研究目的
To develop an efficient system for the detection of suspicious lesions in mammograms to aid in breast cancer diagnosis.
研究成果
The proposed hybrid technique effectively detects suspicious lesions in mammograms with high sensitivity and low false positive rate, outperforming existing methods. It demonstrates the potential for improving computer-aided diagnosis systems in breast cancer detection.
研究不足
The study uses a specific dataset (mini-MIAS) which may limit generalizability; the thresholding method may be sensitive to image quality variations; computational efficiency and real-time application are not addressed.
1：Experimental Design and Method Selection:
The system involves pre-processing using Top-Hat morphological filter and NL means filter, followed by threshold selection using Fuzzy C-means, gradient magnitude, and intensity contrast, and finally thresholding for segmentation. Performance is assessed using the FROC curve.
2：Sample Selection and Data Sources:
57 mammographic images from the mini-MIAS database, divided into 25 normal and 32 abnormal images with manually annotated lesions.
3：List of Experimental Equipment and Materials:
PC with Intel Core 2 Duo 1.60 GHz processor and 2 GB RAM, MATLAB 7.3 software, mammogram images from mini-MIAS database.
4：60 GHz processor and 2 GB RAM, MATLAB 3 software, mammogram images from mini-MIAS database.
Experimental Procedures and Operational Workflow:
4. Experimental Procedures and Operational Workflow: Pre-process images with Top-Hat and NL means filters, apply FCM clustering, compute gradient magnitude, determine threshold using statistical parameters, and segment images. Evaluate using sensitivity and FP/I metrics.
5：Data Analysis Methods:
Quantitative analysis using sensitivity and false positives per image, FROC curve analysis, and comparison with existing methods.
独家科研数据包，助您复现前沿成果，加速创新突破
获取完整内容
