Machine learning classification on texture analyzed T2 maps of osteoarthritic cartilage: oulu knee osteoarthritis study

Which Endpoints Should be Applied in Interventional Trials? -: From Single Uni-dimensional Assessment Tailored to a Drug's Mechanism of Action to Multi-Component Measures and Multi-Domain Composites

Objective

A vast array of structural/imaging and clinical endpoints/outcomes are available today to osteoarthritis epidemiologists or trialists. Which assessments are best suited for which studies remains unsettled. When several assessments are available, these may be analyzed together (simultaneously or hierarchically), using statistical modeling and adjustment. Or, alternatively, they may be combined to form more complex multi-component or composite (potentially multi-domain) endpoints/outcomes. This review describes such concepts and their challenges, using examples from current osteoarthritis imaging research.

Design

A narrative, non-systematic literature search (PubMed and others) was conducted, and informal consultations were held with experts in the field. The identified concepts and experimental findings were then organized to present an integrated framework.

Results

Single imaging assessments can encompass one (uni-dimensional) or more (multi-dimensional) structures. Integration of image assessments of one structure/tissue across anatomical locations provides aggregate measures. This can also be created across heterogeneous (multi-dimensional) types of assessments (multi-component/composite), either within an area (such as imaging - single domain) or across broader areas of health and well-being (multi-domain). Weighting, standardization, and (clinical) usefulness are crucial characteristics of multi-component/composite endpoints. Examples of these concepts are here provided in the context of osteoarthritis imaging.

Conclusions

Options for multi-component/composite endpoints in osteoarthritis research are virtually infinite. Smart research strategies are required to explore and validate these vast possibilities, with appropriate statistical treatment being paramount. A one-size/endpoint-fits-all approach will likely fail in observational and interventional studies. Imaging assessment needs to be tailored to both the drug's unique mechanism of action, and to the participants' morpho-type.

Texture Analysis of Cartilage Repair Tissue Maturation: Comparison of Two Cartilage Repair Methods and Correlation with MOCART 2.0

OBJECTIVE

The objective of this study was to assess the maturation of matrix-associated autologous chondrocyte transplantation (MACT) grafts up to 2 years after the surgery using gray-level co-occurrence matrix (GLCM) texture analysis of quantitative T2 maps, compare the results with the microfracturing technique (MFX) control group, and relate these results to the morphological MOCART 2.0 score.

DESIGN

A subcohort of 37 patients from prospective, multi-center study underwent examination on a 3T MR scanner, including a T2 mapping sequence at 3, 12, and 24 months after surgery. Changes between the time-points in the mean T2 values and 20 GLCM features extracted from T2 maps were assessed in repair tissue, tissue adjacent to the repair site, and the reference cartilage for both procedures.

RESULTS

Significant correlations were found between the MOCART 2.0 and GLCM features for both surgical procedures. There were no significant differences between MACT and MFX. We identified significant intra-group changes in T2 and autocorrelation (3M-12M: P = 0.002; 3M-24M: P = 0.004), dissimilarity (3M-24M: P = 0.01), homogeneity (3M-24M: P = 0.013), and correlation (3M-24M: P = 0.036), sum average (3M-12M: P = 0.001; 3M-24M: P = 0.002), and information measure (3M-24M: P < 0.001) in the MACT repair tissue. MACT models revealed differences in GLCM between all combinations of ROI types at almost all time-points. In the case of MFX, the significant differences were mainly between repair and reference tissue at 12 months.

CONCLUSION

Texture analysis provides a useful extension to T2 mapping. Texture features are correlated to the morphological outcome and reveal differences in the process of maturation between MACT and MFX.

Quantitative MRI methods for the assessment of structure, composition, and function of musculoskeletal tissues in basic research and preclinical applications

Osteoarthritis (OA) is a disabling chronic disease involving the gradual degradation of joint structures causing pain and dysfunction. Magnetic resonance imaging (MRI) has been widely used as a non-invasive tool for assessing OA-related changes. While anatomical MRI is limited to the morphological assessment of the joint structures, quantitative MRI (qMRI) allows for the measurement of biophysical properties of the tissues at the molecular level. Quantitative MRI techniques have been employed to characterize tissues' structural integrity, biochemical content, and mechanical properties. Their applications extend to studying degenerative alterations, early OA detection, and evaluating therapeutic intervention. This article is a review of qMRI techniques for musculoskeletal tissue evaluation, with a particular emphasis on articular cartilage. The goal is to describe the underlying mechanism and primary limitations of the qMRI parameters, their association with the tissue physiological properties and their potential in detecting tissue degeneration leading to the development of OA with a primary focus on basic and preclinical research studies. Additionally, the review highlights some clinical applications of qMRI, discussing the role of texture-based radiomics and machine learning in advancing OA research.

