Computer-aided diagnosis of rheumatoid arthritis with optical tomography, Part 1: feature extraction

Stochastic Gauss-Newton method for estimating absorption and scattering in optical tomography with the Monte Carlo method for light transport

Image reconstruction in optical tomography in the so-called transport regime, where the diffusion approximation is not valid, requires modeling of light transport using the radiative transfer equation. In this work, we approach this problem by utilizing the Monte Carlo method for light transport. In this work, we propose a methodology for absolute imaging of absorption and scattering in this regime utilizing a Monte Carlo method for light transport. The image reconstruction problem is formulated as a minimization problem that is solved using a stochastic Gauss-Newton method. In the construction of the Jacobian matrix for scattering, a perturbation approximation for Monte Carlo is utilized. The approach is evaluated with numerical simulations using an adaptive approach where the number of photon packets is adjusted during the iterations, and with different fixed numbers of photon packets. The simulations show that the Monte Carlo method for light transport can be utilized in the absolute imaging problem of optical tomography and that the absorption and scattering parameters can be estimated simultaneously with good accuracy.

Time-domain diffuse optical imaging technique for monitoring rheumatoid arthritis disease activity: experimental validation in tissue-mimicking finger phantoms

Objective.Effective treatment within 3-5 months of disease onset significantly improves rheumatoid arthritis (RA) prognosis. Nevertheless, 30% of RA patients fail their first treatment, and it takes 3-6 months to identify failure with current monitoring techniques. Time-domain diffuse optical imaging (TD-DOI) may be more sensitive to RA disease activity and could be used to detect treatment failure. In this report, we present the development of a TD-DOI hand imaging system and validate its ability to measure simulated changes in RA disease activity using tissue-mimicking finger phantoms.Approach.A TD-DOI system was built, based on a single-pixel camera architecture, and used to image solid phantoms which mimicked a proximal interphalangeal finger joint. For reference,in silicoimages of virtual models of the solid phantoms were also generated using Monte Carlo simulations. Spatiotemporal Fourier components were extracted from both simulated and experimental images, and their ability to distinguish between phantoms representing different RA disease activity was quantified.Main results.Many spatiotemporal Fourier components extracted from TD-DOI images could clearly distinguish between phantoms representing different states of RA disease activity.Significance.A TD-DOI system was built and validated using finger-mimicking solid phantoms. The findings suggest that the system could be used to monitor RA disease activity. This single-pixel TD-DOI system could be used to acquire longitudinal measures of RA disease activity to detect early treatment failure.

Time-domain diffuse optical imaging technique for monitoring rheumatoid arthritis disease activity: theoretical development and in silico validation

Objective.Effective early treatment-within 3-5 months of disease onset-significantly improves rheumatoid arthritis (RA) prognosis. Nevertheless, 1 in 3 patients experiences treatment failure which takes 3-6 months to detect with current monitoring techniques. The aim of this work is to develop a method for extracting quantitative features from data obtained with time-domain diffuse optical imaging (TD-DOI), and demonstrate their sensitivity to RA disease activity.Approach.80 virtual phantoms of the proximal interphalangeal joint-obtained from a realistic tissue distribution derived from magnetic resonance imaging-were created to simulate RA-induced alterations in 5 physiological parameters. TD-DOI images were generated using Monte Carlo simulations, and Poisson noise was added to each image. Subsequently, each image was convolved with an instrument response function (IRF) to mimic experimental measurements. Various spatiotemporal features were extracted from the images (i.e. statistical moments, temporal Fourier components), corrected for IRF effects, and correlated with the disease index (DI) of each phantom.Main results.Spatiotemporal Fourier components of TD-DOI images were highly correlated with DI despite the confounding effects of noise and the IRF. Moreover, lower temporal frequency components (⩽0.4 GHz) demonstrated greater sensitivity to small changes in disease activity than previously published spatial features extracted from the same images.Significance.Spatiotemporal components of TD-DOI images may be more sensitive to small changes in RA disease activity than previously reported DOI features. TD-DOI may enable earlier detection of RA treatment failure, which would reduce the time needed to identify treatment failure and improve patient prognosis.

