Principal Component Analysis in Medical Image Processing
Research type
Research Study
Full title
Principal Component Analysis in Medical Image Processing: a Diagnostic Aid
IRAS ID
179416
Contact name
Jonathan Taylor
Contact email
Sponsor organisation
Sheffield Teaching Hospitals NHS Foundation Trust
Duration of Study in the UK
2 years, 11 months, 28 days
Research summary
The medical imaging domain refers to Computer Aided Diagnosis (CAD) as the use of computer algorithms to aid the image interpretation process. CAD software is designed to give clinicians an independent, objective assessment of certain qualities of an image. For example, it may be used to automatically detect areas of suspected abnormality in an image or to give an overall normal or abnormal classification to an image.
Despite the vast literature available on CAD it is still not widely used in the clinic, particularly in the UK. This is likely to be due to a number of limitations in the evidence base, in particular the lack of robust testing. This study will apply a novel, progressive approach to delivery of CAD, developing a flexible and robust software system that will encourage wider adoption of automated diagnostic image support in the clinic.
The software will be developed using established methods exploited in a new way, with Principal Component Analysis (PCA) augmented by image registration. Through the inclusion of classification algorithms the software will be optimised for use as an automated classifier to assist clinicians with the interpretation of two types of Nuclear Medicine scan: DMSA (a type of kidney scan) and DaTSCAN (a form of brain scan). The software will also be used to demonstrate CAD for the assessment of Low Grade Glioma (LGG) volumes in Magnetic Resonance Imaging (MRI) data.
The software classification functions will be comprehensively validated against simulated and physical phantom images and against databases of real clinical data. Testing within a clinical workflow is a key focus of this work. The validated software applications will be evaluated locally and at two pilot sites by Radiologists. To demonstrate the clinical impact of the software, DMSA reporting accuracy and consistency will be assessed through Radiologist reporting with and without toolkit assistance.
REC name
London - City & East Research Ethics Committee
REC reference
15/LO/0736
Date of REC Opinion
21 Apr 2015
REC opinion
Favourable Opinion