Cartilage T2 mapping-based radiomics in knee osteoarthritis research: Status, progress and future outlook

Osteoarthritis (OA) affects more than 500 millions people worldwide and places an enormous economic and medical burden on patients and healthcare systems. The knee is the most commonly affected joint. However, there is no effective early diagnostic method for OA. The main pathological feature of OA is cartilage degeneration. Owing to the poor regenerative ability of chondrocytes, early detection of OA and prompt intervention are extremely important. The T2 relaxation time indicates changes in cartilage composition and responds to alterations in the early cartilage matrix. T2 mapping does not require contrast agents or special equipment, so it is widely used. Radiomics analysis methods are used to construct diagnostic or predictive models based on information extracted from clinical images. Owing to the development of artificial intelligence methods, radiomics has made excellent progress in segmentation and model construction. In this review, we summarize the progress of T2 mapping radiomics research methods in terms of T2 map acquisition, image postprocessing, and OA diagnosis or predictive model construction.

Survival analysis on subchondral bone length for total knee replacement

OBJECTIVE

Use subchondral bone length (SBL), a new MRI-derived measure that reflects the extent of cartilage loss and bone flattening, to predict the risk of progression to total knee replacement (TKR).

METHODS

We employed baseline MRI data from the Osteoarthritis Initiative (OAI), focusing on 760 men and 1214 women with bone marrow lesions (BMLs) and joint space narrowing (JSN) scores, to predict the progression to TKR. To minimize bias from analyzing both knees of a participant, only the knee with a higher Kellgren-Lawrence (KL) grade was considered, given its greater potential need for TKR. We utilized the Kaplan-Meier survival curves and Cox proportional hazards models, incorporating raw and normalized values of SBL, JSN, and BML as predictors. The study included subgroup analyses for different demographics and clinical characteristics, using models for raw and normalized SBL (merged, femoral, tibial), BML (merged, femoral, tibial), and JSN (medial and lateral compartments). Model performance was evaluated using the time-dependent area under the curve (AUC), Brier score, and Concordance index to gauge accuracy, calibration, and discriminatory power. Knee joint and region-level analyses were conducted to determine the effectiveness of SBL, JSN, and BML in predicting TKR risk.

RESULTS

The SBL model, incorporating data from both the femur and tibia, demonstrated a predictive capacity for TKR that closely matched the performance of the BML score and the JSN grade. The Concordance index of the SBL model was 0.764, closely mirroring the BML's 0.759 and slightly below JSN's 0.788. The Brier score for the SBL model stood at 0.069, showing comparability with BML's 0.073 and a minor difference from JSN's 0.067. Regarding the AUC, the SBL model achieved 0.803, nearly identical to BML's 0.802 and slightly lower than JSN's 0.827.

CONCLUSION

SBL's capacity to predict the risk of progression to TKR highlights its potential as an effective imaging biomarker for knee osteoarthritis.

Three-dimensional texture analyses of multi-quantitative relaxation time maps for evaluating cartilage repair with the treatment of allogeneic human adipose-derived mesenchymal progenitor cells

BACKGROUND

To explore the ability of three-dimensional texture analyses based on gray-level run-length matrix (GLRLM) for examining the spatial distribution of pixel values on magnetic resonance imaging (MRI) relaxation time maps and detecting the compositional variation of cartilage repair following treatment with allogeneic human adipose-derived mesenchymal progenitor cells (haMPCs).

METHODS

Participants with knee osteoarthritis were randomly divided into three groups with intra-articular haMPCs injections: low-, medium-, and high-dose groups. We analyzed five GLRLM parameters in the T1rho, T2 and T2star maps, including run length non-uniformity (RLNonUni), gray-level non-uniformity (GLevNonU), long run emphasis (LngREmph), short run emphasis (ShrtREmp), and fraction of images in runs. We used the relative D values (the ratio of difference values to baseline) as the objective to avoid errors caused by individual differences. We calculated the two-tailed Pearson's linear correlation coefficient (r) to investigate the correlations of the texture parameters with the Western Ontario and McMaster Universities Osteoarthritis Index (WOMAC) scores.

