Research Associate: Medical Imaging AI - CAPTURE-PH RAIDA
The University of Sheffield
Stand out for this role — generate a tailored resume and cover letter in about a minute.
Benefits offered by this job
Job summary
The University of Sheffield is seeking a Grade 7.5–7.9 Research Associate in Computer Science for CAPTURE‑PH, a two‑year project applying AI to pulmonary hypertension with CT imaging. You will refine, retrain and harden existing RAIDA CT assets for robust, cross‑dataset performance.
You will work with radiologists, clinicians and statisticians to ensure reliable outputs and reproducible validation workflows, across ASPIRE and PHINDER cohorts.
Qualifications
- PhD or equivalent in CS/ML/medical imaging or closely related field.
- Strong experience developing deep‑learning methods for medical‑image segmentation, classification, regression or quantitative image analysis.
- Proficiency in Python and PyTorch with ability to develop and debug research code independently.
- Experience with medical‑imaging data and formats (DICOM/NIfTI), preprocessing, metadata handling and data quality assessment.
- Experience validating AI models across heterogeneous datasets, including external/hold‑out validation and generalisability assessment.
- Ability to build reproducible, version‑controlled computational pipelines and work in secure HPC environments.
- Excellent communication and collaboration with radiologists, clinicians and AI researchers.
- Familiarity with models/architectures such as nnU‑Net, MONAI, vision transformers or DenseNet/Grad‑CAM.
Responsibilities
- Refine and maintain RAIDA CT imaging‑AI assets for CAPTURE‑PH across ASPIRE/PHINDER cohorts.
- Audit RAIDA pipelines for lung parenchymal, cardiac, vascular analysis; prioritise changes for CAPTURE‑PH imaging.
- Develop supervised deep‑learning methods for segmentation, feature extraction, classification and prediction.
- Improve generalisability across scanners, vendors and acquisition protocols.
- Design training, testing and external validation with repeatability, calibration and bias assessment.
- Collaborate with radiologists and clinicians to interpret errors and feed corrections back into models.
- Create reproducible, version‑controlled research software and inference workflows.
- Link RAIDA biomarkers with clinical data to support CAPTURE‑PH analyses and translation.
- Contribute to meetings, reports, manuscripts, and governance.
Skills
Education
Tools
Job description
The University of Sheffield is a remarkable place to work. Our people are at the heart of everything we do. Their diverse backgrounds, abilities and beliefs make Sheffield a world‑class university.
We offer a fantastic range of benefits including a highly competitive annual leave entitlement (with the ability to purchase more), a generous pensions scheme, flexible working opportunities, a commitment to your development and wellbeing, a wide range of retail discounts, and much more. Find out more about our benefits (opens in a new window) and join us to become part of something special.
We are seeking a Grade 7 Research Associate in Computer Science for a two‑year post within CAPTURE‑PH, a multi‑centre programme using chest CT and artificial intelligence (AI) to improve the diagnosis, phenotyping and prognostic assessment of pulmonary hypertension associated with interstitial lung disease (PH‑ILD). The post will build on RAIDA, the group’s existing cardiothoracic imaging‑AI programme, rather than developing a new platform from scratch.
RAIDA brings together AI methods for automated analysis of cardiac chambers, pulmonary vessels and lung parenchyma, with explainable outputs and expert review. Your primary task will be to refine, retrain and technically harden these existing CT assets so that they work reliably across the CAPTURE‑PH datasets, initially ASPIRE and PHINDER and subsequently other approved cohorts.
You will address variation in scanners, acquisition protocols and disease phenotypes; undertake quality control and failure analysis; improve segmentation, feature extraction and prediction pipelines; and establish reproducible validation workflows. You will work closely with radiologists, pulmonary vascular clinicians, clinical scientists and statisticians to link RAIDA‑derived imaging biomarkers with right‑heart catheterisation, clinical measurements and outcomes. The aim is to deliver robust, generalisable research tools and validated imaging outputs that can support CAPTURE‑PH analyses and future clinical translation.
Refine, adapt and maintain existing RAIDA CT imaging‑AI assets for CAPTURE‑PH, with initial focus on robust operation across the ASPIRE and PHINDER cohorts.
Audit existing RAIDA pipelines for lung parenchymal, cardiac chamber, large‑vessel and pulmonary vascular analysis; identify technical limitations and prioritise changes required for heterogeneous CAPTURE‑PH imaging.
Develop and optimise supervised deep‑learning methods for segmentation, feature extraction, classification and prediction, including retraining or fine‑tuning existing models using expert‑reviewed contours and labels where required.
Implement methods to improve generalisability across scanners, vendors, reconstruction methods and acquisition protocols, including appropriate domain‑adaptation, semi‑supervised or related approaches where these add value.
Design and perform rigorous training, testing and hold‑out/external validation, including repeatability, calibration, subgroup performance, bias assessment, quantitative segmentation metrics and clinically relevant diagnostic/prognostic performance measures.
Undertake systematic image review, quality control and failure‑mode analysis; work with clinical experts to understand errors, curate difficult cases and feed corrections back into model development.
Create reproducible, version‑controlled research software and inference workflows that can process DICOM‑derived imaging at scale, generate structured quantitative outputs and operate within approved secure research computing environments.
- Link RAIDA‑derived CT biomarkers with right‑heart catheterisation, lung function, walk testing, NT‑proBNP, MRI and outcome data to support CAPTURE‑PH analyses of PH diagnosis, severity, prognosis and treatment‑responsive phenotypes.
