All talks & lectures

Conference · 29 May 2024

Impacts of AI-enabled neuroimaging on MS treatment plans

A treatment-planning view of AI neuroimaging, spanning mechanism, prognosis, response and monitoring.

38th Annual Meeting of the Consortium of Multiple Sclerosis Centers · Nashville, Tennessee

Opening slide from the original presentation

Delivered: 29 May 2024
Event: 38th Annual Meeting of the Consortium of Multiple Sclerosis Centers
Place: Nashville, Tennessee
Format: Conference
Source deck: 29 slides

From the slides

Deck notes

This talk followed the treatment pathway from selecting a therapy to monitoring its effects. It asked where AI-enabled neuroimaging might add information at each point, and where important gaps remain.

The question

How could quantitative imaging support treatment planning without reducing a complex clinical decision to a single model output?

Deck outline

Artificial intelligence, machine learning and deep learning in clinical imaging
Linking disease mechanisms to treatment selection
Prognosis and prediction of treatment response
Monitoring new lesions, brain volume change and progression
Analysis of heterogeneous, single-sequence clinical MRI
Patient views, adverse effects, cost, diversity and digital exclusion

Take-away

AI neuroimaging may support several parts of a treatment decision, but no one measure captures the full problem. Clinical usefulness also depends on patient priorities, safety, cost and representative evidence.

Original deck

Slide by slide

Read the 29-slide presentation in order, or use the index to jump.

Slide 01
Impacts of AI-Enabled Neuroimaging on MS Treatment Plans
Open full-size slide
Arman Eshaghi, MD, PhD
Queen Square Institute of Neurology
University College London
United Kingdom
Slide 02
Disclosures
Open full-size slide
Received travel support from the National Multiple Sclerosis Society and honorarium from the Journal of Neurology, Neurosurgery and Psychiatry for Editorial Commentaries.
Received research grants from Biogen, Merck and Roche. He acted as a member of Merck Serono and Bristol Myers Squibb advisory board meetings.
Founder and equity stakeholder in Queen Square Analytics Limited.
Serves on the editorial board of Neurology (American Academy of Neurology).
Slide 03
Outline
Open full-size slide
Treatment prescription
- Pathomechanisms
- Prognosis and response prediction
Treatment monitoring
- Disease activity
- Disability worsening/progression
Patient views
Slide 04
Definitions
Open full-size slide
Artificial intelligence
Machine learning
Deep learning
Slide 05
Treatment selection: pathomechanisms
Open full-size slide
Slide 06
Lublin et al, 2014, Neurology
Open full-size slide
References
- Lublin et al, 2014, Neurology — PubMed search
Slide 07
Kuhlmann et al, The Lancet Neurology, 2022
Open full-size slide
References
- Kuhlmann et al, The Lancet Neurology, 2022 — PubMed
Slide 08
Data-driven MS types, Eshaghi et al, Nature Communication, 2021
Open full-size slide
References
- Data-driven MS types, Eshaghi et al, Nature Communication, 2021 — PubMed search
Slide 09
Treatment selection: prognosis
Open full-size slide
Slide 10
Zhao et al, Nature Digital Medicine, 2020
Open full-size slide
References
- Zhao et al, Nature Digital Medicine, 2020 — PubMed search
Slide 11
Gros et al., Neuroimage, 2019
Open full-size slide
References
- Gros et al., Neuroimage, 2019 — PubMed
Slide 12
Treatment selection: response prediction
Open full-size slide
Slide 13
Falet et al, Nature Communications, 2022
Open full-size slide
References
- Falet et al, Nature Communications, 2022 — PubMed
Slide 14
Treatment Monitoring: disease activity
Open full-size slide
Slide 15
Source: MSSEG-2 challenge proceedings, Commonwick et al, MICCAI 2021
Open full-size slide
New lesions
Baseline
Follow up
References
- Source: MSSEG-2 challenge proceedings, Commonwick et al, MICCAI 2021 — PubMed search
Slide 16
Barnett et al., Nature Digital Medicine, 2023
Open full-size slide
References
- Barnett et al., Nature Digital Medicine, 2023 — PubMed
Slide 17
Treatment Monitoring: progression independent of relapse and disease activity
Open full-size slide
Slide 18
Cagol et al, JAMA Neurology, 2022
Open full-size slide
Lublin et al, Brain, 2022
Speaker notes
The effect size, expressed as mean difference in annual percentage cortical thickness change (MD-APC), is graphically displayed in different shades of blue for each of the Desikan-Killiany atlas25 regions presenting significant differences between groups after correction for multiple comparisons.
References
- Cagol et al, JAMA Neurology, 2022 — PubMed
- Lublin et al, Brain, 2022 — PubMed search
Slide 19
Percentage brain volume change
Open full-size slide
Slide 20
Montalban et al, 2017
Open full-size slide
References
- Montalban et al, 2017 — PubMed search
Slide 21
Kappos et al, Lancet, 2018
Open full-size slide
References
- Kappos et al, Lancet, 2018 — PubMed search
Slide 22
Clinical utility
Open full-size slide
Courtesy of Hugh Pemberton and Olivia Goodkin
Slide 23
MindGlide: https://github.com/MS-PINPOINT/mindGlide
Open full-size slide
Annual percentage change
Goebl, et al, Under Review, expected 2024
Lesion
Cortical GM
References
- Goebl et al, Nature Communications, 2025 — PubMed
Slide 24
Zivadinov et al., JNNP, 2022
Open full-size slide
References
- Zivadinov et al., JNNP, 2022 — PubMed
Slide 25
Treatment Planning: Patient Views
Open full-size slide
Slide 26
Zaretsky et al, JAMA Network Open, 2024
Open full-size slide
References
- Zaretsky et al, JAMA Network Open, 2024 — PubMed
Slide 27
Gaps
Open full-size slide
Adverse effects
Real-world clinical trials
Cost-effectiveness
Diversity and Digital Exclusion
Slide 28
Summary
Open full-size slide
Linking pathomechanisms to treatments
Prognosticating outcomes
Predicting response to treatments
Monitoring
Slide 29
Slide 29
Open full-size slide
Email: a.eshaghi@ucl.ac.uk