Machine learning prediction of future amyloid beta positivity in amyloid-negative individuals.

BACKGROUND: The pathophysiology of Alzheimer's disease (AD) involves ß -amyloid (A ß ) accumulation. Early identification of individuals with abnormal ß -amyloid levels is crucial, but A ß quantification with positron emission tomography (PET) and cerebrospinal fluid (CSF) is invasive and expensive. METHODS: We propose a machine learning framework using standard non-invasive (MRI, demographics, APOE, neuropsychology) measures to predict future A ß -positivity in A ß -negative individuals. We separately study A ß -positivity defined by PET and CSF. RESULTS: Cross-validated AUC for 4-year A ß conversion prediction was 0.78 for the CSF-based and 0.68 for the PET-based A ß definitions. Although not trained for the clinical status-change prediction, the CSF-based model excelled in predicting future mild cognitive impairment (MCI)/dementia conversion in cognitively normal/MCI individuals (AUCs, respectively, 0.76 and 0.89 with a separate dataset). CONCLUSION: Standard measures have potential in detecting future A ß -positivity and assessing conversion risk, even in cognitively normal individuals. The CSF-based definition led to better predictions than the PET-based definition.

Near-Infrared Spectroscopy Enables Arthroscopic Histologic Grading of Human Knee Articular Cartilage.

Purpose: To develop the means to estimate cartilage histologic grades and proteoglycan content in ex vivo arthroscopy using near-infrared spectroscopy (NIRS). Methods: In this experimental study, arthroscopic NIR spectral measurements were performed on both knees of 9 human cadavers, followed by osteochondral block extraction and in vitro measurements: reacquisition of spectra and reference measurements (proteoglycan content, and three histologic scores). A hybrid model, combining principal component analysis and linear mixed-effects model (PCA-LME), was trained for each reference to investigate its relationship with in vitro NIR spectra. The performance of the PCA-LME model was validated with ex vivo spectra before and after the exclusion of outlying spectra. Model performance was evaluated based on Spearman rank correlation (ρ) and root-mean-square error (RMSE). Results: The PCA-LME models performed well (independent test: average ρ = 0.668, RMSE = 0.892, P < .001) in the prediction of the reference measurements based on in vitro data. The performance on ex vivo arthroscopic data was poorer but improved substantially after outlier exclusion (independent test: average ρ = 0.462 to 0.614, RMSE = 1.078 to 0.950, P = .019 to .008). Conclusions: NIRS is capable of nondestructive evaluation of cartilage integrity (i.e., histologic scores and proteoglycan content) under similar conditions as in clinical arthroscopy. Clinical Relevance: There are clear clinical benefits to the accurate assessment of cartilage lesions in arthroscopy. Visual grading is the current standard of care. However, optical techniques, such as NIRS, may provide a more objective assessment of cartilage damage.

Near infrared spectroscopic evaluation of biochemical and crimp properties of knee joint ligaments and patellar tendon.

Knee ligaments and tendons play an important role in stabilizing and controlling the motions of the knee. Injuries to the ligaments can lead to abnormal mechanical loading of the other supporting tissues (e.g., cartilage and meniscus) and even osteoarthritis. While the condition of knee ligaments can be examined during arthroscopic repair procedures, the arthroscopic evaluation suffers from subjectivity and poor repeatability. Near infrared spectroscopy (NIRS) is capable of non-destructively quantifying the composition and structure of collagen-rich connective tissues, such as articular cartilage and meniscus. Despite the similarities, NIRS-based evaluation of ligament composition has not been previously attempted. In this study, ligaments and patellar tendon of ten bovine stifle joints were measured with NIRS, followed by chemical and histological reference analysis. The relationship between the reference properties of the tissue and NIR spectra was investigated using partial least squares regression. NIRS was found to be sensitive towards the water (R2CV = .65) and collagen (R2CV = .57) contents, while elastin, proteoglycans, and the internal crimp structure remained undetectable. As collagen largely determines the mechanical response of ligaments, we conclude that NIRS demonstrates potential for quantitative evaluation of knee ligaments.

Functional and structural properties of human patellar articular cartilage in osteoarthritis.

