Microscopy is crucial for diagnosing sickle cell disease by enabling the examination of blood samples at the cellular level. This technique reveals abnormalities of red blood cells, aiding in the identification of disease-specific anomalies. However, image quality is frequently degraded by random ‘salt and pepper’ noise, complicating subsequent processing tasks such as segmentation and feature extraction. This study investigates the Noise2Void deep learning denoising model to enhance the quality of microscopic images of blood smears from sickle cell disease patients. By evaluating the model’s performance on clinical data, we demonstrated that N2V outperforms traditional denoising methods as well as the Noise2Noise model, achieving PSNR and SSIM values of 43.98 dB and 0.98, respectively. Our results underscore the potential of N2V to significantly enhance data quality for the development of accurate sickle cell disease detection methods.