As　a　powerful　machine　learning　technique,　deep　learning　has　been　widely　applied　to　lesion　detection　in　medical　image　processing.　This　review　summarizes　the　research　progress　of　deep　learning　applications　in　lesion　detection.　Firstly,　the　characteristics　of　medical　image　data　are　introduced,　and　the　datasets　and　evaluation　metrics　of　lesion　detection　are　summarized.　Then,　the　main　contents　of　deep　learning,　including　Convolutional　Neural　Networks　(CNN),　Recurrent　Neural　Networks　(RNN),　Generative　Adversarial　Networks　(GAN),　YOLO　algorithm,　and　SAM,　have　demonstrated　good　performance　in　medical　image　processing.　Meanwhile,　the　applications　of　lesion　detection　in　different　medical　image　modalities　are　discussed,　and　the　advantages　of　deep　learning　in　different　lesion　types　are　highlighted,　such　as　high　automation,　good　performance,　and　transferability.　In　addition,　some　challenges　of　deep　learning　in　lesion　detection　are　discussed,　such　as　sample　scarcity,　interpretability,　and　reliability.　Finally,　the　future　development　directions　of　deep　learning　in　lesion　detection　are　discussed,　such　as　multimodal　fusion,　transfer　learning,　and　labeled　data.　This　review　provides　a　comprehensive　overview　of　the　research　progress　of　deep　learning　in　the　field　of　lesion　detection,　which　offers　guidance　and　reference　for　related　research　and　applications.　©　2023　IEEE.