Changes in the fibril-reinforced poroelastic (FRPE) mechanical material parameters of human patellar cartilage at different stages of osteoarthritis (OA) are not known. Further, the patellofemoral joint loading is thought to include more sliding and shear compared to other knee joint locations, thus, the relations between structural and functional changes may differ in OA. Thus, our aim was to determine the patellar cartilage FRPE properties followed by associating them with the structure and composition. Osteochondral plugs (n = 14) were harvested from the patellae of six cadavers. Then, the FRPE material properties were determined, and those properties were associated with proteoglycan content, collagen fibril orientation angle, optical retardation (fibril parallelism), and the state of OA of the samples. The initial fibril network modulus and permeability strain-dependency factor were 72% and 63% smaller in advanced OA samples when compared to early OA samples. Further, we observed a negative association between the initial fibril network modulus and optical retardation (r = -0.537, p < 0.05). We also observed positive associations between 1) the initial permeability and optical retardation (r = 0.547, p < 0.05), and 2) the initial fibril network modulus and optical density (r = 0.670, p < 0.01).These results suggest that the reduced pretension of the collagen fibrils, as shown by the reduced initial fibril network modulus, is linked with the loss of proteoglycans and cartilage swelling in human patellofemoral OA. The characterization of these changes is important to improve the representativeness of knee joint models in tissue and cell scale.

Transfer Learning in Magnetic Resonance Brain Imaging: A Systematic Review.

(1) Background: Transfer learning refers to machine learning techniques that focus on acquiring knowledge from related tasks to improve generalization in the tasks of interest. In magnetic resonance imaging (MRI), transfer learning is important for developing strategies that address the variation in MR images from different imaging protocols or scanners. Additionally, transfer learning is beneficial for reutilizing machine learning models that were trained to solve different (but related) tasks to the task of interest. The aim of this review is to identify research directions, gaps in knowledge, applications, and widely used strategies among the transfer learning approaches applied in MR brain imaging; (2) Methods: We performed a systematic literature search for articles that applied transfer learning to MR brain imaging tasks. We screened 433 studies for their relevance, and we categorized and extracted relevant information, including task type, application, availability of labels, and machine learning methods. Furthermore, we closely examined brain MRI-specific transfer learning approaches and other methods that tackled issues relevant to medical imaging, including privacy, unseen target domains, and unlabeled data; (3) Results: We found 129 articles that applied transfer learning to MR brain imaging tasks. The most frequent applications were dementia-related classification tasks and brain tumor segmentation. The majority of articles utilized transfer learning techniques based on convolutional neural networks (CNNs). Only a few approaches utilized clearly brain MRI-specific methodology, and considered privacy issues, unseen target domains, or unlabeled data. We proposed a new categorization to group specific, widely-used approaches such as pretraining and fine-tuning CNNs; (4) Discussion: There is increasing interest in transfer learning for brain MRI. Well-known public datasets have clearly contributed to the popularity of Alzheimer's diagnostics/prognostics and tumor segmentation as applications. Likewise, the availability of pretrained CNNs has promoted their utilization. Finally, the majority of the surveyed studies did not examine in detail the interpretation of their strategies after applying transfer learning, and did not compare their approach with other transfer learning approaches.

Quantitative dual contrast photon-counting computed tomography for assessment of articular cartilage health.

Photon-counting detector computed tomography (PCD-CT) is a modern spectral imaging technique utilizing photon-counting detectors (PCDs). PCDs detect individual photons and classify them into fixed energy bins, thus enabling energy selective imaging, contrary to energy integrating detectors that detects and sums the total energy from all photons during acquisition. The structure and composition of the articular cartilage cannot be detected with native CT imaging but can be assessed using contrast-enhancement. Spectral imaging allows simultaneous decomposition of multiple contrast agents, which can be used to target and highlight discrete cartilage properties. Here we report, for the first time, the use of PCD-CT to quantify a cationic iodinated CA4+ (targeting proteoglycans) and a non-ionic gadolinium-based gadoteridol (reflecting water content) contrast agents inside human osteochondral tissue (n = 53). We performed PCD-CT scanning at diffusion equilibrium and compared the results against reference data of biomechanical and optical density measurements, and Mankin scoring. PCD-CT enables simultaneous quantification of the two contrast agent concentrations inside cartilage and the results correlate with the structural and functional reference parameters. With improved soft tissue contrast and assessment of proteoglycan and water contents, PCD-CT with the dual contrast agent method is of potential use for the detection and monitoring of osteoarthritis.

Quantitative Longitudinal Predictions of Alzheimer's Disease by Multi-Modal Predictive Learning.