Diagnostic Evaluation of Rheumatoid Arthritis (RA) in Finger Joints Based on the Third-Order Simplified Spherical Harmonics (SP3) Light Propagation Model

This work focuses on the evaluation of third-order simplified spherical harmonics (SP3) model-based image reconstruction with respect to its clinical utility to diagnose rheumatoid arthritis (RA). The existing clinical data of 219 fingers was reconstructed for both absorption and scattering maps in fingers by using the reduced-Hessian sequential quadratic programming (rSQP) algorithm that employs the SP3 model of light propagation. The k-fold cross validation method was used for feature extraction and classification of SP3-based tomographic images. The performance of the SP3 model was compared to the DE and ERT models with respect to diagnostic accuracy and computational efficiency. The results presented here show that the SP3 model achieves clinically relevant sensitivity (88%) and specificity (93%) that compare favorably to the ERT while maintaining significant computational advantage over the ERT (i.e., the SP3 model is 100 times faster than the ERT). Furthermore, it is also shown that the SP3 is similar in speed but superior in diagnostic accuracy to the DE. Therefore, it is expected that the method presented here can greatly aid in the early diagnosis of RA with clinically relevant accuracy in near real-time at a clinical setting.

Quantum-Inspired Interpertable AI-Empowered Decision Support System for Detection of Early-Stage Rheumatoid Arthritis in Primary Care Using Scarce Dataset

Rheumatoid arthritis (RA) is a chronic inflammatory and long-term autoimmune disease that can lead to joint and bone erosion. This can lead to patients’ disability if not treated in a timely manner. Early detection of RA in settings such as primary care (as the first contact with patients) can have an important role on the timely treatment of the disease. We aim to develop a web-based Decision Support System (DSS) to provide a proper assistance for primary care providers in early detection of RA patients. Using Sparse Fuzzy Cognitive Maps, as well as quantum-learning algorithm, we developed an online web-based DSS to assist in early detection of RA patients, and subsequently classify the disease severity into six different levels. The development process was completed in collaborating with two specialists in orthopedic as well as rheumatology orthopedic surgery. We used a sample of anonymous patient data for development of our model which was collected from Shohada University Hospital, Tabriz, Iran. We compared the results of our model with other machine learning methods (e.g., linear discriminant analysis, Support Vector Machines, and K-Nearest Neighbors). In addition to outperforming other methods of machine learning in terms of accuracy when all of the clinical features are used (accuracy of 69.23%), our model identified the relation of the different features with each other and gave higher explainability comparing to the other methods. For future works, we suggest applying the proposed model in different contexts and comparing the results, as well as assessing its usefulness in clinical practice.

Practical Implementation of Artificial Intelligence-Based Deep Learning and Cloud Computing on the Application of Traditional Medicine and Western Medicine in the Diagnosis and Treatment of Rheumatoid Arthritis

Rheumatoid arthritis (RA), an autoimmune disease of unknown etiology, is a serious threat to the health of middle-aged and elderly people. Although western medicine, traditional medicine such as traditional Chinese medicine, Tibetan medicine and other ethnic medicine have shown certain advantages in the diagnosis and treatment of RA, there are still some practical shortcomings, such as delayed diagnosis, improper treatment scheme and unclear drug mechanism. At present, the applications of artificial intelligence (AI)-based deep learning and cloud computing has aroused wide attention in the medical and health field, especially in screening potential active ingredients, targets and action pathways of single drugs or prescriptions in traditional medicine and optimizing disease diagnosis and treatment models. Integrated information and analysis of RA patients based on AI and medical big data will unquestionably benefit more RA patients worldwide. In this review, we mainly elaborated the application status and prospect of AI-assisted deep learning and cloud computation-oriented western medicine and traditional medicine on the diagnosis and treatment of RA in different stages. It can be predicted that with the help of AI, more pharmacological mechanisms of effective ethnic drugs against RA will be elucidated and more accurate solutions will be provided for the treatment and diagnosis of RA in the future.

Flexible optical imaging band system for the assessment of arthritis in patients with systemic lupus erythematosus

We have developed a flexible optical imaging system (FOIS) to assess systemic lupus erythematosus (SLE) arthritis in the finger joints. While any part of the body can be affected, arthritis in the finger joints is one of the most common SLE manifestations. There is an unmet need for accurate, low-cost assessment of lupus arthritis that can be easily performed at every clinic visit. Current imaging methods are imprecise, expensive, and time consuming to allow for frequent monitoring. Our FOIS can be wrapped around joints, and multiple light sources and detectors gather reflected and transmitted light intensities. Using data from two SLE patients and two healthy volunteers, we demonstrate the potential of this FOIS for assessment of arthritis in SLE patients.

