Semantic segmentation of medical images presents an enormous potential for diagnosis and surgery. However, achieving precise results involves designing and training complex Deep Learning (DL) models specifically for this task, which is only available to some. Segment Anything Model (SAM) is a model developed by Meta capable of segmenting objects present in virtually any type of image. This paper showcases SAM’s robustness and exceptional performance in medical image segmentation, even in the absence of direct training on these image types (lung Computed TomographyComputed Tomographies (CTs) and chest X-rays, in particular). Additionally, it achieves this impressive outcome while requiring minimal user intervention. Although the dataset used to train SAM does not contain a single sample of both medical image types, processing a popular dataset comprised of 20 volumes with a total of 3520 slices using the ViTL version of the model yields an average Jaccard index of 91.45% and an average Dice score of 94.95 %. The same version of the model achieves a 93.19%Dice score and a 87.45%Jaccard index when segmenting a frequently-used chest X-ray dataset. The values obtained are above the 70% mark recommended in the literature, and close to state-of-the art models developed specifically for medical segmentation. These results are achieved without user interaction by providing the model with positive prompts based on the masks of the dataset used and a negative prompt located in the center of bounding box that contains the masks.

Semantic segmentation of medical images holds significant potential for enhancing diagnostic and surgical procedures. Radiology specialists can benefit from automated segmentation tools that facilitate identifying and isolating regions of interest in medical scans. Nevertheless, to obtain precise results, sophisticated deep learning models tailored to this specific task must be developed and trained, a capability not universally accessible. Segment Anything Model (SAM) 2 is a foundational model designed for image and video segmentation tasks, built on its predecessor, SAM. This paper introduces a novel approach leveraging SAM 2’s video segmentation capabilities to reduce the prompts required to segment an entire volume of medical images. The study first compares SAM and SAM 2’s performance in medical image segmentation. Evaluation metrics such as the Jaccard index and Dice score are used to measure precision and segmentation quality. Then, our novel approach is introduced. Statistical tests include comparing precision gains and computational efficiency, focusing on the trade-off between resource use and segmentation time. The results show that SAM 2 achieves an average improvement of 1.76% in the Jaccard index and 1.49% in the Dice score compared to SAM, albeit with a ten-fold increase in segmentation time. Our novel approach to segmentation reduces the number of prompts needed to segment a volume of medical images by 99.95%. We demonstrate that it is possible to segment all the slices of a volume and, even more, of a whole dataset, with a single prompt, achieving results comparable to those obtained by state-of-the-art models explicitly trained for this task. Our approach simplifies the segmentation process, allowing specialists to devote more time to other tasks. The hardware and personnel requirements to obtain these results are much lower than those needed to train a deep learning model from scratch or to modify the behavior of an existing one using model modification techniques.

This study evaluates the effectiveness of three loss functions Asymmetric Unified Focal Loss (AUFL), Dice Similarity Coefficient Loss (DSCL), and Cross‐Entropy (CE) for segmenting COVID‐19 lung infections in computed tomography images. Detailed analyses using the intersection over union metric assessed each function's accuracy. AUFL achieved an average Dice Similarity Coefficient (DSC) of 85.18% ± 8.86%, outperforming DSCL 85.18% ± 8.86%, which had the same average DSC but less precise segmentation, and CE, which had an average DSC of 78.31% ± 11.93%. Segmentations using AUFL demonstrated more defined contours and better alignment with actual anatomical structures than those obtained with DSCL and CE. Observations revealed that AUFL‐generated segmentations had more precise boundaries and were more consistent with the expected anatomical regions of lung infections. This study is the first to quantitatively and qualitatively compare the effectiveness of AUFL, DSCL, and CE in segmenting COVID‐19 lung infections, providing concrete evidence of AUFL's superiority in segmentation performance and reliability for clinical applications. The findings underscore the importance of selecting appropriate loss functions to enhance segmentation in medical imaging, highlighting their crucial role in improving image‐based diagnostics and treatment. The study emphasizes the need for ongoing research to optimize these segmentation techniques further.

This paper investigates the limits of foundation models in medical image segmentation, mainly focusing on SAM by Meta. While previous research demonstrated SAM’s potential for cost-efficient segmentation, this study explores its performance enhancement through integration with prompt enhancement optimization and genetic algorithms, aiming to minimize user input further. As a proof of concept, we apply this novel approach to lung segmentation tasks using public axial lung CT scans, frontal chest X-ray datasets, and spleen MRIs. Our findings reveal that the genetic algorithm optimization significantly improves SAM’s segmentation accuracy, bringing it closer to the state-of-the-art performance achieved by specifically trained models. In particular, when compared with our previous approach, this technique reaches a 94.85 % Jaccard Index (+3.77 delta) and a 97.17 % Dice Score (+2.50 delta) for lung CT scans, a 93.39 % Jaccard Index (+5.95 delta) and a 96.57 %Dice Score (+3.38 delta) for chest X-rays, and a 91.00 % Jaccard Index (+6.51 delta) and a 95.07 % Dice Score (+4.12 delta) for spleen MRIs. Notably, this improvement is achieved without retraining or modifying SAM’s architecture. However, our analysis also identifies an inherent limitation in this optimization approach, revealing a performance ceiling that cannot be surpassed despite further genetic algorithm iterations. The implications of these findings emphasize the potential of combining foundation models with non-intrusive optimization techniques for cost-effective and accessible medical image segmentation. While dataset-related limitations may affect generalizability, validating the approach across broader clinical scenarios remains essential. Future work should explore applications to additional organs, diverse datasets, and the integration of expert-in-the-loop strategies to enhance clinical utility.