Improving Sickle Cell Disease Classification: A Fusion of Conventional Classifiers, Segmented Images, and Convolutional Neural Networks

TL;DR

This work tackles automated sickle cell classification from microscopy images with a focus on reducing computational overhead. It proposes a hybrid pipeline that uses CNN-derived features as input to conventional classifiers (SVM and Naive Bayes) and compares results on original versus segmented images, using three backbones (DenseNet-169, ResNet-50, MobileNet) as feature extractors. The key finding is that MobileNet features with segmented images fed to SVM achieve $96.80\%$ accuracy, outperforming other configurations while DenseNet remains robust to segmentation. The approach yields substantial speedups (approximately 14× faster) compared with end-to-end CNN classification, making it suitable for resource-constrained settings and informing future integration of segmentation with traditional classifiers in medical-image analysis.

Abstract

Sickle cell anemia, which is characterized by abnormal erythrocyte morphology, can be detected using microscopic images. Computational techniques in medicine enhance the diagnosis and treatment efficiency. However, many computational techniques, particularly those based on Convolutional Neural Networks (CNNs), require high resources and time for training, highlighting the research opportunities in methods with low computational overhead. In this paper, we propose a novel approach combining conventional classifiers, segmented images, and CNNs for the automated classification of sickle cell disease. We evaluated the impact of segmented images on classification, providing insight into deep learning integration. Our results demonstrate that using segmented images and CNN features with an SVM achieves an accuracy of 96.80%. This finding is relevant for computationally efficient scenarios, paving the way for future research and advancements in medical-image analysis.

Figures (4)

• Figure 1: Steps of the proposed approach.
• Figure 2: Image instances from the erythrocytesIDB dataset showing original (top) and segmented (bottom) images.
• Figure 3: Examples of visualization of feature activation obtained by DenseNet.
• Figure 4: Comparison between each strategy evaluated considering accuracy performance.