مشروع البحث: SEZGĠSEL OPTĠMĠZAS ON YONTEMĠ ĠLE ĠNSAN RETĠNA G RÜNTÜLERĠNDE OPTĠK DĠSK SEGMENTAS ONU VE DERĠN ĞRENME ĠLE HASTALIK TEġHĠSĠ
تحميل...
المساهمين
الممولين
رقم التعريف
665
الباحث
حميدة علي المشرقي
الوصف
Comprehensive experiments are conducted to identify optimal hyperparameters, significantly enhancing the accuracy of Glokom detection. GWO achieved an average sensitivity of 96.04%, a specificity of 99.58%, an accuracy of 99.39%, a DICE coefficient of 94.15%, and a Jaccard index of 90.4% on the Drishti-GS1 dataset for segmentation. For classification, the Swin Transformer model achieved high test accuracy of 100% for hold-out validation and 88.18% for 3-fold cross-validation on the Drishti-GS1 dataset. Additionally, on the ORIGA dataset, the Swin Transformer model obtained remarkable results with 96.15% test accuracy for hold-out validation and InceptionV3 achieved 93.84% test accuracy for 5-fold cross-validation. Compared to previous studies, the proposed models in this work have demonstrated superior performance.
الكلمات الدالة
OĞRENME ĠLE HASTALIK TEġHĠSĠ”