Related Experiment Video
Updated: Aug 26, 2025
Author Spotlight: AI-Driven Trypanosome Species Detection from Microscopic Images
Published on: October 27, 2023
A novel deep learning approach for sickle cell anemia detection in human RBCs using an improved wrapper-based feature
Alagu S1, Kavitha Ganesan1, Bhoopathy Bagan K1
1Department of Electronics Engineering, Madras Institute of Technology, Chennai, India.
Insights
This study introduces a deep learning method for diagnosing Sickle Cell Anemia (SCA) using red blood cell images. The approach significantly improves detection accuracy, aiding early clinical decisions.
Area of Science:
- Medical Imaging
- Computational Biology
- Machine Learning in Healthcare
Background:
- Sickle Cell Anemia (SCA) is a prevalent red blood cell disorder affecting vulnerable populations like young children and pregnant women.
- Early diagnosis of SCA is critical for effective management and improved patient outcomes.
- Computer-aided diagnosis systems offer a promising avenue for accurate and timely SCA detection.
Purpose of the Study:
- To propose a novel and effective deep learning approach for the identification of Sickle Cell Anemia.
- To enhance the accuracy of SCA detection by integrating advanced feature selection and classification techniques.
- To provide a reliable tool for pathologists to support early clinical decision-making.
Main Methods:
- Utilized a dataset of approximately 900 microscopic red blood cell images from the 'erythrocytes IDB' database.
- Extracted 2048 deep features using the pre-trained InceptionV3 model.
- Implemented an improved wrapper-based feature selection technique with Multi-Objective Binary Grey Wolf Optimization (MO-BGWO) combined with K-Nearest Neighbors (KNN) and Support Vector Machine (SVM) classifiers.
Main Results:
- The proposed deep learning system demonstrated superior performance compared to the standard InceptionV3 model with a SoftMax layer.
- The Support Vector Machine (SVM) classifier achieved a high accuracy of approximately 96%.
- The optimal subset of deep features selected by MO-BGWO, coupled with SVM, significantly enhanced system performance.
Conclusions:
- The developed deep learning approach offers a highly accurate and efficient method for Sickle Cell Anemia detection.
- The integration of advanced feature selection and optimization techniques is crucial for improving diagnostic performance.
- This system serves as a valuable tool for pathologists, facilitating earlier and more informed clinical decisions regarding SCA.
Abstract:
Sickle Cell Anemia (SCA) is a disorder in Red Blood Cells (RBCs) of human blood. Children under five years and pregnant women are mostly affected by SCA. Early diagnosis of this ailment can save lives. In recent years, the computer aided diagnosis of SCA is preferred to resolve this issue. A novel and effective deep learning approach for identification of sickle cell anemia is proposed in this work. Around nine hundred microscopic images of human red blood cells are obtained from the public database 'erythrocytes IDB'. All the images are resized uniformly. About 2048 deep features are extracted from the fully connected layer of pre-trained model InceptionV3. These features are further subjected to classification using optimization-based methods. An improved wrapper-based feature selection technique is implemented using Multi- Objective Binary Grey Wolf Optimization (MO-BGWO) approach with KNN and SVM for classification. The detection of sickle cell is also performed using typical InceptionV3 model by using SoftMax layer. It is observed that the performance of the proposed system seems to be high when compared to the classification using the original InceptionV3 model. The results are validated by various evaluation metrics such as accuracy, precision, sensitivity, specificity and F1-score. The SVM classifier yields high accuracy of about 96%. The optimal subset of deep features along with SVM enhances the system performance in the proposed work. Thus, the proposed approach is appropriate for pathologists to take early clinical decisions on detection of sickle cells.
