A comparative analysis of automatic classification and grading methods for knee osteoarthritis focussing on X-ray images

Radiomics-based nomogram for predicting total knee replacement in knee osteoarthritis patients

BACKGROUND

The incidence of total knee replacement (TKR) surgeries has increased, partly attributed to healthcare policies that cause premature and potentially unwarranted interventions. This has raised concerns regarding a potential trend of excessive surgeries.

PURPOSE

This study aimed to propose a predictive model based on digital radiography (DR) radiomics to objectively assess the need for TKR surgery in patients with knee osteoarthritis (KOA) and to improve risk stratification, thereby avoiding unnecessary surgeries.

METHODS

A retrospective study was conducted on 1,785 KOA patients from January 2017 to December 2022. Radiomics features were extracted from DR images to quantify lesion phenotypes, followed by a two-step feature selection to derive robust signatures. Multiple models were constructed using independent risk factors and radiomics features, and these models were validated using logistic regression. The performance of the models was evaluated via receiver operating characteristic (ROC) curves, area under the curve (AUC), calibration curves, and decision curve analysis. A multivariable Cox regression-derived nomogram was used to predict operation-free survival (OFS), and the patients were categorized into high- or low-risk groups based on risk stratification. Kaplan-Meier curves were used to compare OFS between the two groups.

RESULTS

During a follow-up period of at least one year, 962 of 1785 (53.89 %) patients underwent TKR. Age, presence of radiographic KOA (RKOA), and Kellgren-Lawrence (KL) grading were identified as independent risk factors for OFS. The combined RKOA model (including age, presence of RKOA, and Radscore; AUC = 0.969) and combined KL model (including age, KL grading, and Radscore; AUC = 0.968) showed similar performance, with both significantly outperforming other models (p < 0.001). The 1-, 2-, and 3-year AUCs for the RKOA nomogram were 0.891, 0.916, and 0.920, respectively, whereas those for the KL nomogram were 0.890, 0.914, and 0.931. The thresholds of 68.92 (RKOA nomogram) and 64.41 (KL nomogram) were derived from the median nomogram scores and used to stratify patients into high- and low-risk groups. K-M curves demonstrated that the risk stratification system effectively distinguished between high- and low-risk groups, with the high-risk group being more likely to require TKR.

CONCLUSIONS

Two nomograms incorporating age, RKOA (or KL grading), and Radscore were developed to predict 3-years OFS for KOA patients and establish risk thresholds, potentially guiding personalized non-surgical treatments during the OFS period.

Optimizing knee osteoarthritis severity prediction on MRI images using deep stacking ensemble technique

Knee osteoarthritis (KOA) represents a well-documented degenerative arthropathy prevalent among the elderly population. KOA is a persistent condition, also referred to as progressive joint Disease, stemming from the continual deterioration of cartilage. Predominantly afflicting individuals aged 45 and above, this ailment is commonly labeled as a "wear and tear" joint disorder, targeting joints such as the knee, hand, hips, and spine. Osteoarthritis symptoms typically increase gradually, contributing to the deterioration of articular cartilage. Prominent indicators encompass pain, stiffness, tenderness, swelling, and the development of bone spurs. Diagnosis typically involves the utilization of Radiographic X-ray images, Magnetic Resonance Imaging (MRI), and Computed Tomography (CT) Scan by medical professionals and experts. However, this conventional approach is time-consuming, and also sometimes tedious for medical professionals. In order to address the limitation of time and expedite the diagnostic process, deep learning algorithms have been implemented in the medical field. In the present investigation, four pre-trained models, specifically CNN, AlexNet, ResNet34 and ResNet-50, were utilized to predict the severity of KOA. Further, a Deep stack ensemble technique was employed to achieve optimal performance resulting to the accuracy of 99.71%.

