# Prompting with the human-touch: evaluating model-sensitivity of foundation models for musculoskeletal CT segmentation

arXiv: https://arxiv.org/abs/2603.10541 · HTML: https://mlsanity.com/paper/2603.10541/

## Key facts

- No citing papers in the map yet (data through 14 Oct 2026).
- Builds on 24 of its 37 references that are in the map.
- First posted on arXiv 11 Mar 2026.

## Builds on (most cited first, 24 of 24)

- [SAM 2: Segment Anything in Images and Videos](https://mlsanity.com/paper/2408.00714/) (2024)
- [SAM 3: Segment Anything with Concepts](https://mlsanity.com/paper/2511.16719/) (2025)
- [TotalSegmentator: robust segmentation of 104 anatomical structures in CT images](https://mlsanity.com/paper/2208.05868/) (2022)
- [Segment Anything Model for Medical Image Analysis: an Experimental Study](https://mlsanity.com/paper/2304.10517/) (2023)
- [Metrics reloaded: Recommendations for image analysis validation](https://mlsanity.com/paper/2206.01653/) (2022)
- [Medical SAM 2: Segment medical images as video via Segment Anything Model 2](https://mlsanity.com/paper/2408.00874/) (2024)
- [VerSe: A Vertebrae Labelling and Segmentation Benchmark for Multi-detector CT Images](https://mlsanity.com/paper/2001.09193/) (2020)
- [SAM-Med3D: Towards General-purpose Segmentation Models for Volumetric Medical Images](https://mlsanity.com/paper/2310.15161/) (2023)
- [Segment Anything Model for Medical Image Segmentation: Current Applications and Future Directions](https://mlsanity.com/paper/2401.03495/) (2024)
- [SegVol: Universal and Interactive Volumetric Medical Image Segmentation](https://mlsanity.com/paper/2311.13385/) (2023)
- [SAM on Medical Images: A Comprehensive Study on Three Prompt Modes](https://mlsanity.com/paper/2305.00035/) (2023)
- [Foundational Models in Medical Imaging: A Comprehensive Survey and Future Vision](https://mlsanity.com/paper/2310.18689/) (2023)
- [VISTA3D: A Unified Segmentation Foundation Model For 3D Medical Imaging](https://mlsanity.com/paper/2406.05285/) (2024)
- [ScribblePrompt: Fast and Flexible Interactive Segmentation for Any Biomedical Image](https://mlsanity.com/paper/2312.07381/) (2023)
- [Deep Learning to Segment Pelvic Bones: Large-scale CT Datasets and Baseline Models](https://mlsanity.com/paper/2012.08721/) (2020)
- [Computer-Vision Benchmark Segment-Anything Model (SAM) in Medical Images: Accuracy in 12 Datasets](https://mlsanity.com/paper/2304.09324/) (2023)
- [Segment anything model 2: an application to 2D and 3D medical images](https://mlsanity.com/paper/2408.00756/) (2024)
- [SAM 2 in Robotic Surgery: An Empirical Evaluation for Robustness and Generalization in Surgical Video Segmentation](https://mlsanity.com/paper/2408.04593/) (2024)
- [Is SAM 2 Better than SAM in Medical Image Segmentation?](https://mlsanity.com/paper/2408.04212/) (2024)
- [Zero-shot performance of the Segment Anything Model (SAM) in 2D medical imaging: A comprehensive evaluation and practical guidelines](https://mlsanity.com/paper/2305.00109/) (2023)
- [Towards a Comprehensive, Efficient and Promptable Anatomic Structure Segmentation Model using 3D Whole-body CT Scans](https://mlsanity.com/paper/2403.15063/) (2024)
- [Vision Foundation Models in Medical Image Analysis: Advances and Challenges](https://mlsanity.com/paper/2502.14584/) (2025)
- [RadioActive: 3D Radiological Interactive Segmentation Benchmark](https://mlsanity.com/paper/2411.07885/) (2024)
- [Zero-shot capability of SAM-family models for bone segmentation in CT scans](https://mlsanity.com/paper/2411.08629/) (2024)

## Abstract (arXiv)

Promptable Foundation Models (FMs), initially introduced for natural image segmentation, have also revolutionized medical image segmentation. The increasing number of models, along with evaluations varying in datasets, metrics, and compared models, makes direct performance comparison between models difficult and complicates the selection of the most suitable model for specific clinical tasks. In our study, 11 promptable FMs are tested using non-iterative 2D and 3D prompting strategies on a private and public dataset focusing on bone and implant segmentation in four anatomical regions (wrist, shoulder, hip and lower leg). The Pareto-optimal models are identified and further analyzed using human prompts collected through a dedicated observer study. Our findings are: 1) The segmentation performance varies a lot between FMs and prompting strategies; 2) The Pareto-optimal models in 2D are SAM and SAM2.1, in 3D nnInteractive and Med-SAM2; 3) Localization accuracy and rater consistency vary with anatomical structures, with higher consistency for simple structures (wrist bones) and lower consistency for complex structures (pelvis, tibia, implants); 4) The segmentation performance drops using human prompts, suggesting that performance reported on "ideal" prompts extracted from reference labels might overestimate the performance in a human-driven setting; 5) All models were sensitive to prompt variations. While two models demonstrated intra-rater robustness, it did not scale to inter-rater settings. We conclude that the selection of the most optimal FM for a human-driven setting remains challenging, with even high-performing FMs being sensitive to variations in human input prompts. Our code base for prompt extraction and model inference is available: https://github.com/CarolineMagg/segmentation-FM-benchmark/