BACKGROUND: Quantitatively predicting the progression of Alzheimer's disease (AD) in an individual on a continuous scale, such as the Alzheimer's Disease Assessment Scale-cognitive (ADAS-cog) scores, is informative for a personalized approach as opposed to qualitatively classifying the individual into a broad disease category. OBJECTIVE: To evaluate the hypothesis that the multi-modal data and predictive learning models can be employed for future predicting ADAS-cog scores. METHODS: Unimodal and multi-modal regression models were trained on baseline data comprised of demographics, neuroimaging, and cerebrospinal fluid based markers, and genetic factors to predict future ADAS-cog scores for 12, 24, and 36 months. We subjected the prediction models to repeated cross-validation and assessed the resulting mean absolute error (MAE) and cross-validated correlation (ρ) of the model. RESULTS: Prediction models trained on multi-modal data outperformed the models trained on single modal data in predicting future ADAS-cog scores (MAE12, 24 & 36 months= 4.1, 4.5, and 5.0, ρ12, 24 & 36 months= 0.88, 0.82, and 0.75). Including baseline ADAS-cog scores to prediction models improved predictive performance (MAE12, 24 & 36 months= 3.5, 3.7, and 4.6, ρ12, 24 & 36 months= 0.89, 0.87, and 0.80). CONCLUSION: Future ADAS-cog scores were predicted which could aid clinicians in identifying those at greater risk of decline and apply interventions at an earlier disease stage and inform likely future disease progression in individuals enrolled in AD clinical trials.

Orientation anisotropy of quantitative MRI parameters in degenerated human articular cartilage.

Quantitative magnetic resonance (MR) relaxation parameters demonstrate varying sensitivity to the orientation of the ordered tissues in the magnetic field. In this study, the orientation dependence of multiple relaxation parameters was assessed in cadaveric human cartilage with varying degree of natural degeneration, and compared with biomechanical testing, histological scoring, and quantitative histology. Twelve patellar cartilage samples were imaged at 9.4 T MRI with multiple relaxation parameters, including T1 , T2 , CW - T1ρ , and adiabatic T1ρ , at three different orientations with respect to the main magnetic field. Anisotropy of the relaxation parameters was quantified, and the results were compared with the reference measurements and between samples of different histological Osteoarthritis Research Society International (OARSI) grades. T2 and CW - T1ρ at 400 Hz spin-lock demonstrated the clearest anisotropy patterns. Radial zone anisotropy for T2 was significantly higher for samples with OARSI grade 2 than for grade 4. The proteoglycan content (measured as optical density) correlated with the radial zone MRI orientation anisotropy for T2 (r = 0.818) and CW - T1ρ with 400 Hz spin-lock (r = 0.650). Orientation anisotropy of MRI parameters altered with progressing cartilage degeneration. This is associated with differences in the integrity of the collagen fiber network, but it also seems to be related to the proteoglycan content of the cartilage. Samples with advanced OA had great variation in all biomechanical and histological properties and exhibited more variation in MRI orientation anisotropy than the less degenerated samples. Understanding the background of relaxation anisotropy on a molecular level would help to develop new MRI contrasts and improve the application of previously established quantitative relaxation contrasts.

Structural Brain Imaging Phenotypes of Mild Cognitive Impairment (MCI) and Alzheimer's Disease (AD) Found by Hierarchical Clustering.

A hierarchical clustering algorithm was applied to magnetic resonance images (MRI) of a cohort of 751 subjects having a mild cognitive impairment (MCI), 282 subjects having received Alzheimer's disease (AD) diagnosis, and 428 normal controls (NC). MRIs were preprocessed to gray matter density maps and registered to a stereotactic space. By first rendering the gray matter density maps comparable by regressing out age, gender, and years of education, and then performing the hierarchical clustering, we found clusters displaying structural features of typical AD, cortically-driven atypical AD, limbic-predominant AD, and early-onset AD (EOAD). Among these clusters, EOAD subjects displayed marked cortical gray matter atrophy and atrophy of the precuneus. Furthermore, EOAD subjects had the highest progression rates as measured with ADAS slopes during the longitudinal follow-up of 36 months. Striking heterogeneities in brain atrophy patterns were observed with MCI subjects. We found clusters of stable MCI, clusters of diffuse brain atrophy with fast progression, and MCI subjects displaying similar atrophy patterns as the typical or atypical AD subjects. Bidirectional differences in structural phenotypes were found with MCI subjects involving the anterior cerebellum and the frontal cortex. The diversity of the MCI subjects suggests that the structural phenotypes of MCI subjects would deserve a more detailed investigation with a significantly larger cohort. Our results demonstrate that the hierarchical agglomerative clustering method is an efficient tool in dividing a cohort of subjects with gray matter atrophy into coherent clusters manifesting different structural phenotypes.

Dual contrast in computed tomography allows earlier characterization of articular cartilage over single contrast.