Deep learning to combat knee osteoarthritis and severity assessment by using CNN-based classification

BACKGROUND

In today's digital age, various diseases drastically reduce people's quality of life. Arthritis is one amongst the most common and debilitating maladies. Osteoarthritis affects several joints, including the hands, knees, spine, and hips. This study focuses on the medical disorder underlying Knee Osteoarthritis (KOA) which severely impairs people's quality of life. KOA is characterised by restricted mobility, stiffness, and terrible pain and can be caused by a range of factors such as ageing, obesity, and traumas. This degenerative disorder leads to progressive wear and tear of the knee joint.

METHODS

To combat arthritis in the kneecap, this study employs a 12-layer Convolutional Neural Network (CNN) to reach deep learning capabilities. A collection of data from the Osteoarthritis Initiative (OAI) is used to classify KOA. Through the use of medical image processing; the study ascertains whether an individual has this ailment. A sophisticated CNN architecture created especially for binary classification and KOA severity utilising deep learning algorithms is the main component of this work.

RESULTS

The cross-entropy loss function is an important component of the model's laborious design that classifies data into two groups. The remaining section uses the Kellgren-Lawrence (KL) grade to classify the disease's severity. In the binary classification, the proposed algorithm outperforms previous methods with an accuracy rate of 92.3%, and in the multiclassification, its accuracy rate is 78.4% which is superior to the previous findings.

CONCLUSION

Looking ahead, the research broadens the scope of this work by gathering information from various sources and using these methods on a wider range of datasets and situations. The potential for major advancements in the field of osteoarthritis detection and classification is highlighted by this forward-looking approach. Furthermore, this method reduces the intervention of medical practitioners and ultimately results in accurate diagnosis.

CLINICAL TRIAL NUMBER

Not applicable.

The effects of pulsed electromagnetic field therapy on muscle strength and pain in patients with end-stage knee osteoarthritis: a randomized controlled trial

Background

Osteoarthritis (OA) of the knee is one of the most common chronic degenerative joint conditions affecting aging population.

Aim

To investigate the effectiveness of a combination of home-based exercise and pulsed electromagnetic field (PEMF) therapy to improve muscle strength, physical function, and pain.

Methods

Sixty patients were randomly assigned to either home-based exercise alone (control group; n = 30) or combined with PEMF therapy (treatment group; n = 30) twice a week for eight weeks. Knee extension, flexion muscle strength, gait speed (GS), 5 time sit-to-stand test (5STS), Visual Analogue Scale (VAS) pain and Knee Injury and Osteoarthritis Outcome Score (KOOS) were recorded at baseline and 4 and 8 weeks.

Results

Significant improvements in symptomatic knee extension muscle strength (SKE, p = 0.001), flexion strength (SKF, p = 0.011), contralateral knee extension muscle strength (CKE, p = 0.002), and flexion strength (CKF, p = 0.009) were observed for the PEMF treatment group at 8 weeks. Significant reductions in VAS pain scores were observed in both the treatment (p < 0.001, partial η2 = 0.505) and control (p < 0.001, partial η2 = 0.268) groups. Significant differences were reported between groups in the 4 (p = 0.010, partial η2 = 0.111) and 8 (p = 0.046, partial η2 = 0.068) week assessment in VAS pain. A significant time difference was found in GS and 5STS between baseline and week 8 (GS: difference 0.051, p = 0.026; 5STS: difference 2.327, p < 0.001) in the treatment group. The significant group difference at week 8 was observed in SKE (p = 0.013) in female patients while pain in male patients (p = 0.026). Patients aged over 70 years have a significantly superior improvement in SKE, SKF, and CKF after 8 weeks of PEMF therapy.

Conclusion

The combination of PEMF therapy and home-based exercise superiorly improved knee muscle strength and reduced pain in end-stage knee OA subjects and showed a promising tendency to improve performance-based physical function. PEMF therapy was shown to preferentially benefit knee muscle strength in female patients and patients aged over 70 years, whereas male patients were more responsive to PEMF therapy in the form of pain relief.

