Related Experiment Video
Updated: Jul 27, 2025
08:59
Morphology-Based Distinction Between Healthy and Pathological Cells Utilizing Fourier Transforms and Self-Organizing Maps
Published on: October 28, 2018
7.1K
Fourier ptychographic and deep learning using breast cancer histopathological image classification
Leena Thomas1,2,3, M K Sheeja1,2
1Department of Electronics & Communication Engineering, Sree Chitra Thirunal College of Engineering, Thiruvananthapuram, Kerala, India.
Journal of Biophotonics
|June 10, 2023
Summary
This study introduces a novel method combining Fourier ptychography (FP) and deep learning for accurate breast cancer histological image classification. The approach enhances tumor detection, outperforming traditional methods in identifying malignant growths.
Area of Science:
- Medical Imaging
- Computational Pathology
- Biomedical Engineering
Background:
- Accurate classification of breast cancer histological images is vital for detecting malignant tumors.
- Histopathological image analysis presents challenges in automated and precise diagnosis.
- Existing methods may lack the accuracy required for reliable clinical application.
Purpose of the Study:
- To develop and evaluate an automated classification system for breast cancer histopathological images.
- To integrate Fourier ptychography (FP) with deep learning for enhanced image analysis.
- To improve the accuracy and efficiency of malignant tumor detection.
Main Methods:
- Utilized Fourier ptychography (FP) to reconstruct high-resolution complex holograms from low-resolution multi-view images.
- Employed iterative retrieval with FP constraints for hologram reconstruction.
- Implemented feature extraction including entropy, geometrical, and textural features, with entropy-based normalization.
- Developed an Enhanced Neural Network (ENDNN) for classifying breast cancer images as normal or abnormal.
Main Results:
- The proposed method successfully classified breast cancer images into normal or abnormal categories.
- Feature extraction and optimization using entropy-based normalization improved classification performance.
- Experimental results demonstrated that the presented technique surpasses traditional methods in accuracy.
- The integration of FP and deep learning provided high-resolution imaging for detailed analysis.
Conclusions:
- The combined Fourier ptychography and deep learning approach offers a powerful tool for accurate breast cancer histological image classification.
- The ENDNN model, with optimized features, shows significant potential for improving diagnostic accuracy in histopathology.
- This technique represents a promising advancement over conventional methods for breast cancer detection.