Cationic computed tomography contrast agents are more sensitive for detecting cartilage degeneration than anionic or non-ionic agents. However, osteoarthritis-related loss of proteoglycans and increase in water content contrarily affect the diffusion of cationic contrast agents, limiting their sensitivity. The quantitative dual-energy computed tomography technique allows the simultaneous determination of the partitions of iodine-based cationic (CA4+) and gadolinium-based non-ionic (gadoteridol) agents in cartilage at diffusion equilibrium. Normalizing the cationic agent partition at diffusion equilibrium with that of the non-ionic agent improves diagnostic sensitivity. We hypothesize that this sensitivity improvement is also prominent during early diffusion time points and that the technique is applicable during contrast agent diffusion. To investigate the validity of this hypothesis, osteochondral plugs (d = 8 mm, N = 33), extracted from human cadaver (n = 4) knee joints, were immersed in a contrast agent bath (a mixture of CA4+ and gadoteridol) and imaged using the technique at multiple time points until diffusion equilibrium. Biomechanical testing and histological analysis were conducted for reference. Quantitative dual-energy computed tomography technique enabled earlier determination of cartilage proteoglycan content over single contrast. The correlation coefficient between human articular cartilage proteoglycan content and CA4+ partition increased with the contrast agent diffusion time. Gadoteridol normalized CA4+ partition correlated significantly (P < .05) with Mankin score at all time points and with proteoglycan content after 4 hours. The technique is applicable during diffusion, and normalization with gadoteridol partition improves the sensitivity of the CA4+ contrast agent.

Open-source python module for automated preprocessing of near infrared spectroscopic data.

Near infrared spectroscopy (NIRS) is an analytical technique for determining the chemical composition or structure of a given sample. For several decades, NIRS has been a frequently used analysis tool in agriculture, pharmacology, medicine, and petrochemistry. The popularity of NIRS is constantly growing as new application areas are discovered. Contrary to mid infrared spectral region, the absorption bands in near infrared spectral region are often non-specific, broad, and overlapping. Analysis of NIR spectra requires multivariate methods which are highly subjective to noise arising from instrumentation, scattering effects, and measurement setup. NIRS measurements are also frequently performed outside of a laboratory which further contributes to the presence of noise. Therefore, preprocessing is a critical step in NIRS as it can vastly improve the performance of multivariate models. While extensive research regarding various preprocessing methods exists, selection of the best preprocessing method is often determined through trial-and-error. A more powerful approach for optimizing preprocessing in NIRS models would be to automatically compare a large number of preprocessing techniques (e.g., through grid-search or hyperparameter tuning). To enable this, we present, nippy, an open-source Python module for semi-automatic comparison of NIRS preprocessing methods (available at https://github.com/uef-bbc/nippy). We provide here a brief overview of the capabilities of nippy and demonstrate the typical usage through two examples with public datasets.

Synchrotron MicroCT Reveals the Potential of the Dual Contrast Technique for Quantitative Assessment of Human Articular Cartilage Composition.

Dual contrast micro computed tomography (CT) shows potential for detecting articular cartilage degeneration. However, the performance of conventional CT systems is limited by beam hardening, low image resolution (full-body CT), and long acquisition times (conventional microCT). Therefore, to reveal the full potential of the dual contrast technique for imaging cartilage composition we employ the technique using synchrotron microCT. We hypothesize that the above-mentioned limitations are overcome with synchrotron microCT utilizing monochromatic X-ray beam and fast image acquisition. Human osteochondral samples (n = 41, four cadavers) were immersed in a contrast agent solution containing two agents (cationic CA4+ and non-ionic gadoteridol) and imaged with synchrotron microCT at an early diffusion time point (2 h) and at diffusion equilibrium (72 h) using two monochromatic X-ray energies (32 and 34 keV). The dual contrast technique enabled simultaneous determination of CA4+ (i.e., proteoglycan content) and gadoteridol (i.e., water content) partitions within cartilage. Cartilage proteoglycan content and biomechanical properties correlated significantly (0.327 < r < 0.736, p < 0.05) with CA4+ partition in superficial and middle zones at both diffusion time points. Normalization of the CA4+ partition with gadoteridol partition within the cartilage significantly (p < 0.05) improved the detection sensitivity for human osteoarthritic cartilage proteoglycan content, biomechanical properties, and overall condition (Mankin, Osteoarthritis Research Society International, and International Cartilage Repair Society grading systems). The dual energy technique combined with the dual contrast agent enables assessment of human articular cartilage proteoglycan content and biomechanical properties based on CA4+ partition determined using synchrotron microCT. Additionally, the dual contrast technique is not limited by the beam hardening artifact of conventional CT systems. © 2019 The Authors. Journal of Orthopaedic Research® published by Wiley Periodicals, Inc. on behalf of Orthopaedic Research Society. J Orthop Res 38:563-573, 2020.

