Skip to main content
University of Oxford Department of Computer Science
  Contact Us     Search  
  1. Home
  2. About Us
  3. People
  4. Students
  5. Allison Clement

Allison Clement

Personal photo - Allison Clement

Allison Clement

Doctoral Student

Student, Keble College

E: allison.clement@cs.ox.ac.uk

Wolfson Building, Parks Road, Oxford OX1 3QD

Projects:

Interests

I am a DPhil Student a part of a Medical Imaging research group lead by Irina Voiculescu. I am part of the Computational and Health Informatics theme. I have an undergraduate degree in Biomechanical Engineering and a Master's in Biomedical Engineering. I have worked on many computational projects in biomechanical engineering. This has led me to Computer Science. 

My research focuses on techniques aimed at creating a better understanding of the skeleton and developing technologies for diagnosis/treatment of orthopaedic disease.

My work to date focuses on screening babies for hip dislocations. It is important to improve screening methods to allow for more accessible patient care. To do this, I have developed landmark-based methods incorporating the clinical decision to improve landmark localisation. 

My most recent accepted work (MICCAI 2025) focuses on creating and validating methods for reporting a machine's confidence in angle predictions for orthopaedic tasks.

Multimodal (text and image data)

 

Currently, I am focusing on developing and evaluating the incorporation of multimodal (text and image data) to improve landmark-based orthopaedic screening. My recent paper "Quantifying Contributions within Multimodal Fusion for Clinical Decisions" was accepted to MICCAI 2026 ML-CDS 2026: Multimodal Learning and Fusion Across Scales for Clinical Decision Support  Workshop. Here we established a method for per patient contributions of each input on the final decision of the multimoda (image-text) pipeline. Some results are shown below.

See more publications and work on my Google Scholar Page.

Current work includes:

  • Improving automated ultrasound screening methods.
  • Exploring the use of integrating clinical classification into a loss function to improve ultrasound screening.
  • Combined landmark−based screening method to improve disease classification.
  • Creating angle metrics to evaluate agreement with classification.
  • Incorporating multimodal (text and image data) to improve orthopaedic screening.

Previous work includes:

  • Skull to face predictive models (UNets) using metadata and CT images.
  • Skull implants generation for defects from CT images ( Auto Implant Challenge MICCAI 2021)
  • Facial recognition CNNs for intraoperative tools to improve nasal reconstruction surgery.
  • Automated tumour segmentation in preclinical and clinical models.
  • Material, mechanical and quantitative characterisation of bone architecture and quality.

Selected Publications

View AllManage publications

Supervisor

Irina Voiculescu

About Us

Back to Top