Back to Projects

AI-enabled Imaging

Synergistic image analysis of longitudinal cardiac MRI

Project ID: 2022_001

1st Supervisor: Andrew King, King’s College London
2nd Supervisor: Reza Razavi, King’s College London
Additional supervisor: Esther Puyol-Anton, King’s College London
Clinical supervisor: Bram Ruijsink, King’s College London
Industrial Supervisor: Paul Aljabar, Perspectum

Aim of the PhD Project:

  • Develop and evaluate frameworks for improving the estimation of cardiac biomarkers and their changes from longitudinal MR imaging 
  • Investigate techniques ranging from separate independent analysis to ‘synergistic’ and simultaneous analysis of scans from multiple time points  
  • Establish the repeatability of cardiac biomarker estimation using repeat scan data 
  • Compare longitudinal analysis results to baseline repeatability performance 

Project description/background:

Recent years have seen the emergence of state-of-the-art deep learning segmentations tools, reaching human-level performance on a range of tasks [1]. In cardiac MR, such techniques have been developed to derive imaging biomarkers for fully automated quality-controlled functional quantification [2]. In both clinical practice and research studies, it is common for subjects to be scanned multiple times. The resulting longitudinal data will contain significant similarities, as well as differences due to changes in anatomy and pathology over time as well as variations due to differences in acquisition. However, the image analysis tools that are employed to segment structures and derive biomarkers from imaging data do not typically exploit these similarities and treat the data as if they were independent. There is a need for methods that can sensitively identify biomarker changes, especially when trialling a new drug or treatment, as this would enable the trial size/duration to be reduced, which can have a significant impact on cost and the ability to determine the efficacy of the technology.  

Some other strands of research, such as image reconstruction, have developed ‘synergistic’ approaches for processing longitudinal data [3]. Some related work has also been performed in image analysis, but to date this has mostly focused on exploiting longitudinal data for classification problems such as predicting treatment response [4,5], tumour detection [6,7] or assessing disease severity [8]. 

This project will seek to develop a ‘synergistic’ analysis framework for longitudinal imaging data in a novel application to biomarker estimation. Our focus will be on cardiac MR imaging, but we believe that the methods developed will be applicable to a wider range of modalities and problems, for example the methods developed may be extended to further organs to assess health in multi-organ conditions such as Covid-19. We will aim to exploit the similarities between scans in a synergistic way to enable more accurate and robust estimates to be made of morphological and functional biomarkers from all scans. We will also use repeat-scan data from the same subjects to establish a level of repeatability/reproducibility and evaluate our synergistic approaches against this. To begin with, we will develop methods for synergistically analysing two scans from the same scanner at different time points, but then seek to generalise the framework to multiple scans from multiple scanners at multiple time points. We will predominantly focus on deriving biomarkers via segmentations but will also investigate the possibility of synergistic direct estimation of biomarkers from the imaging data. 

The candidate for this project should have good computational skills and experience in, or a strong desire to learn about, artificial intelligence/deep learning for medical imaging.

 

References

  1. Isensee, et al, “nnU-Net: A Self-configuring Method for Deep Learning-based Biomedical Image Segmentation,” Nat Methods, 18, pp203–211, 2021.
  2. Ruijsink, et al, “Fully Automated, Quality-Controlled Cardiac Analysis From CMR: Validation and Large-Scale Application to Characterize Cardiac Function”, JACC: Cardiovasc Imag, 13(3), pp684-695, 2020.
  3. Ellis and Reader, “Simultaneous maximum a posteriori longitudinal PET image reconstruction,” Phys Med Biol, 62(17), pp6963–6979, 2017.
  4. Jin, et al, “Predicting Treatment Response From Longitudinal Images Using Multi-task Deep Learning,” Nat Commun, 12, pp1851, 2021.
  5. Zhang et al, “Predicting Rectal Cancer Response to Neoadjuvant Chemoradiotherapy Using Deep Learning of Diffusion Kurtosis MRI,” Radiology, 296(1), pp56-64, 2020.
  6. Ardila, et al, “End-to-end Lung Cancer Screening With Three-dimensional Deep Learning on Low-dose Chest Computed Tomography,” Nat Med, 25, pp954–961, 2019.
  7. Salem et al, “A Fully Convolutional Neural Network for New T2-w Lesion Detection in Multiple Sclerosis,” Neuroimage: Clinical, 25(102149), 2020.
  8. Li, et al, “Siamese Neural Networks for Continuous Disease Severity Evaluation and Change Detection in Medical Imaging,” npj Digit Med, 3(48), 2020.

Back to Projects