Related Experiment Video
Updated: Mar 19, 2026
08:27
Author Spotlight: Efficient Image Recognition Using Directional Gradient Histogram Technique and Support Vector Machines
Published on: January 5, 2024
1.7K
Digital image classification with the help of artificial neural network by simple histogram
Pranab Dey1, Nirmalya Banerjee1, Rajwant Kaur1
1Department of Cytopathology, Postgraduate Institute of Medical Education and Research, Chandigarh, Punjab and Haryana, India.
Journal of Cytology
|June 10, 2016
Summary
This study demonstrates that an artificial neural network (ANN) can effectively classify malignant and benign cells using simple histogram data from digital images, aiding cytopathologists in routine work.
Area of Science:
- Computational pathology
- Medical image analysis
- Artificial intelligence in diagnostics
Background:
- Visual image classification presents a significant challenge for cytopathologists in daily practice.
- Artificial neural networks (ANNs) offer a potential solution for automating cell classification tasks.
Purpose of the Study:
- To classify digital images of malignant and benign cells in effusion cytology smears.
- To evaluate the utility of simple histogram data and ANNs for cell classification.
Main Methods:
- A dataset of 404 digital images (168 benign, 236 malignant) was utilized.
- Histogram data was extracted, and a 6-3-1 architecture ANN was constructed using Alyuda Neurointelligence software.
- An on-line backpropagation training algorithm was employed with training, validation, and test sets.
Main Results:
- The ANN system underwent 10,000 training iterations at a speed of 609.81/s.
- The trained ANN model successfully identified all 34 malignant cell images.
- The model correctly identified 24 out of 26 benign cell images.
Conclusions:
- An ANN model utilizing simple histogram data can accurately identify individual malignant cells.
- This approach holds promise for future applications in identifying malignant cells in unknown cytological samples.
