ChatGPT V.4.0 Enhances Digital Ki-67 Quantification in Pathology

The Journal of Clinical Pathology has published research highlighting the innovative use of ChatGPT V.4.0 in digital pathology. The study specifically focuses on the application of the model's color segmentation feature for quantifying Ki-67, a crucial biomarker, in neuroendocrine tumors. This development is significant because it illustrates how advanced general-purpose AI, particularly multimodal large language models, can be effectively repurposed for highly specialized medical tasks without requiring custom model development from scratch. The findings suggest that conventional microscope camera images are sufficiently accurate for QuPath-based Ki-67 quantification, democratizing access to digital image analysis (DIA). The ability to adapt existing powerful AI models like ChatGPT V.4.0 for such intricate tasks opens new avenues for efficiency and accessibility in pathology workflows. This adaptability could reduce the barrier to entry for many institutions seeking to integrate AI into their diagnostic processes. However, the study also emphasizes the critical need for systematic validation of successive model versions. As multimodal large language models continue to advance, ensuring reproducibility and reliability across different iterations and API updates will be paramount. This research not only showcases a practical application of multimodal AI in healthcare but also contributes to the ongoing discussion about the responsible and effective deployment of AI in clinical settings.