ChronoMID-Cross-modal neural networks for 3-D temporal medical imaging data

ChronoMID-neural networks for temporally-varying, hence Chrono, Medical Imaging Data-makes the novel application of cross-modal convolutional neural networks (X-CNNs) to the medical domain. In this paper, we present multiple approaches for incorporating temporal information into X-CNNs and compare their performance in a case study on the classification of abnormal bone remodelling in mice. Previous work developing medical models has predominantly focused on either spatial or temporal aspects, but rarely both. Our models seek to unify these complementary sources of information and derive insights in a bottom-up, data-driven approach. As with many medical datasets, the case study herein exhibits deep rather than wide data; we apply various techniques, including extensive regularisation, to account for this. After training on a balanced set of approximately 70000 images, two of the models-those using difference maps from known reference points-outperformed a state-of-the-art convolutional neural network baseline by over 30pp (> 99% vs. 68.26%) on an unseen, balanced validation set comprising around 20000 images. These models are expected to perform well with sparse data sets based on both previous findings with X-CNNs and the representations of time used, which permit arbitrarily large and irregular gaps between data points. Our results highlight the importance of identifying a suitable description of time for a problem domain, as unsuitable descriptors may not only fail to improve a model, they may in fact confound it.

Assessing spectral imaging of the human finger for detection of arthritis

Rheumatoid arthritis causes changes in the optical properties of tissues in the joints, which could be detected using spectral imaging. This has the potential for development of low cost, non-contact method for early detection of the disease. In this work, hyperspectral imaging system was used to obtain 24 images of proximal interphalangeal joints of 12 healthy volunteers. A large inter-subject variability was observed, but still an increase in transmittance in the spectral range of 600 nm - 950 nm could be associated to the joint in all images. The results of experiments were compared to detailed simulations of light propagation trough tissue. For the simulations, voxelized 3D models of unaffected and inflamed human joints with realistic tissue distributions were constructed from an in-vivo MRI scan of a healthy human finger. The simulated model of healthy finger successfully reproduced the experimental data, while the affected models indicated that the inflammation introduces detectable differences in the spectral and spatial features. The results were used to guide the design of a dedicated imaging system for detection of rheumatoid arthritis, that will be used in an upcoming clinical study.

Detecting inflammation in rheumatoid arthritis using Fourier transform analysis of dorsal optical transmission images from a pilot study

A clinical need exists for low-cost and noninvasive imaging tools capable of detecting inflammation in the joints of inflammatory arthritis patients. Previous studies have reported an optical contrast between inflamed and noninflamed joints resulting from distinct absorption and scattering properties. Accurate classification using nonocclusion-based continuous wave, transillumination imaging was limited to patient-specific changes during follow-up examination as opposed to single time-point examination, which was attributed to high intersubject variability. In distinction from previous work, optical images were acquired from the dorsal side with illumination on the palmar side and features about the spatial distribution of transmitted light along the joint were assessed using a normalized Fourier transform method. Results using this approach demonstrated an area under receiver operator curve of up to 0.888 for detecting inflammation in a pilot study involving single time-point examination of 144 joints from 21 rheumatology patients. This workflow 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.

Multispectral, non-contact diffuse optical tomography of healthy human finger joints

Rheumatoid arthritis (RA) is an inflammatory joint disease often affecting the hands, which if untreated causes disability. Diffuse optical tomography (DOT) provides information about the underlying functional properties of biological tissue. To detect pathophysiological changes in inflamed RA joints, a good understanding of the baseline values for healthy subjects is first required. Finger joints from healthy subjects were imaged using a non-contact, multispectral, continuous wave DOT system, recovering physiological parameters of oxygen saturation, total haemoglobin, water concentration and scatter amplitude. Reconstructed values across the cohort demonstrated good consistency between finger joints from the same participant, with greater variation seen between subjects.

