This study presents a comprehensive and critical review of segmentation algorithms applied to digital fundus images, aiming to identify computational strategies that balance diagnostic accuracy with practical feasibility in clinical environments. A systematic search following PRISMA guidelines was conducted for studies published between 2014 and 2025, encompassing deep learning, classical machine learning, hybrid, and semi-supervised approaches. The review examines how each methodological family performs in segmenting key anatomical structures such as blood vessels, the optic disc, and the fovea, considering both algorithmic and clinical metrics. Findings reveal that advanced deep learning models—particularly U-Net and CNN-based architectures—achieve superior accuracy in delineating complex and low-contrast structures but demand high computational resources. In contrast, traditional and hybrid methods offer efficient alternatives for real-time or low-resource settings, maintaining acceptable precision while minimizing cost. Importantly, the analysis underscores the persistent gap between methodological innovation and clinical translation, emphasizing the need for lightweight, clinically interpretable models that integrate algorithmic performance with medical relevance.

Background/Objectives: Clinical foundation models (CFMs) are increasingly being applied to analyze unstructured clinical text, supporting tasks such as clinical reasoning, clinical documentation generation, and information extraction. However, their evaluation remains limited and lacks standardization, which represents a significant challenge for their safe integration into healthcare systems. This scoping review aimed to examine how performance, clinical utility, safety, and generalization are assessed in current evaluation frameworks and methodologies for CFMs applied to electronic health record text and to identify existing gaps and challenges in their evaluation. Methods: This scoping review was reported in accordance with PRISMA-ScR guidance, and the study selection process was summarized using a PRISMA flow diagram. A total of 448 records were identified in the initial search, of which 16 met the eligibility criteria and were included in the final synthesis. Results: The findings were organized into three categories: methodological frameworks, performance and clinical utility, and ethical and safety considerations. The results show progress in structured methods, evaluation schemes, and simulated environments. Nevertheless, inconsistencies persist in operational robustness, particularly regarding clinical reasoning and hallucinations. The literature also highlights important ethical concerns, including potential biases and regulatory gaps. Conclusions: Although CFMs show strong potential to transform clinical text analysis, no comprehensive framework currently guides their evaluation and safe use in real clinical settings. These findings highlight the need for a structured conceptual framework that integrates methodological, technical, clinical, and ethical criteria to support responsible implementation in clinical environments.