RESULTS

Compared with the base time, significant reduction of WOMAC score was observed in both high and medium doses groups at terminal time, indicating relief of pain symptoms in high and medium groups with the treatment of allogeneic haMPCs. Significant differences were observed in the GLRLM parameters of cartilage MR relaxation time maps in different doses groups. In both T1rho and T2 relaxation time maps, the high-dose group showed significant increases in relative D values of RLNonUni, GLevNonU, LngREmph and ShrtREmp, which indicated significant changes in the uniformity of relaxation time maps. For T2star map, GLRLM parameters such as GLevNonU and ShrtREmp, especially LngREmph, showed significant increases in relative D values in high-dose group. Among all GLRLM features, LngREmph of three relaxation time maps had performed excellent linear correlations with WOMAC scores.

CONCLUSIONS

Texture analysis of the cartilage may allow the detection of compositional variation in cartilage repair with the treatment of allogeneic haMPCs. This technique displays potential applications in understanding the mechanism of stem cell repair of the cartilage and assessing the treatment response.

Radiomics signature of osteoarthritis: Current status and perspective

Osteoarthritis (OA) is one of the fast-growing disability-related diseases worldwide, which has significantly affected the quality of patients' lives and brings about substantial socioeconomic burdens in medical expenditure. There is currently no cure for OA once the bone damage is established. Unfortunately, the existing radiological examination is limited to grading the disease's severity and is insufficient to precisely diagnose OA, detect early OA or predict OA progression. Therefore, there is a pressing need to develop novel approaches in medical image analysis to detect subtle changes for identifying early OA development and rapid progressors. Recently, radiomics has emerged as a unique approach to extracting high-dimensional imaging features that quantitatively characterise visible or hidden information from routine medical images. Radiomics data mining via machine learning has empowered precise diagnoses and prognoses of disease, mainly in oncology. Mounting evidence has shown its great potential in aiding the diagnosis and contributing to the study of musculoskeletal diseases. This paper will summarise the current development of radiomics at the crossroads between engineering and medicine and discuss the application and perspectives of radiomics analysis for OA diagnosis and prognosis.

The translational potential of this article

Radiomics is a novel approach used in oncology, and it may also play an essential role in the diagnosis and prognosis of OA. By transforming medical images from qualitative interpretation to quantitative data, radiomics could be the solution for precise early OA detection, progression tracking, and treatment efficacy prediction. Since the application of radiomics in OA is still in the early stages and primarily focuses on fundamental studies, this review may inspire more explorations and bring more promising diagnoses, prognoses, and management results of OA.

Artificial intelligence in osteoarthritis detection: A systematic review and meta-analysis

OBJECTIVES

As an increasing number of studies apply artificial intelligence (AI) algorithms in osteoarthritis (OA) detection, we performed a systematic review and meta-analysis to pool the data on diagnostic performance metrics of AI, and to compare them with clinicians' performance.

MATERIALS AND METHODS

A search in PubMed and Scopus was performed to find studies published up to April 2022 that evaluated and/or validated an AI algorithm for the detection or classification of OA. We performed a meta-analysis to pool the data on the metrics of diagnostic performance. Subgroup analysis based on the involved joint and meta-regression based on multiple parameters were performed to find potential sources of heterogeneity. The risk of bias was assessed using Prediction Model Study Risk of Bias Assessment Tool reporting guidelines.

RESULTS

Of the 61 studies included, 27 studies with 91 contingency tables provided sufficient data to enter the meta-analysis. The pooled sensitivities for AI algorithms and clinicians on internal validation test sets were 88% (95% confidence interval [CI]: 86,91) and 80% (95% CI: 68,88) and pooled specificities were 81% (95% CI: 75,85) and 79% (95% CI: 80,85), respectively. At external validation, the pooled sensitivity and specificity for AI algorithms were 94% (95% CI: 90,97) and 91% (95% CI: 77,97), respectively.

CONCLUSION

Although the results of this meta-analysis should be interpreted with caution due to the potential pitfalls in the included studies, the promising role of AI as a diagnostic adjunct to radiologists is indisputable.