Clinical trial registration

clinicalTrials.gov, NCT05550428.

Automatic grading of knee osteoarthritis with a plain radiograph radiomics model: combining anteroposterior and lateral images

OBJECTIVES

To establish a radiomics-based automatic grading model for knee osteoarthritis (OA) and evaluate the influence of different body positions on the model's effectiveness.

MATERIALS AND METHODS

Plain radiographs of a total of 473 pairs of knee joints from 473 patients (May 2020 to July 2021) were retrospectively analyzed. Each knee joint included anteroposterior (AP) and lateral (LAT) images which were randomly assigned to the training cohort and the testing cohort at a ratio of 7:3. First, an assessment of knee OA severity was done by two independent radiologists with Kallgren-Lawrence grading scale. Then, another two radiologists independently delineated the region of interest for radiomic feature extraction and selection. The radiomic classification features were dimensionally reduced and a machine model was conducted using logistic regression (LR). Finally, the classification efficiency of the model was evaluated using receiver operating characteristic curves and the area under the curve (AUC).

RESULTS

The AUC (macro/micro) of the model using a combination of AP and LAT (AP&LAT) images were 0.772/0.778, 0.818/0.799, and 0.864/0.879, respectively. The radiomic features from the combined images achieved better classification performance than the individual position image (p < 0.05). The overall accuracy of the radiomic model with AP&LAT images was 0.727 compared to 0.712 and 0.417 for radiologists with 4 years and 2 years of musculoskeletal diagnostic experience.

CONCLUSIONS

A radiomic model constructed by combining the AP&LAT images of the knee joint can better grade knee OA and assist clinicians in accurate diagnosis and treatment.

CRITICAL RELEVANCE STATEMENT

A radiomic model based on plain radiographs accurately grades knee OA severity. By utilizing the LR classifier and combining AP&LAT images, it improves accuracy and consistency in grading, aiding clinical decision-making, and treatment planning.

KEY POINTS

Radiomic model performed more accurately in K/L grading of knee OA than junior radiologists. Radiomic features from the combined images achieved better classification performance than the individual position image. A radiomic model can improve the grading of knee OA and assist in diagnosis and treatment.

Machine learning-based bioimpedance assessment of knee osteoarthritis severity

This study proposes a multiclass model to classify the severity of knee osteoarthritis (KOA) using bioimpedance measurements. The experimental setup considered three types of measurements using eight electrodes: global impedance with adjacent pattern, global impedance with opposite pattern, and direct impedance measurement, which were taken using an electronic device proposed by authors and based on the Analog Devices AD5933 impedance converter. The study comprised 37 participants, 25 with healthy knees and 13 with three different degrees of KOA. All participants performed 20 repetitions of each of the following five tasks: (i) sitting with the knee bent, (ii) sitting with the knee extended, (iii) sitting and performing successive extensions and flexions of the knee, (iv) standing, and (v) walking. Data from the 15 experimental setups (3 types of measurements×5 exercises) were used to train a multiclass random forest. The training and validation cycle was repeated 100 times using random undersampling. At each of the 100 cycles, 80% of the data were used for training and the rest for testing. The results showed that the proposed approach achieved average sensitivities and specificities of 100% for the four KOA severity grades in the extension, cyclic, and gait tasks. This suggests that the proposed method can serve as a screening tool to determine which individuals should undergo x-rays or magnetic resonance imaging for further evaluation of KOA.

A proposed modification to the Kellgren and Lawrence classification for knee osteoarthritis using a compartment-specific approach

Purpose

Since Kellgren and Lawrence (KL) originally classified knee osteoarthritis, several authors have reported varying levels of reliability and a lack of uniformity in the use of this classification system. We propose several modifications to the KL classification including the use of a compartment-specific approach that we hypothesize will lead to a better understanding of knee OA while maintaining an adequate interobserver and intraobserver reliability.

