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 19, 2026

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images

Published on: April 13, 2013

CALM-VLM: CALIBRATION AND SELECTIVE PREDICTION IN VISION-LANGUAGE MODELS FOR RELIABLE BRAIN MRI CLASSIFICATION.

Nikhil J Dhinagar, Chirag Jagad, Pavithra Senthilkumar

Biorxiv : the Preprint Server for Biology

|April 27, 2026

Summary

Related Concept Videos

Calibration Curves: Linear Least Squares

Calibration Curves: Linear Least Squares

A calibration curve is a plot of the instrument's response against a series of known concentrations of a substance. This curve is used to set the instrument response levels, using the substance and its concentrations as standards. Alternatively, or additionally, an equation is fitted to the calibration curve plot and subsequently used to calculate the unknown concentrations of other samples reliably.
For data that follow a straight line, the standard method for fitting is the linear...

You might also read

Related Articles

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

Sort by

Same author

Exercise as a regulator of glymphatic function.

Trends in neurosciences·2026

Same author

Evaluation of an intensive, private, community, upper limb rehabilitation program for people with chronic stroke: a mixed methods case series.

Frontiers in rehabilitation sciences·2026

Same author

Transcutaneous auricular vagus nerve stimulation paired with task-specific training for improving walking and balance in chronic stroke: a double-blind, randomised controlled feasibility trial.

Journal of neuroengineering and rehabilitation·2026

Same author

Acute and longitudinal magnetic resonance imaging abnormalities in antibody-mediated encephalitis.

Brain communications·2026

Same author

Quantitative susceptibility mapping demonstrates ageing-related long-term structural brain changes in adult rats after mild ischaemic stroke.

Brain communications·2026

Same author

Polygenic Risk Identifies Older Adults Who May Benefit From Aspirin for the Primary Prevention of Ischemic Stroke.

Stroke·2026

We developed CALM-VLM, a calibrated vision-language model (VLM) for medical imaging. This AI tool improves diagnostic accuracy and confidence in predicting Alzheimer's disease and stroke from MRI scans.

Area of Science:

Artificial Intelligence
Medical Imaging
Neuroscience

Background:

Vision-language models (VLMs) show promise in medical image analysis but lack reliable confidence estimation.
Uncertainty in VLM diagnostic predictions hinders clinical adoption.

Purpose of the Study:

To introduce CALM-VLM, a novel framework integrating confidence calibration and selective prediction into generative 3D VLMs.
To enhance the reliability and diagnostic accuracy of VLMs for neuroimaging tasks.

Main Methods:

Fine-tuned a Med3DVLM architecture for Alzheimer's disease (AD) and stroke classification.
Implemented temperature scaling for confidence calibration of VLM generative outputs.
Enabled selective prediction abstention when the model is uncertain.

Related Experiment Videos

Last Updated: May 19, 2026

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images

Automated Midline Shift and Intracranial Pressure Estimation based on Brain CT Images

Published on: April 13, 2013

Main Results:

CALM-VLM demonstrated improved confidence compared to uncalibrated VLMs across multi-site MRI datasets.
Coverage-adjusted ROC-AUC increased by 5-13% for AD and stroke classification on independent test sets.
Achieved high test ROC-AUC scores: 0.951 for AD and 0.905 for stroke classification.

Conclusions:

Calibrated, uncertainty-aware VLMs are crucial for trustworthy neuroimaging AI.
CALM-VLM offers a reliable approach for AI-assisted diagnosis in neurology.
The findings support the clinical integration of calibrated generative VLMs.