Frequency-domain optical tomographic image reconstruction algorithm with the simplified spherical harmonics (SP3) light propagation model

We introduce here the finite volume formulation of the frequency-domain simplified spherical harmonics model with n-th order absorption coefficients (FD-SP_N) that approximates the frequency-domain equation of radiative transfer (FD-ERT). We then present the FD-SP_N based reconstruction algorithm that recovers absorption and scattering coefficients in biological tissue. The FD-SP_N model with 3^rd order absorption coefficient (i.e., FD-SP₃) is used as a forward model to solve the inverse problem. The FD-SP₃ is discretized with a node-centered finite volume scheme and solved with a restarted generalized minimum residual (GMRES) algorithm. The absorption and scattering coefficients are retrieved using a limited-memory Broyden-Fletcher-Goldfarb-Shanno (L-BFGS) algorithm. Finally, the forward and inverse algorithms are evaluated using numerical phantoms with optical properties and size that mimic small-volume tissue such as finger joints and small animals. The forward results show that the FD-SP₃ model approximates the FD-ERT (S₁₂) solution within relatively high accuracy; the average error in the phase (<3.7%) and the amplitude (<7.1%) of the partial current at the boundary are reported. From the inverse results we find that the absorption and scattering coefficient maps are more accurately reconstructed with the SP₃ model than those with the SP₁ model. Therefore, this work shows that the FD-SP₃ is an efficient model for optical tomographic imaging of small-volume media with non-diffuse properties both in terms of computational time and accuracy as it requires significantly lower CPU time than the FD-ERT (S₁₂) and also it is more accurate than the FD-SP₁.

In Vivo, Non-Invasive Characterization of Human Bone by Hybrid Broadband (600-1200 nm) Diffuse Optical and Correlation Spectroscopies

Non-invasive in vivo diffuse optical characterization of human bone opens a new possibility of diagnosing bone related pathologies. We present an in vivo characterization performed on seventeen healthy subjects at six different superficial bone locations: radius distal, radius proximal, ulna distal, ulna proximal, trochanter and calcaneus. A tailored diffuse optical protocol for high penetration depth combined with the rather superficial nature of considered tissues ensured the effective probing of the bone tissue. Measurements were performed using a broadband system for Time-Resolved Diffuse Optical Spectroscopy (TRS) to assess mean absorption and reduced scattering spectra in the 600-1200 nm range and Diffuse Correlation Spectroscopy (DCS) to monitor microvascular blood flow. Significant variations among tissue constituents were found between different locations; with radius distal rich of collagen, suggesting it as a prominent location for bone related measurements, and calcaneus bone having highest blood flow among the body locations being considered. By using TRS and DCS together, we are able to probe the perfusion and oxygen consumption of the tissue without any contrast agents. Therefore, we predict that these methods will be able to evaluate the impairment of the oxygen metabolism of the bone at the point-of-care.

Joint blood flow is more sensitive to inflammatory arthritis than oxyhemoglobin, deoxyhemoglobin, and oxygen saturation

Joint hypoxia plays a central role in the progression and perpetuation of rheumatoid arthritis (RA). Thus, optical techniques that can measure surrogate markers of hypoxia such as blood flow, oxyhemoglobin, deoxyhemoglobin, and oxygen saturation are being developed to monitor RA. The purpose of the current study was to compare the sensitivity of these physiological parameters to arthritis. Experiments were conducted in a rabbit model of RA and the results revealed that joint blood flow was the most sensitive to arthritis and could detect a statistically significant difference (p<0.05, power = 0.8) between inflamed and healthy joints with a sample size of only four subjects. Considering that this a quantitative technique, the high sensitivity to arthritis suggests that joint perfusion has the potential to become a potent tool for monitoring disease progression and treatment response in RA.

Overview of diffuse optical tomography and its clinical applications

Near-infrared diffuse optical tomography (DOT), one of the most sophisticated optical imaging techniques for observations through biological tissue, allows 3-D quantitative imaging of optical properties, which include functional and anatomical information. With DOT, it is expected to be possible to overcome the limitations of conventional near-infrared spectroscopy (NIRS) as well as offering the potential for diagnostic optical imaging. However, DOT has been under development for more than 30 years, and the difficulties in development are attributed to the fact that light is strongly scattered and that diffusive photons are used for the image reconstruction. The DOT algorithm is based on the techniques of inverse problems. The radiative transfer equation accurately describes photon propagation in biological tissue, while, because of its high computation load, the diffusion equation (DE) is often used as the forward model. However, the DE is invalid in low-scattering and/or highly absorbing regions and in the vicinity of light sources. The inverse problem is inherently ill-posed and highly undetermined. Here, we first summarize NIRS and then describe various approaches in the efforts to develop accurate and efficient DOT algorithms and present some examples of clinical applications. Finally, we discuss the future prospects of DOT.

