People-Centred Mammogram Analysis (PecMan)

Project Overview

Current generic AI models for mammogram analysis provide biased results for patients and inflexible analysis for radiologists, reducing patients’ and radiologists’ trust in such models. In this project, we introduce an innovative design strategy for the development of new mammogram analysis AI models to increase their usability and trust by cooperating and personalising to radiologists and producing fair and accurate classification for all patient cohorts. In addition to measuring model accuracy for all patients, this proposal will introduce new assessment measures to evaluate the integration of the model into clinical practice in terms of radiologist’s performance improvement and workflow disruptiveness, and also to test the generalisation of the model to all patient sub-groups. Figure 1 compares the tradition data-centred approach with our proposed people-centred strategy.

Principal Investigators

CVSSP: Gustavo Carneiro , Kevin Wells

NHS Royal Wolverhampton : David Rosewarne

NHS Royal Surrey: Mark Halling-Brown

SVI – Melbourne: Helen Frazer

Univ. Sydney : Catherine Jones

DetectedX + Univ. Sydney : Patrick Brennan

Monash University: Thanh-Toan Do

PDRA1:Cuong Nguyen

PDRA2:Tahir Hassan

PhD Students: Zheng Zhang, Wenjie Ai, Milad Masroor

Funding Information

Funding Agency: EPSRC

Grant Number: EP/Y018036/1

Funding Period: 01/10/2023 to 01/04/2025

News and Announcements

Keynote speech at 4th DELTA Endometriosis Symposium.
Paper on interpretable and trustworthy image recognition accepted to TPAMI in 2025.
Paper on ordinal classification accepted to CVPR 2025.
Paper on hierarchical prototypes for medical image interpretation accepted to JBHI 2025
Paper on learning to defer accepted to ICLR 2025 as Oral Presentation!
1x Physics in Medicine & Biology paper on Human–AI collaborative multi-modal multi-rater learning.
Keynote speech at the The 5th International Conference on Medical Imaging and Computer-Aided Diagnosis in November 2024.
1x Pattern Recognition paper accepted on Noisy-label learning.
3x ECCV'24 papers accepted on Human-AI Collaborative Classification, one on Noisy-label Learning, and another on Object Detection with Crowdsourced Annotation in July 2024
Co-organisation of the ECCV 2024 Workshop on More Exploration, Less Exploitation in October 2024
Seminar at the AI-DLDA: International Summer School on Artificial Intelligence in July 2024 on Learning to Complement and to Defer to Multiple Users using Personalized Human-AI Collaborative Classifiers
Talk at IEEE Diversity, Equity and Inclusion Forum-Engineering and Women Health in June 2024 on Image-based Endometriosis Diagnosis
Seminar at the European Congress of Endocrinology 2024 EndoCompass Meeting in May 2024.
Access to OPTIMAM granted in April 2024
Seminar of People Centred Medical Image Analysis at CAMP on Generative Models in April 2024
Publication of the book Machine Learning with Noisy Labels (Definitions, Theory, Techniques and Solutions) in February 2024
Podcast interview in Dec 2023: Will AI Cure Cancer?
Seminar at the EndoCompass Event: Artificial Intelligence - Research and Clinical Impact in Endocrinology, Workshop in November 2023
Project announcement in August 2023

Publications

Cuong C. Nguyen, Thanh-Toan Do, Gustavo Carneiro. "Probabilistic Learning to Defer: Handling Missing Expert Annotations and Controlling Workload Distribution" International Conference on Learning Representation 2025.

<code to come >

Chong Wang, Yuanhong Chen, Fengbei Liu, Yuyuan Liu, Davis James McCarthy, Helen Frazer, and Gustavo Carneiro. "Mixture of gaussian-distributed prototypes with generative modelling for interpretable and trustworthy image recognition." IEEE Transactions on Pattern Analysis and Machine Intelligence (2025).

Chong Wang, Fengbei Liu, Yuanhong Chen, Chun Fung Kwok, Michael Elliott, Carlos Pena-Solorzano, Davis J. McCarthy, Helen Frazer, Gustavo Carneiro. “Progressive Mining and Dynamic Distillation of Hierarchical Prototypes for Disease Classification and Localisation”. IEEE Journal of Biomedical and Health Informatics (2025).

<code to come >

Dileepa Pitawela, Gustavo Carneiro, Hsiang-Ting Chen. CLOC: Contrastive Learning for Ordinal Classification with Multi-Margin N-pair Loss. In Proceedings of the IEEE/CVF Conference on Computer Vision and Pattern Recognition (CVPR) 2025.

<code to come >

Zhang, Zheng, Wenjie Ai, Kevin Wells, David Rosewarne, Thanh-Toan Do, and Gustavo Carneiro. "Learning to Complement and to Defer to Multiple Users." European Conference on Computer Vision 2024.

Wang, Hu, David Butler, Yuan Zhang, Jodie Avery, Steven Knox, Congbo Ma, Louise Hull, and Gustavo Carneiro. "Human-AI collaborative multi-modal multi-rater learning for endometriosis diagnosis." Physics in Medicine and Biology (2024).

Garg, Arpit, Cuong Nguyen, Rafael Felix, Thanh-Toan Do, and Gustavo Carneiro. "Instance-dependent Noisy-label Learning with Graphical Model Based Noise-rate Estimation." European Conference on Computer Vision 2024.

Tan, Zhi Qin, Olga Isupova, Gustavo Carneiro, Xiatian Zhu, and Yunpeng Li. "Bayesian Detector Combination for Object Detection with Crowdsourced Annotations." European Conference on Computer Vision 2024.

Cordeiro, Filipe R., and Gustavo Carneiro. "ANNE: Adaptive Nearest Neighbors and Eigenvector-based sample selection for robust learning with noisy labels." Pattern Recognition (2024): 111132.

Page updated

Google Sites

Report abuse