Detection of joint inflammation in rheumatoid arthritis using multispectral diffuse optical imaging

Lighter, Daniel (2020). Detection of joint inflammation in rheumatoid arthritis using multispectral diffuse optical imaging. University of Birmingham. Ph.D.

Text - Accepted Version
Available under License All rights reserved.

Download (7MB) | Preview


Rheumatoid arthritis is a chronic autoimmune disease, characterised by joint inflammation, which if untreated causes disability. A clinical need exists for novel, low-cost and noninvasive imaging tools capable of detecting inflammation in the joints for the diagnosis
and monitoring of patients with rheumatoid arthritis. Diffuse optical imaging provides information about the underlying functional properties of biological tissue and previous studies have reported an optical contrast between inflamed and non-inflamed joints, with former displaying localised increases in absorption and scattering attributed to underlying pathophysiological changes. In this work, a novel, multispectral diffuse optical imaging system for imaging human hand joints was presented, which combined surface imaging and optical transmission imaging in a single work-flow to reconstruct maps of clinically relevant parameters such as oxygen saturation, total haemoglobin, water and scattering amplitude in three dimensions. The system was designed to provide accurate, robust and rapid data acquisition, particularly through the novel application to joint imaging of a galvanometer-based unit for fast source repositioning allowing full datasets to be acquired in 2mins per joint, such as to be sufficient for implementation in a clinical setting. This clinical prototype system was then comprehensively studied through experiments involving
biological tissue mimicking optical phantoms, to assess performance against a ground truth set of known parameters.
Preliminary studies involving healthy volunteers gave useful insight into the systems in vivo performance and provided a good understanding of baseline values in healthy subjects, with significantly greater variability observed between subjects than when comparing joints within the same subject. A pilot clinical study was then carried out, involving 144 joints from 21 rheumatology patients with ultrasound imaging and clinical examination as reference comparisons, to assess the systems diagnostic accuracy capabilities. A degree of sensitivity was observed from three dimensional maps of total haemoglobin and scattering amplitude to pathophysiological changes in the joint during longitudinal monitoring of
either recovery from acute injury in a single healthy subject or the response to therapy in rheumatoid arthritis patients. From single time-point examination data, classification accuracies when considering the entire cohort were limited, with areas under the receiver
operator curve of up to 0.657 achieved, with similar conclusions reached to those in comparable single-wavelength, continuous-wave studies previously reported despite the multiple wavelength acquisition. A normalised Fourier transform methodology was then presented, engineered to extract features related to the spatial signature of the transmitted light through the joint that were less sensitive to inter-subject variability in total flux for the assessment of optical transmission images. For the first time within the academic community, to the authors knowledge, the impact on diffuse optical imaging signals of the spatially asymmetrical prevalence of inflammation in hand joints of rheumatoid arthritis patients was addressed. In distinction from previous work, optical images were acquired from the dorsal side with illumination on the palmar side and results when using the proposed normalised fast Fourier transform methodology demonstrated accurate detection of inflamed joints from single time-point examinations, with of area under the receiver operator curve values up to 0.888 together with sensitivities and specificities of up to 77.9% and 90.9% respectively achieved for this specific dataset. This work-flow may enable future development of clinically viable, low-cost devices for assessing inflammation in arthritis patients, without the need for cuff occlusion or comparison to baseline. It will be important to assess the generalisation of these accuracies in future work, using a larger patient cohort and testing different machine learning classification schemes.

Type of Work: Thesis (Doctorates > Ph.D.)
Award Type: Doctorates > Ph.D.
Licence: All rights reserved
College/Faculty: Colleges (2008 onwards) > College of Engineering & Physical Sciences
School or Department: Physical Sciences for Health Doctoral Training Centre
Funders: Engineering and Physical Sciences Research Council
Subjects: T Technology > TA Engineering (General). Civil engineering (General)


Request a Correction Request a Correction
View Item View Item


Downloads per month over past year