In　medical　diagnostics,　the　invention　of　the　computer-aided　identification　method　has　played　a　significant　role　in　making　essential　decisions　for　human　diseases.　Lung　cancer　requires　a　greater　focus　among　various　diagnostic　processes　because　both　men　and　women　are　affected,　contributing　to　high　mortality　rates.　In　addition,　lung　cancer　is　one　of　the　leading　causes　of　death　worldwide.　It　can　be　treated　if　diagnosed　at　an　early　stage.　Detecting　and　classifying　lung　lesions　is　challenging　for　radiologists.　Radiologists　typically　use　computer-aided　diagnostic　systems　to　screen　for　lung　cancer.　In　recent　years,　computer　specialists　have　proposed　many　techniques　for　diagnosing　lung　cancer.　Conventional　lung　cancer　prediction　methods　have　failed　to　maintain　the　precision　needed　because　the　low-quality　picture　affects　the　segmentation　process.　Here,　we　propose　a　well-performing　method　to　detect　and　classify　lung　cancer.　We　applied　the　Grey　Wolf　Optimization　algorithm　with　a　weighted　filter　to　reduce　noise　in　images,　followed　by　segmentation　using　watershed　transformation　and　dilation　operations.　In　the　end,　we　classified　lung　cancer　among　three　classes　using　our　method　that　showed　high　performance　compared　to　previous　studies:　98.33%　accuracy,　100%　sensitivity,　and　93.33%　specificity.