Improving medical image analysis with AI

Using deep learning to predict individual outcome and response to treatment for patients with neurological diseases.

A head x-ray taken by a computer-assisted tomographic (CAT) scanner.

Background

Personalized medicine tailors treatment to the individual patient, and can be particularly crucial in chronic diseases such as multiple sclerosis (MS), where timely and effective treatment is essential to prevent morbidity and reduce disability.

While current decisions are based on general statistics and population-based clinical markers, harnessing patient images for personalized predictions can enable individualized treatment recommendations that better account for variability in drug efficacy across patients.

Yet these models remain underexplored, and the open challenges presented by real-world clinical settings hamper their safe deployment.

Objectives

We develop deep learning models leveraging medical images for real-world clinical contexts and improved patient care. This could result in tools to assist clinicians with the selection of optimal treatments for individual patients with chronic, incurable disease, as well as improved clinical trial analysis.

We seek to develop modern deep learning frameworks which accurately predict future patient outcomes and individual treatment effects, requiring them to overcome challenges presented by real-world data.

We also need to ensure the safety and reliability of the models for successful clinical deployment.

We have developed the first causal inference model for personalized medicine, based on 3D magnetic resonance images (MRI) acquired from patients suffering from neurological diseases during clinical trials.

In particular, this model can accurately predict the future worsening of a patient's disease under the effect of all possible therapies (and a placebo). The aim was also to identify individuals likely to respond to different treatments in heterogeneous populations, by predicting future individual treatment effects. At the individual level, this can be used to improve treatment recommendations: the framework helped identify sub-populations of patients who responded to treatments.

We are striving to improve model safety and reliability by building the first uncertainty-aware causal models for imaging-based personalized medicine. Communicating the probability of response to treatment would enable more informed recommendations to be made.

We tested the model on a large clinical trial dataset of multiple sclerosis patients, provided by the International Progressive MS Alliance.

Patients analyzed

Over 1,800 MS patients have been analyzed as part of the project.

Treatments evaluated

Experiments were conducted on 5 different treatments.

Randomized clinical trials

The model was tested in 4 randomized clinical trials.

Resources

Publications Directory: Probabilistic Vision Group (PVG)

The PVG is an interdisciplinary research lab dedicated to the development of probabilistic machine learning frameworks in computer vision for a wide range of real-world applications in neurology and neurosurgery.

View publications https://www.cim.mcgill.ca/~pvg/publications/

Improving Image-Based Precision Medicine with Uncertainty-Aware Causal Models

Interview with PhD student Joshua Durso-Finley about his paper and his research for MICCAI 2023.

Read the article https://www.rsipvision.com/MICCAI2023-Tuesday/8/

Publications

Improving Image-Based Precision Medicine with Uncertainty-Aware Causal Models

Joshua D. Durso-Finley

Jean-Pierre R. Falet

Raghav Mehta

Douglas Arnold

Nick Pawlowski

Tal Arbel

Image-based precision medicine aims to personalize treatment decisions based on an individual's unique imaging features so as to improve the… (see more)ir clinical outcome. Machine learning frameworks that integrate uncertainty estimation as part of their treatment recommendations would be safer and more reliable. However, little work has been done in adapting uncertainty estimation techniques and validation metrics for precision medicine. In this paper, we use Bayesian deep learning for estimating the posterior distribution over factual and counterfactual outcomes on several treatments. This allows for estimating the uncertainty for each treatment option and for the individual treatment effects (ITE) between any two treatments. We train and evaluate this model to predict future new and enlarging T2 lesion counts on a large, multi-center dataset of MR brain images of patients with multiple sclerosis, exposed to several treatments during randomized controlled trials. We evaluate the correlation of the uncertainty estimate with the factual error, and, given the lack of ground truth counterfactual outcomes, demonstrate how uncertainty for the ITE prediction relates to bounds on the ITE error. Lastly, we demonstrate how knowledge of uncertainty could modify clinical decision-making to improve individual patient and clinical trial outcomes.

2023-10-08

OpenReview.net/Archive (published)

doi.org

openreview.net

Personalized Prediction of Future Lesion Activity and Treatment Effect in Multiple Sclerosis from Baseline MRI

Joshua D. Durso-Finley

Jean-Pierre R. Falet

Brennan Nichyporuk

Douglas Arnold

Tal Arbel

Precision medicine for chronic diseases such as multiple sclerosis (MS) involves choosing a treatment which best balances efficacy and side … (see more)effects/preferences for individual patients. Making this choice as early as possible is important, as delays in finding an effective therapy can lead to irreversible disability accrual. To this end, we present the first deep neural network model for individualized treatment decisions from baseline magnetic resonance imaging (MRI) (with clinical information if available) for MS patients which (a) predicts future new and enlarging T2 weighted (NE-T2) lesion counts on follow-up MRI on multiple treatments and (b) estimates the conditional average treatment effect (CATE), as defined by the predicted future suppression of NE-T2 lesions, between different treatment options relative to placebo. Our model is validated on a proprietary federated dataset of 1817 multi-sequence MRIs acquired from MS patients during four multi-centre randomized clinical trials. Our framework achieves high average precision in the binarized regression of future NE-T2 lesions on five different treatments, identifies heterogeneous treatment effects, and provides a personalized treatment recommendation that accounts for treatment-associated risk (side effects, patient preference, administration difficulties).

2022-02-28

MIDL.io/2022/Conference (published)

doi.org

openreview.net

Estimating individual treatment effect on disability progression in multiple sclerosis using deep learning

Jean-Pierre R. Falet

Joshua D. Durso-Finley

Brennan Nichyporuk

Julien Schroeter

Francesca Bovis

Maria-Pia Sormani

Doina Precup