Background/Objective: The rising global incidence of skin cancer emphasizes the urgent need for reliable and accurate diagnostic tools to aid early intervention. This study introduces YOLOSAMIC (YOLO and SAM in Cancer Imaging), a fully automated segmentation framework that integrates YOLOv8 for lesion detection, and the Segment Anything Model (SAM)-Box for precise segmentation. The objective is to develop a reliable segmentation system that handles complex skin lesion characteristics without requiring manual intervention. Methods: A hybrid database comprising 3463 public and 765 private dermoscopy images was built to enhance model generalizability. YOLOv8 was employed to localize lesions through bounding box detection, while SAM-Box refined the segmentation process. The model was trained and evaluated under four scenarios to assess its robustness. Additionally, an ablation study examined the impact of grayscale conversion, image blur, and model pruning on segmentation performance. Results: YOLOSAMIC demonstrated high segmentation accuracy, achieving Dice and Jaccard scores of 0.9399 and 0.9112 on the public database and 0.8990 and 0.8445 on the hybrid dataset. Conclusions: The proposed YOLOSAMIC framework provides a robust, fully automated solution for skin lesion segmentation, eliminating the need for manual annotation. Integrating YOLOv8 and SAM-Box enhances segmentation precision, making it a valuable decision-support tool for dermatologists.
30 April 2023 in Düzce Üniversitesi Bilim ve Teknoloji Dergisi
Çalışmamızda Q-öğrenme tabanlı adaptif PID kontrolörün gerçek zamanlı bir sistemdeki performansı incelenmiştir. Gerçek zamanlı sistem olarak DA makine hız kontrolü sistemi tercih edilmiştir. DA makine sisteminden gelen hata sinyali üzerinden sistemin durum bilgisi ve Q-öğrenme yöntemi için ödül sinyali hesaplanmaktadır. Durum bilgisi ve ödül sinyali yardımı ile PID katsayıları artırılıp azaltılarak optimal katsayılara ulaşılmaktadır. Her PID katsayısı için bir adet Q-tablosu tanımlanmıştır. Simülasyon çalışması ve gerçek zamanlı uygulama ile kontrolör performansı incelenmiştir. Pekiştirmeli öğrenme ile tasarlanan kontrolcünün klasik PID yapısı gibi başarılı olduğu tespit edilmiştir.
31 December 2024 in Sakarya University Journal of Computer and Information Sciences
Skin lesion segmentation for recognizing and defining the boundaries of skin lesions in images is proper for automated analysis of skin lesion images, especially for the early diagnosis and detection of skin cancers. Deep learning architectures are an efficient way to implement segmentation once a skin lesion dataset is provided with ground truth images. This study evaluates deep learning architectures on a hybrid dataset, including a private dataset collected from a hospital and a public ISIC dataset. Four different test cases exist in the analysis where the combinations of public and private datasets are used as train and test datasets. Experimental results include Unet, Unet++, DeepLabV3, DeepLabV3++, and FPN segmentation architectures. According to the comparative evaluations, mixed datasets, where public and private datasets were used together, provided the best results. The evaluations also show that the collected dataset with ground truth images provided promising results.
About Me
Research Assistant at Sakarya University Electical and Electronics Engineering
Research Assistant at Sakarya University Electical and Electronics Engineering