Hemodynamic signals in fNIRS

Near-infrared spectroscopy (NIRS) was originally designed for clinical monitoring of tissue oxygenation, and it has also been developed into a useful tool in neuroimaging studies, with the so-called functional NIRS (fNIRS). With NIRS, cerebral activation is detected by measuring the cerebral hemoglobin (Hb), where however, the precise correlation between NIRS signal and neural activity remains to be fully understood. This can in part be attributed to the situation that NIRS signals are inherently subject to contamination by signals arising from extracerebral tissue. In recent years, several approaches have been investigated to distinguish between NIRS signals originating in cerebral tissue and signals originating in extracerebral tissue. Selective measurements of cerebral Hb will enable a further evolution of fNIRS. This chapter is divided into six sections: first a summary of the basic theory of NIRS, NIRS signals arising in the activated areas, correlations between NIRS signals and fMRI signals, correlations between NIRS signals and neural activities, and the influence of a variety of extracerebral tissue on NIRS signals and approaches to this issue are reviewed. Finally, future prospects of fNIRS are described.

Fast linear solver for radiative transport equation with multiple right hand sides in diffuse optical tomography

It is well known that radiative transfer equation (RTE) provides more accurate tomographic results than its diffusion approximation (DA). However, RTE-based tomographic reconstruction codes have limited applicability in practice due to their high computational cost. In this article, we propose a new efficient method for solving the RTE forward problem with multiple light sources in an all-at-once manner instead of solving it for each source separately. To this end, we introduce here a novel linear solver called block biconjugate gradient stabilized method (block BiCGStab) that makes full use of the shared information between different right hand sides to accelerate solution convergence. Two parallelized block BiCGStab methods are proposed for additional acceleration under limited threads situation. We evaluate the performance of this algorithm with numerical simulation studies involving the Delta-Eddington approximation to the scattering phase function. The results show that the single threading block RTE solver proposed here reduces computation time by a factor of 1.5~3 as compared to the traditional sequential solution method and the parallel block solver by a factor of 1.5 as compared to the traditional parallel sequential method. This block linear solver is, moreover, independent of discretization schemes and preconditioners used; thus further acceleration and higher accuracy can be expected when combined with other existing discretization schemes or preconditioners.

Hyperspectral imaging for detection of arthritis: feasibility and prospects

Rheumatoid arthritis (RA) is a disease that frequently leads to joint destruction. It has a high incidence rate worldwide, and the disease significantly reduces patients’ quality of life. Detecting and treating inflammatory arthritis before structural damage to the joint has occurred is known to be essential for preventing patient disability and pain. Existing diagnostic technologies are expensive, time consuming, and require trained personnel to collect and interpret data. Optical techniques might be a fast, noninvasive alternative. Hyperspectral imaging (HSI) is a noncontact optical technique which provides both spectral and spatial information in one measurement. In this study, the feasibility of HSI in arthritis diagnostics was explored by numerical simulations and optimal imaging parameters were identified. Hyperspectral reflectance and transmission images of RA and normal human joint models were simulated using the Monte Carlo method. The spectral range was 600 to 1100 nm. Characteristic spatial patterns for RA joints and two spectral windows with transmission were identified. The study demonstrated that transmittance images of human joints could be used as one parameter for discrimination between arthritic and unaffected joints. The presented work shows that HSI is a promising imaging modality for the diagnostics and follow-up monitoring of arthritis in small joints.

Computer-aided diagnosis of rheumatoid arthritis with optical tomography, Part 2: image classification

This is the second part of a two-part paper on the application of computer-aided diagnosis to diffuse optical tomography (DOT) for diagnosing rheumatoid arthritis (RA). A comprehensive analysis of techniques for the classification of DOT images of proximal interphalangeal joints of subjects with and without RA is presented. A method for extracting heuristic features from DOT images was presented in Part 1. The ability of five classification algorithms to accurately label each DOT image as belonging to a subject with or without RA is analyzed here. The algorithms of interest are the k-nearest-neighbors, linear and quadratic discriminant analysis, self-organizing maps, and support vector machines (SVM). With a polynomial SVM classifier, we achieve 100.0% sensitivity and 97.8% specificity. Lower bounds for these results (at 95.0% confidence level) are 96.4% and 93.8%, respectively. Image features most predictive of RA are from the spatial variation of optical properties and the absolute range in feature values. The optimal classifiers are low-dimensional combinations (<7 features). These results underscore the high potential for DOT to become a clinically useful diagnostic tool and warrant larger prospective clinical trials to conclusively demonstrate the ultimate clinical utility of this approach.