Virtually Objective Quantification of in vitro Wound Healing Scratch Assays with the Segment Anything Model

TL;DR

This work tackles the subjectivity in quantifying wound closure in in vitro scratch assays by employing an interactive, prompt-based segmentation approach with the Segment Anything (SAM) framework, enabling training-free, generalizable analysis. ISAMS leverages user clicks to generate segmentation masks in real time and is evaluated against a WHST baseline and multi-expert ground truth, using Dice, precision, recall, and the $ \Delta A$ metric. Results demonstrate that ISAMS delivers higher segmentation accuracy and markedly reduced intra- and inter-observer variability, while maintaining rapid processing times, suggesting it can provide objective, scalable wound healing quantification. The approach offers practical impact for reproducible large-scale studies and is complemented by publicly shared data and code for reproducibility and further development.

Abstract

The in vitro scratch assay is a widely used assay in cell biology to assess the rate of wound closure related to a variety of therapeutic interventions. While manual measurement is subjective and vulnerable to intra- and interobserver variability, computer-based tools are theoretically objective, but in practice often contain parameters which are manually adjusted (individually per image or data set) and thereby provide a source for subjectivity. Modern deep learning approaches typically require large annotated training data which complicates instant applicability. In this paper, we make use of the segment anything model, a deep foundation model based on interactive point-prompts, which enables class-agnostic segmentation without tuning the network's parameters based on domain specific training data. The proposed method clearly outperformed a semi-objective baseline method that required manual inspection and, if necessary, adjustment of parameters per image. Even though the point prompts of the proposed approach are theoretically also a source for subjectivity, results attested very low intra- and interobserver variability, even compared to manual segmentation of domain experts.

Figures (4)

• Figure 1: Overview of the experiments performed in this study. While the right box shows all sub-experiments performed for both interactive segmentations, manual and ISAMS, the left box shows the overall big picture including evaluation.
• Figure 2: Results of the methods ISAMS and WHST based on the metrics Dice, precision, recall and $\Delta A$.
• Figure 3: Comparison of intra- and interobserver variability based on the measures $\Delta A$ and Dice for manual segmentation and ISAMS.
• Figure 4: Differences in wound segmentation between experts during manual segmentation (first row), during interactive segmentation using ISAMS (second row) and between the different methods (third row) at three stages (see different columns)