In　recent　years,　deep　convolutional　neural　network　(DCNN)　models　have　shown　great　promise　in　the　automated　detection　of　brain　stroke　from　CT　scan　images.　However,　existing　DCNN　models　may　not　be　optimized　for　early　detection　of　stroke.　In　this　study,　we　present　a　novel　DCNN　model　for　the　early　detection　of　brain　stroke　using　CT　scan　images.　The　proposed　DCNN　model　consists　of　three　main　components:　a　feature　extractor,　a　feature　fusion　module,　and　a　stroke　detection　module.　The　feature　extractor　consists　of　multiple　convolutional　and　pooling　layers　that　extract　high-level　features　from　the　input　CT　scan　images.　The　feature　fusion　module　combines　the　features　extracted　from　the　different　layers　of　the　feature　extractor　to　create　a　more　informative　representation　of　the　input　image.　The　stroke　detection　module　uses　the　fused　features　to　detect　the　presence　of　stroke　in　the　image.　To　evaluate　the　proposed　model,　we　used　a　dataset　of　CT　scan　images　from　stroke　patients　and　healthy　controls.　The　dataset　was　divided　into　training　and　testing　sets,　and　the　model　was　trained　using　a　combination　of　supervised　and　unsupervised　learning　techniques.　We　compared　the　performance　of　the　proposed　model　to　several　state-of-the-art　DCNN　models　for　stroke　detection,　including　VGG16,　ResNet50,　and　InceptionV3.　The　results　of　our　experiments　showed　that　the　proposed　model　achieved　a　higher　accuracy,　sensitivity,　and　specificity　than　the　other　DCNN　models　for　stroke　detection.　The　proposed　model　also　outperformed　existing　methods　for　early　detection　of　stroke,　achieving　an　accuracy　of　96.5%　in　detecting　stroke　within　6 hours　of　onset.　Our　code　is　available　at　(“https://github.com/FatimaAyub12/DCNN”).　©　The　Author(s),　under　exclusive　licence　to　Springer　Science+Business　Media,　LLC,　part　of　Springer　Nature　2024.