Search research articles

ABOUT JoVE

Overview Leadership Blog JoVE Help Center

AUTHORS

Publishing Process Editorial Board Scope & Policies Peer Review FAQ Submit

LIBRARIANS

Testimonials Subscriptions Access Resources Library Advisory Board FAQ

RESEARCH

JoVE Journal Methods Collections JoVE Encyclopedia of Experiments Archive

EDUCATION

JoVE Core JoVE Business JoVE Science Education JoVE Lab Manual Faculty Resource Center Faculty Site

Terms & Conditions of Use

Search research articles

Related Experiment Video

Updated: May 28, 2026

A Multimodal Imaging Framework to Advance Phenotyping of Living Label-free Breast Cancer Cells

A Multimodal Imaging Framework to Advance Phenotyping of Living Label-free Breast Cancer Cells

Published on: August 22, 2025

Artificial Intelligence in Cancer Research: Modality Dependence and Limited Visual-Spatial Integration in Multimodal

Ibrahim Güler^1,2, Armin Kraus¹, Gerrit Grieb^3,4

¹Department of Plastic, Aesthetic and Hand Surgery, Otto-von-Guericke University, 39120 Magdeburg, Germany.

Life (Basel, Switzerland)

|May 27, 2026

Summary

Related Concept Videos

You might also read

Related Articles

Articles linked to this work by shared authors, journal, and citation graph.

Sort by

Same author

Geographic and Socioeconomic Variation in Aesthetic Plastic Surgery Pricing Across the United States: A State-Level Analysis of Income, Gross Domestic Product, and Real Economic Burden.

Aesthetic surgery journal. Open forum·2026

Same author

Evaluation of Aesthetic Outcomes Following Botulinum Toxin Treatment Using Multimodal Large Language Models: A Paired Before-and-After Analysis.

Aesthetic surgery journal. Open forum·2026

Same author

AI in Dermato-Oncology: Diagnostic Performance and Prompt-Injection Vulnerability of Vision-Language Models in Dermoscopic Skin Cancer Assessment.

Cancers·2026

Same author

Artificial Intelligence in Medical Assessment: Reliability and Performance of Multimodal Large Language Models in a High-Stakes Licensing Examination.

Behavioral sciences (Basel, Switzerland)·2026

Same author

AI in Hand and Wrist Radiography: Multimodal Large Language Models for Distal Radius Fracture Detection and Characterization.

Diagnostics (Basel, Switzerland)·2026

Same author

Artificial Intelligence for Biomedical Diagnostics: Diagnostic Accuracy and Reliability of Multimodal Large Language Models in Electrocardiogram Interpretation.

Life (Basel, Switzerland)·2026

This summary is machine-generated.

Multimodal large language models (MLLMs) struggle with cancer histopathology. They perform poorly on segmentation masks alone and do not improve when masks are combined with images, indicating limitations for diagnostic support.

Area of Science:

Computational pathology
Artificial intelligence in medicine
Large language models

Background:

Multimodal large language models (MLLMs) show promise for cancer diagnostic support.
Their effectiveness in interpreting histopathological images requires thorough evaluation.

Purpose of the Study:

To assess the performance of general-purpose MLLMs in breast cancer histopathology interpretation.
To evaluate MLLM performance using images, nuclei segmentation masks, and both modalities.

Main Methods:

Six contemporary MLLMs were tested on 58 H&E-stained breast cancer histopathology images and segmentation masks.
Each case was classified five times per model under image-only, mask-only, and combined conditions.

Main Results:

Keywords:

AI safety in oncology H&E images artificial intelligence breast cancer histopathology cancer diagnostics model variability multimodal large language models reproducibility segmentation masks texture dependence

Related Experiment Videos

Last Updated: May 28, 2026

A Multimodal Imaging Framework to Advance Phenotyping of Living Label-free Breast Cancer Cells

A Multimodal Imaging Framework to Advance Phenotyping of Living Label-free Breast Cancer Cells

Published on: August 22, 2025

Mean accuracy decreased significantly when using masks only (49.6%) compared to images only (69.4%).
Combining image and mask data did not improve classification accuracy and, for one model, decreased performance.
Models maintained high confidence even with near-random accuracy on masks, and reasoning shifted between modalities.

Conclusions:

Current general-purpose MLLMs rely heavily on visual features and struggle to integrate spatial structural information from histopathology.
These limitations highlight concerns for the safe deployment of MLLMs in clinical cancer diagnostics.