Deep Learning-Based Red Blood Cell Classification for Sickle Cell Anemia Diagnosis Using Hybrid CNN-LSTM Model

A mutation in the beta-globin gene results in the blood condition known as Sickle cell anemia. It is estimated that number of individuals affected by sickle cell anemia worldwide exceeded 20 million and 250 million individuals globally bear the gene accountable for sickle cell disease and other hemoglobinopathies. SCA is characterized by aberrant red blood cells with a crescent form that can obstruct blood arteries and result in a number of health issues. Manual detection of SCA is time-consuming and costly. The goal of this study is to increase the efficiency and precision of sickle cell anemia detection by employing deep learning techniques on microscopic blood cell images, which can result in early identification and better patient care by employing deep learning techniques on microscopic blood cell images. Data augmentation techniques are applied to expand the dataset and enhance the model's performance. Otsu thresholding followed by watershed segmentation and region-based segmentation is performed for image processing. We have employed Extreme Learning Machine (ELM), a pretrained CNN InceptionV3 architecture, and a hybridized model of CNN + LSTM for classifying images into circular, elongated, and other categories. This amalgamated architecture facilitates deep feature extraction through CNN and detection via the extracted features using LSTM. The proposed hybridized model exhibits superior performance and accuracy which surpasses the results achieved by alternative approaches, making it a reliable tool for early SCA detection and improved patient care.