سامانه همایش های دانشگاه آزاد اسلامی واحد نجف آباد

صفحه اصلی / اولین همایش بین المللی تحول خدمات سلامت: تلفیق مراقبت، درمان، بازتوانی و امیدبخشی

Advancements in Computer-Aided Diagnosis Systems for Medical Imaging: A Comprehensive Review of Accuracy Enhancement and Segmentation Techniques

نویسندگان :

Maryam Saeedi Tazangi¹ Alireza Zarei²

1- موسسه آموزش عالی آپادانا 2- دانشگاه بین المللی امام رضا علیه السلام

کلمات کلیدی :

Diagnosis،Computer-Assisted،Diagnostic Imaging،Image Segmentation،Artificial Intelligence،Deep Learning،Machine Learning

چکیده :

Abstract: Background: Integrating Deep Learning (DL) into Computer-Aided Diagnosis (CAD) has revolutionized medical imaging by elevating diagnostic precision. However, clinical translation remains hindered by data heterogeneity, segmentation complexities, and the opaque “black-box” nature of neural networks. This systematic review evaluates recent CAD advancements, focusing on segmentation techniques and algorithmic interpretability.[1 - 3] Method: Adhering to PRISMA 2020 guidelines, a comprehensive search utilizing MeSH terms was executed across major databases[21]. From 600 initial records, 23 studies (2016-2026) were included after rigorous screening and quality assessment using the MINORS criteria. Due to clinical heterogeneity, a narrative synthesis mapped the evolution of diagnostic methodologies. Finding: A distinct shift toward advanced DL and hybrid architectures (e.g., U-Net, Vision Transformers) offers superior boundary delineation. Integrating Explainable AI (XAI) frameworks (Grad-CAM, SHAP) has emerged as mandatory to demystify black-box predictions. Critically, this review identified key limitations: English-only language restriction, high methodological heterogeneity among primary studies, and lack of full-text access for certain articles, which currently impact clinical generalizability[5]. Conclusion: While advanced DL frameworks provide unprecedented accuracy, widespread clinical adoption requires overcoming existing barriers. Future CAD development must prioritize designing lightweight architectures for real-time applications, establishing robust data standardization protocols, and expanding XAI integration to ensure physician trust.