA_BT._MD._YUSOF.pdf https://etd.uum.edu.my/1304/2/1.NORLIA_BT._MD._YUSOF.pdf |
| Tags: |
Add Tag
No Tags, Be the first to tag this record!
|
| Summary: | Breast cancer has become the number one cause of cancer deaths amongst women. Once a breast cancer is detected, it can be classified a benign (not cancerous tissue) or malignant (cancerous tissue). However, it is very difficult to distinguish benign from one that is malignant due to its variability associated with the appearances ofthe cancer. The problem leads to a motivation for a researcher in finding a technique that can enhance the performance of the previous breast cancer test detections. Among the techniques that could assist a specialist in diagnosing the breast cancer disease is computer-aided detection and diagnosis, abbreviated as CAD. CAD tools have exploited a wide range of AI technique since these technique are able to support CAD's needs. Hence, there is a need for multiple AI approach to support CAD. In this study, the Neural Network (NN) simulator with backpropagation algorithm was developed to predict the condition of the breast cancer tumor whether it is benign or maligant and Case-Base Reasoning (CBR) engine developed to classify the cancer stages as well as suggesting appropriate treatment to the patient. In CBR, mono symbolic valued was used for training and testing purpose. NN model obtained 98.60% accuracy clasification. This implies that NN model can be used as an inductive, or exploratory, analytical tool for the prediction for the breast cancer tissue. Experimental result also shows that CBR is able to classify the stage correctly and display appropriate treatment planning based on the doctor evaluation. The results from this study indicate that CBR coupled with NN techniques have great potentials to be used for a critical domain like medical. The proposed system is developed in the web-based platform, so that it can be accessed anytime, anywhere regardless of the geographical location. |
|---|