Automated Sickle Cell Detection Using Seagull Optimization-Enhanced Densenet-121 Deep Network

This study presents a comprehensive framework
for automated Sickle Cell Detection (SCD) through classifying medical images by combining all necessary steps such as preprocessing, segmentation, optimization and deep learning into one. Pre-processing steps use adaptive histogram equalization (AHE) to enhance image contrast while reducing noise. After every pre-processed image is enhanced, a fuzzy c-means (FCM)
segmentation algorithm is applied to provide accurate
extractions from the pre-processed images. The FCM sorted
images are classified using Denset-121 augmented by the Seagull
Optimization (SOP) to improve feature extraction and return it
back to the user in less time. The experimental results show that
the new system has a 95% accuracy, precision, recall and F1
score and an AUC = 0.951, indicating that the classification
results are reliable. It is demonstrated in these experimental
results that the new developed framework provide efficient,
accurate and strong solutions for automated medical image
classification.