A deep learning self-attention cross residual network with Info-WGANGP for mitotic cell identification in HEp-2 medical microscopic images

Asaad Anaam, Mugahed A. Al-antari, Akio Gofuku

Abstract: Highlights•An end-to-end self-attention deep cross-residual network (Att-DCRNet) is proposed for efficient mitotic and interphase cell image classification.•A deep learning Info-WGANGP is used to synthesize new mitotic cell images for oversampling the minority mitotic cell patterns.•Combining GAN-synthesized with classic augmentation improved the classification performance.•Incorporating a self-attention mechanism enhanced the DCRNet classification performance.