Dataset on equine cartilage near infrared spectra, composition, and functional properties.

Near infrared (NIR) spectroscopy is a well-established technique that is widely employed in agriculture, chemometrics, and pharmaceutical engineering. Recently, the technique has shown potential in clinical orthopaedic applications, for example, assisting in the diagnosis of various knee-related diseases (e.g., osteoarthritis) and their pathologies. NIR spectroscopy (NIRS) could be especially useful for determining the integrity and condition of articular cartilage, as the current arthroscopic diagnostics is subjective and unreliable. In this work, we present an extensive dataset of NIRS measurements for evaluating the condition, mechanical properties, structure, and composition of equine articular cartilage. The dataset contains NIRS measurements from 869 different locations across the articular surfaces of five equine fetlock joints. A comprehensive library of reference values for each measurement location is also provided, including results from a mechanical indentation testing, digital densitometry imaging, polarized light microscopy, and Fourier transform infrared spectroscopy. The published data can either be used as a model of human cartilage or to advance equine veterinary research.

Quantitative Dual Contrast CT Technique for Evaluation of Articular Cartilage Properties.

Impact injuries of cartilage may initiate post-traumatic degeneration, making early detection of injury imperative for timely surgical or pharmaceutical interventions. Cationic (positively-charged) CT contrast agents detect loss of cartilage proteoglycans (PGs) more sensitively than anionic (negatively-charged) or non-ionic (non-charged, i.e., electrically neutral) agents. However, degeneration related loss of PGs and increase in water content have opposite effects on the diffusion of the cationic agent, lowering its sensitivity. In contrast to cationic agents, diffusion of non-ionic agents is governed only by steric hindrance and water content of cartilage. We hypothesize that sensitivity of an iodine(I)-based cationic agent may be enhanced by simultaneous use of a non-ionic gadolinium(Gd)-based agent. We introduce a quantitative dual energy CT technique (QDECT) for simultaneous quantification of two contrast agents in cartilage. We employ this technique to improve the sensitivity of cationic CA4+ (q =+4) by normalizing its partition in cartilage with that of non-ionic gadoteridol. The technique was evaluated with measurements of contrast agent mixtures of known composition and human osteochondral samples (n = 57) after immersion (72 h) in mixture of CA4+ and gadoteridol. Samples were arthroscopically graded and biomechanically tested prior to QDECT (50/100 kV). QDECT determined contrast agent mixture compositions correlated with the true compositions (R2= 0.99, average error = 2.27%). Normalizing CA4+ partition in cartilage with that of gadoteridol improved correlation with equilibrium modulus (from ρ = 0.701 to 0.795). To conclude, QDECT enables simultaneous quantification of I and Gd contrast agents improving diagnosis of cartilage integrity and biomechanical status.

Optimal Regression Method for Near-Infrared Spectroscopic Evaluation of Articular Cartilage.

Near-infrared (NIR) spectroscopy has been successful in nondestructive assessment of biological tissue properties, such as stiffness of articular cartilage, and is proposed to be used in clinical arthroscopies. Near-infrared spectroscopic data include absorbance values from a broad wavelength region resulting in a large number of contributing factors. This broad spectrum includes information from potentially noisy variables, which may contribute to errors during regression analysis. We hypothesized that partial least squares regression (PLSR) is an optimal multivariate regression technique and requires application of variable selection methods to further improve the performance of NIR spectroscopy-based prediction of cartilage tissue properties, including instantaneous, equilibrium, and dynamic moduli and cartilage thickness. To test this hypothesis, we conducted for the first time a comparative analysis of multivariate regression techniques, which included principal component regression (PCR), PLSR, ridge regression, least absolute shrinkage and selection operator (Lasso), and least squares version of support vector machines (LS-SVM) on NIR spectral data of equine articular cartilage. Additionally, we evaluated the effect of variable selection methods, including Monte Carlo uninformative variable elimination (MC-UVE), competitive adaptive reweighted sampling (CARS), variable combination population analysis (VCPA), backward interval PLS (BiPLS), genetic algorithm (GA), and jackknife, on the performance of the optimal regression technique. The PLSR technique was found as an optimal regression tool (R2Tissue thickness = 75.6%, R2Dynamic modulus = 64.9%) for cartilage NIR data; variable selection methods simplified the prediction models enabling the use of lesser number of regression components. However, the improvements in model performance with variable selection methods were found to be statistically insignificant. Thus, the PLSR technique is recommended as the regression tool for multivariate analysis for prediction of articular cartilage properties from its NIR spectra.