Methods

We propose the addition of the lateral and skyline-view radiographs to the standard anteroposterior (AP) and lateral projections in the evaluation. Also suggest a more precise definition of the evaluated parameters; the addition of the subchondral cancellous bone as parameter of evaluation; and the assessment of medial tibiofemoral compartment (MTFC), lateral tibiofemoral compartment (LTFC) and patellofemoral compartment (PFC) separately resulting in a compartment-specific KL staging score rather than a single overall KL score. Six evaluators (two knee surgeons, two radiologists and two knee fellows) used the modified KL classification to classify 230 randomly selected knees on two separate occasions. Reliabilities were assessed by calculating Krippendorff's ⍺ coefficients.

Results

Two hundred and ten knees were included for final evaluation and analyses (53% left knees; 65% females; mean age 56 years old). Average interobserver reliability was moderate for all compartments (0.51 for the MTFC; 0.51 for the LTFC; and 0.56 for the PFC). Average intraobserver reliability was substantial for all compartments (0.63 for the MTFC; 0.65 for the LTFC; and 0.7 for the PFC). Experienced evaluators showed a higher intraobserver reliability than less-experienced evaluators.

Conclusions

A modified compartment-specific KL classification enables a practical and detailed description of knee OA involvement and demonstrates acceptable interobserver and intraobserver reliability. Level of Evidence: Level III.

DC-AAE: Dual channel adversarial autoencoder with multitask learning for KL-grade classification in knee radiographs

Knee osteoarthritis (OA) is a frequent musculoskeletal disorder that leads to physical disability in older adults. Manual OA assessment is performed via visual inspection, which is highly subjective as it suffers from moderate to high inter-observer variability. Many deep learning-based techniques have been proposed to address this issue. However, owing to the limited amount of labelled data, all existing solutions have limitations in terms of performance or the number of classes. This paper proposes a novel fully automatic Kellgren and Lawrence (KL) grade classification scheme in knee radiographs. We developed a semi-supervised multi-task learning-based approach that enables the exploitation of additional unlabelled data in an unsupervised as well as supervised manner. Specifically, we propose a dual-channel adversarial autoencoder, which is first trained in an unsupervised manner for reconstruction tasks only. To exploit the additional data in a supervised way, we propose a multi-task learning framework by introducing an auxiliary task. In particular, we use leg side identification as an auxiliary task, which allows the use of more datasets, e.g., CHECK dataset. The work demonstrates that the utilization of additional data can improve the primary task of KL-grade classification for which only limited labelled data is available. This semi-supervised learning essentially helps to improve the feature learning ability of our framework, which leads to improved performance for KL-grade classification. We rigorously evaluated our proposed model on the two largest publicly available datasets for various aspects, i.e., overall performance, the effect of additional unlabelled samples and auxiliary tasks, robustness analysis, and ablation study. The proposed model achieved the accuracy, precision, recall, and F1 score of 75.53%, 74.1%, 78.51%, and 75.34%, respectively. Furthermore, the experimental results show that the suggested model not only achieves state-of-the-art performance on two publicly available datasets but also exhibits remarkable robustness.

Investigation of Vitamin D Status, Age, and Body Mass Index as Determinants of Knee Osteoarthritis Severity Using the Kellgren-Lawrence Grading System in a Saudi Arabian Cohort: A Cross-Sectional Study

