Fig. 7: Deep learning algorithms for mHealth platforms.

a Image enhancement. The figure shows the process of using deep learning to enhance low-quality images captured by mobile phones (figure adapted with permission from Rivenson et al. ¹⁰⁷). b Image segmentation. The figure illustrates a deep learning workflow employing the U-net architecture for sickle cell analysis (figure adapted with permission from Haan et al. ³⁶). c Image classification. The figure illustrates the detection of an on-chip bubble signal by using a CNN employing Inception v3 architecture (figure adapted with permission from Draz et al. ¹⁰⁹). d Regression. The figure illustrates the regression CNN structure for counting bubbles (figure adapted with permission from Chen et al. ¹¹⁰). e Augmentation of the image dataset. The figure illustrates the structure of GAN for generating realistic synthetic microfluidic chip images to augment the image dataset (figure adapted with permission from Shokr et al. ¹¹¹).