Support human‑in‑the‑loop review and explainability of model outputs, including integration with approved image‑review workflows such as MIM where appropriate, and contribute to documentation needed for reproducibility and future translation.
Contribute to project meetings, milestone reporting, manuscripts, conference presentations and collaborative work with CAPTURE‑PH and RAIDA investigators, while following research governance, data‑security and quality‑assurance requirements.
- Carry out other duties, commensurate with the grade and remit of the post
The selection process will take place following the closing date. This will consist of an interview and a short technical/research presentation or practical discussion relevant to medical‑imaging AI. We will let candidates know if they have progressed to the selection stage following the closing date. If you need any support, equipment or adjustments to enable you to participate in any element of the recruitment process you can contact ClinMed‑Staffing@sheffield.ac.uk.
A PhD (submitted, achieved or equivalent research/professional experience) in computer science, machine learning, medical imaging, biomedical engineering, data science or a closely related discipline.
Strong practical experience developing deep‑learning methods for medical‑image segmentation, classification, regression or quantitative image analysis, preferably with 3D CT data.
Proficiency in Python and a modern deep‑learning framework such as PyTorch, with the ability to develop, debug, profile and optimise research code independently.
Experience working with medical‑imaging data and associated formats/workflows, including DICOM and/or NIfTI, preprocessing, metadata handling and data‑quality assessment.
Experience adapting or validating AI models across heterogeneous datasets, including external/hold‑out validation, failure analysis and assessment of generalisability across sites, scanners or acquisition protocols.
Ability to build reproducible, version‑controlled computational pipelines and to work effectively with secure research computing or high‑performance computing environments.
Strong analytical problem‑solving skills and the ability to take ownership of a defined technical work programme, prioritise tasks and deliver agreed milestones over a two‑year project.
Effective communication and collaborative working skills, including the ability to work with radiologists, clinicians, clinical scientists, statisticians and other AI researchers.
Experience with medical‑image segmentation frameworks/architectures such as nnU‑Net, MONAI, vision transformers, or related approaches, and/super explainable image‑classification models such as DenseNet/Grad‑CAM.
Knowledge of thoracic CT, pulmonary hypertension/ILD, domain adaptation or semi‑supervised learning, or human‑in‑the‑loop clinical AI validation.
Grade 7.5 - 7.9
£41,064 - £46,049
Full‑time (hybrid/flexible working options considered where compatible with secure data and computing requirements)
Fixed‑term for 24 months
RAIDA / CAPTURE‑PH academic lead
None
If you do not currently hold the right to work in the UK, you can find more information here to help determine your visa eligibility. Additional guidance is also available on the UK Visa & Immigration website.
[Insert School/Division website]
For informal enquiries about this job contact Andrew Swift, Professor of Cardiothoracic Radiology, a.j.swift@sheffield.ac.uk.
The selection process will take place following the closing date. This will consist of an interview and a short technical/research presentation or practical discussion relevant to medical‑imaging AI. We will let candidates know if they have progressed to the selection stage following the closing date. If you need any support, equipment or adjustments to enable you to participate in any element of the recruitment process you can contact ClinMed‑Staffing@sheffield.ac.uk.
We are the University of Sheffield. This is our vision: sheffield.ac.uk/vision (opens in new window).
- A minimum of 41 days annual leave including bank holiday and closure days (pro rata) with the ability to purchase more.
- Flexible working opportunities, including hybrid working for some roles.
- Generous pension scheme.
- A wide range of discounts and rewards on shopping, eating out and travel.
- A variety of staff networks, providing opportunities for social interaction, peer support and personal development (for example, Race Equality, LGBT+, Women’s and Parent’s networks).
- Recognition Awards to reward staff who go above and beyond in their role.
- A commitment to your development, with access to learning and mentoring schemes.
- A range of generous family‑friendly policies
- paid time off for parenting and caring emergencies
- and more
Closing Date : 05/10/2026
We are a research university with a global reputation for excellence. Our ideas and expertise change the world for the better, making a real difference to society. We know that when people come together with different views, approaches and insights it can lead to richer, more creative and innovative teaching and research and the highest levels of student experience. Our University Vision (www.sheffield.ac.uk/vision) outlines our commitment to building a diverse community of staff and students that recognises and values the abilities, backgrounds, beliefs and ways of living for everyone.
'/%3e%3cpath%20d='M101.744%2012.4726L102.894%2016.1947L99.0021%2013.2201L101.744%2012.4726ZM105.3%208.40937H100.488L99.0026%203.59473L97.5116%208.41047L92.7%208.40385L96.5965%2011.3834L95.1055%2016.1942L99.0021%2013.2196L101.408%2011.3834L105.3%208.40992V8.40937Z'%20fill='white'/%3e%3cdefs%3e%3clinearGradient%20id='paint0_linear_5593_1381'%20x1='90'%20y1='9.44471'%20x2='99'%20y2='9.44471'%20gradientUnits='userSpaceOnUse'%3e%3cstop%20stop-color='%2300AD70'/%3e%3cstop%20offset='0.998594'%20stop-color='%2300B67A'/%3e%3cstop%20offset='1'%20stop-color='%23DCDCE6'/%3e%3c/linearGradient%3e%3c/defs%3e%3c/svg%3e){kind=small}