Osteoarthritis year in review 2022: imaging

PURPOSE

This narrative review summarizes original research focusing on imaging in osteoarthritis (OA) published between April 1st 2021 and March 31st 2022. We only considered English publications that were in vivo human studies.

METHODS

The PubMed, Medline, Embase, Scopus, and ISI Web of Science databases were searched for "Osteoarthritis/OA" studies based on the search terms: "Radiography", "Ultrasound/US", "Computed Tomography/CT", "DXA", "Magnetic Resonance Imaging/MRI", "Artificial Intelligence/AI", and "Deep Learning". This review highlights the anatomical focus of research on the structures within the tibiofemoral, patellofemoral, hip, and hand joints. There is also a noted focus on artificial intelligence applications in OA imaging.

RESULTS

Over the last decade, the increasing trend of using open-access large databases has reached a plateau (from 17 to 37). Compositional MRI has had the most prominent use in OA imaging and its biomarkers have been used in the detection of preclinical OA and prediction of OA outcomes. Most noteworthy, there has been an accelerated rate of publications on the implications of artificial intelligence, used in developing prediction models and performing trabecular texture analysis, in OA imaging (from 17 to 154).

CONCLUSIONS

While imaging has maintained its key role in OA research, publication trends have shown an emphasis on the integration of AI. During the past year, MRI has maintained the highest prevalence in usage while US and CT remain as readily available modalities. Finally, there has been a notable uptake in the development and validation of AI techniques used to perform texture analysis and predict OA progression.

Osteoarthritis year in review 2021: imaging

PURPOSE

To provide a narrative review of original articles on imaging of osteoarthritis (OA) published between January 1, 2020 and March 31, 2021, with a special focus on imaging of inflammation, imaging of bone, cartilage and bone-cartilage interactions, imaging of peri-articular tissues, imaging scoring methods for OA, and artificial intelligence (AI) applied to OA imaging.

METHODS

The Embase, Pubmed, Medline, Cochrane databases were searched for original research articles in the English language on human, in vivo, imaging of OA published between January 1, 2020 and March 31, 2021. Search terms related to osteoarthritis combined with all imaging modalities and artificial intelligence were applied. A selection of articles reporting on one of the focus topics was discussed further.

RESULTS

The search resulted in 651 articles, of which 214 were deemed relevant to human OA imaging. Among the articles included, the knee joint (69%) and magnetic resonance imaging (MRI) (52%) were the predominant anatomical area and imaging modality studied. There were also a substantial number of papers (n = 46) reporting on AI applications in the field of OA imaging.

CONCLUSION

Imaging continues to play an important role in the assessment of OA. Recent advances in OA imaging include quantitative, non-contrast, and hybrid imaging techniques for improved characterization of multiple tissue processes in OA. In addition, an increasing effort in AI techniques is undertaken to enhance OA imaging acquisition and analysis.

Combining Spectral and Texture Features of UAV Images for the Remote Estimation of Rice LAI throughout the Entire Growing Season

Leaf area index (LAI) estimation is very important, and not only for canopy structure analysis and yield prediction. The unmanned aerial vehicle (UAV) serves as a promising solution for LAI estimation due to its great applicability and flexibility. At present, vegetation index (VI) is still the most widely used method in LAI estimation because of its fast speed and simple calculation. However, VI only reflects the spectral information and ignores the texture information of images, so it is difficult to adapt to the unique and complex morphological changes of rice in different growth stages. In this study we put forward a novel method by combining the texture information derived from the local binary pattern and variance features (LBP and VAR) with the spectral information based on VI to improve the estimation accuracy of rice LAI throughout the entire growing season. The multitemporal images of two study areas located in Hainan and Hubei were acquired by a 12-band camera, and the main typical bands for constituting VIs such as green, red, red edge, and near-infrared were selected to analyze their changes in spectrum and texture during the entire growing season. After the mathematical combination of plot-level spectrum and texture values, new indices were constructed to estimate rice LAI. Comparing the corresponding VI, the new indices were all less sensitive to the appearance of panicles and slightly weakened the saturation issue. The coefficient of determination (R2) can be improved for all tested VIs throughout the entire growing season. The results showed that the combination of spectral and texture features exhibited a better predictive ability than VI for estimating rice LAI. This method only utilized the texture and spectral information of the UAV image itself, which is fast, easy to operate, does not need manual intervention, and can be a low-cost method for monitoring crop growth.