Patrick Riley - PhD Student
LJMU Machine Learning Research Group
Research project
My PhD Research project is studying diagnosis of prostate cancer, through multiparametric MRI. This work extends to clinical decision support systems, including research in breast cancer with mammography images. Diagnosis of prostate cancer through MRI is the second line of defence, after an overspecific test (PSA blood test & TRUS biopsy). Early detection of prostate cancer is becoming increasingly important, with two-fifths of prostate cancers diagnosed at a late stage, according to the chartity Orchid (UK). MRI diagnosis is progressing to the first line of diagnosis, with clinical trials taking place within the UK. With an abundance of available clinical data, particularly multiparametric MRI, it may be challenging to know where to start. My research project focuses on both the abundance of clinical data available from MRI scans and areas in demand from clinicians.
Research Hypothesis Applying machine learning methodologies that are able to deal with multiparametric MRI, in an unsupervised or a semi-supervised manner, can improve the diagnosis and segmentation of prostate cancer.
Research interests I am interested in machine learning methodologies and interpretability, allowing for meaningful and impactful results. Previous work has looked at using methods such as Support Vector Machines for classification of prostate lesions from publically available datasets, such as PROSTATEx. This work is being conducted in collaboration with the Clatterbridge Cancer Centre, to progress towards real-world clinical impact.
Publications
2019
- Marnell, S; Riley, P; Olier, I; Rea, M; Ortega-Martorell, S (2019). A Comparative Assessment of Feed-Forward and Convolutional Neural Networks to assist in the diagnosis of prostate cancer. doi: https://doi.org/10.1007/978-3-030-33617-2_15
- Ortega-Martorell, S; Candiota, AP; Thomson, R; Riley, P; Julia-Sape, M; Olier, I (2019). Embedding MRI information into MRSI data source extraction improves brain tumour delineation in animal models. PLoS ONE 14(8): e0220809. doi: https://doi.org/10.1371/journal.pone.0220809
- Riley, P; Olier, I; Rea, M; Lisboa, P; Ortega-Martorell, S (2019). A Voting Ensemble Method to Assist the Diagnosis of Prostate Cancer Using Multiparametric MRI. doi: https://doi.org/10.1007/978-3-030-19642-4_29
- Srivastava, M; Olier, I; Riley, P; Lisboa, P; Ortega-Martorell, S (2019). Classifying and Grouping Mammography Images into Communities Using Fisher Information Networks to Assist the Diagnosis of Breast Cancer. doi: https://doi.org/10.1007/978-3-030-19642-4_30
Presentations
2019
- IDEAL 2019 Conference. A Comparative Assessment of Feed-Forward and Convolutional Neural Networks to assist in the diagnosis of prostate cancer.
- WSOM+ 2019 conference. A Voting Ensemble Method to Assist the Diagnosis of Prostate Cancer Using Multiparametric MRI
Awards
- Best poster - LJMU Faculty Research Week, 2019.
- Best potential impact - LJMU Faculty Research Week, 2019.
Biography
I attained both my Data Science MSc and my Mathematics BSc (Hons), at Liverpool John Moores University, (LJMU). During the MSc I completed a Curriculum Enhancement Internship for creating better clustering analysis tools in R. In 2018 I attained acceptance onto my PhD within the Machine Learning Research Group at LJMU; my supervisor is Dr Sandra Ortega Martorell
Skills
Programming and coding I am proficient in using the R programming language, in which I have completed various research projects with data science and machine leraning techniques. This includes R Markdown and some skills in development in R Shiny. I have acquired skills in Python and MATLAB.
Previous degrees
2018 - Data Science MSc, Liverpool John Moores University. Grade: Distinction.
2017 - Mathematics BSc (Hons), Liverpool John Moores University. Grade: 1st Class.
Outreach & teaching activities
I have conducted a set of voluntry workshops within the Department of Applied Mathematics at LJMU, to deliver content on R/RStudio to multidisciplinary students - to those with, and without, prior programming knowledge. Students have fed back positively on the workshops. Topics include a crash course as well as material on ggplot2 and dplyr.
For 2019/20 I am President of the Faculty Researcher Representative Society. From 2019 I have worked alongside the Outreach team at the university part time as a Student Advocate, assisting in various duties around the unviersity and with prospective applicants.
Funding
I am funded by the Liverpool John Moores University PhD Scholarship fund.
