MALUS-HF Version 2.1
Research type
Research Study
Full title
Machine Learning–Assisted Lung Ultrasound for Detection of Pulmonary Congestion in Heart Failure (MALUS_HF)
IRAS ID
368897
Contact name
Andrew Walden
Contact email
Sponsor organisation
Royal Berkshire hospital
Duration of Study in the UK
1 years, 6 months, 7 days
Research summary
Heart failure affects approximately 1 million people in the UK, with around 200,000 new diagnoses each year. It is one of the commonest reasons for emergency hospital admission in older adults. When heart failure worsens, fluid can build up in the lungs (pulmonary congestion), causing breathlessness and requiring treatment with diuretic (“water”) medication. Accurately assessing lung congestion is important because both under-treatment and over-treatment can be harmful.
At present, assessment usually involves clinical examination with a stethoscope, blood tests, and chest X-rays, but these methods can be unreliable for detecting early or mild congestion. Lung ultrasound is a more accurate bedside test for identifying fluid in the lungs, but it currently requires specialist training and experience to perform and interpret.
The purpose of this study is to explore whether computer-based image analysis (“machine learning”) can automatically identify features of pulmonary congestion from lung ultrasound images. If successful, this could help make lung ultrasound easier to use in routine care and potentially improve assessment of patients with heart failure in hospital and community settings.
30 adult participants will take part in the study. This will include patients with heart failure who are undergoing lung ultrasound as part of their routine clinical care, as well as a small number of participants without heart failure whose scans will provide comparison images.
Participants will not undergo any additional treatments or invasive procedures for the study. Ultrasound images collected during routine care will be stored in anonymised form and analysed by the research team. Relevant clinical information, such as observations and heart function test results, will also be collected from the medical record. The study will assess whether ultrasound images can be reliably analysed using machine learning techniques and will help inform the design of larger future studies.
REC name
East Midlands - Derby Research Ethics Committee
REC reference
26/EM/0135
Date of REC Opinion
9 Jul 2026
REC opinion
Further Information Favourable Opinion