Background Knee osteoarthritis (OA) is a common and disabling condition affecting millions worldwide. This cross-sectional study endeavors to investigate the relationship between vitamin D status, age, body mass index (BMI), and knee OA in a cohort of individuals in Saudi Arabia. Methods In this cross-sectional study, we assessed vitamin D serum levels, conducted knee radiographs, and evaluated the severity of knee OA using the Kellgren-Lawrence grading system (KLGS). The analysis incorporated both descriptive and inferential statistics, including chi-square tests and a regression model to investigate the relationship between KLGS grades as indicators of knee OA severity and vitamin D levels, considering demographics as covariants. Results The study included 93 participants with suspected knee OA, of which a substantial portion of the sample population presented with knee OA (58 [62.4%]). Knee OA exhibited a higher prevalence among females, comprising 47 (50.54%) of the total, while 11 (11.83%) were male. The largest age group with knee OA was those older than 58 years, 27 (29.03%), followed by the age group of 48-58 years, 19 (20.43%). Obesity was a prevalent factor among knee OA patients (36 [38.7%]), with grade 2 (17 [18.3%]) and grade 3 (24 [25.8%]) being the most frequent. Vitamin D deficiency was prevalent in 54 (58%) of patients. Among knee OA cases, bilateral involvement was predominant in 46 (79%), with a substantial portion, 36 (62%), presenting deficient vitamin D levels. The regression model revealed that age (95% CI: 0.54-1.03, p < 0.001) and BMI (95% CI: 0.01-0.60, p = 0.04) significantly predict higher KLGS grades, indicating that increasing age and higher BMI are associated with higher KLGS grades. However, Vitamin D levels did not show a significant impact on the severity of knee OA. Conclusions The findings from this study highlight the importance of monitoring and maintaining adequate vitamin D levels to potentially reduce the risk of knee OA and the need for early detection and intervention to manage knee OA, particularly in females, older poplulation, and obese adults. They may guide healthcare providers in developing comprehensive approaches to reduce the risk of this condition.

Recent targets of osteoarthritis research

Osteoarthritis is one of the most common diseases and poses a significant medical burden worldwide. Currently, the diagnosis and treatment of osteoarthritis primarily rely on clinical symptoms and changes observed in radiographs or other image modalities. However, identification based on reliable biomarkers would greatly improve early diagnosis, help with precise monitoring of disease progression, and provide aid for accurate treatment. In recent years, several biomarkers for osteoarthritis have been identified, including image modalities and biochemical biomarkers such as collagen degradation products, pro- or anti-inflammatory cytokines, micro RNAs, long non-coding RNAs, and circular RNAs. These biomarkers offer new insights in the pathogenesis of osteoarthritis and provide potential targets for further research. This article reviews the evolution of osteoarthritis biomarkers from the perspective of pathogenesis and emphasizes the importance of continued research to improve the diagnosis, treatment, and management of osteoarthritis.

Osteo-NeT: An Automated System for Predicting Knee Osteoarthritis from X-ray Images Using Transfer-Learning-Based Neural Networks Approach

Knee osteoarthritis is a challenging problem affecting many adults around the world. There are currently no medications that cure knee osteoarthritis. The only way to control the progression of knee osteoarthritis is early detection. Currently, X-ray imaging is a central technique used for the prediction of osteoarthritis. However, the manual X-ray technique is prone to errors due to the lack of expertise of radiologists. Recent studies have described the use of automated systems based on machine learning for the effective prediction of osteoarthritis from X-ray images. However, most of these techniques still need to achieve higher predictive accuracy to detect osteoarthritis at an early stage. This paper suggests a method with higher predictive accuracy that can be employed in the real world for the early detection of knee osteoarthritis. In this paper, we suggest the use of transfer learning models based on sequential convolutional neural networks (CNNs), Visual Geometry Group 16 (VGG-16), and Residual Neural Network 50 (ResNet-50) for the early detection of osteoarthritis from knee X-ray images. In our analysis, we found that all the suggested models achieved a higher level of predictive accuracy, greater than 90%, in detecting osteoarthritis. However, the best-performing model was the pretrained VGG-16 model, which achieved a training accuracy of 99% and a testing accuracy of 92%.

Tomato Leaf Disease Recognition on Leaf Images Based on Fine-Tuned Residual Neural Networks

Humans depend heavily on agriculture, which is the main source of prosperity. The various plant diseases that farmers must contend with have constituted a lot of challenges in crop production. The main issues that should be taken into account for maximizing productivity are the recognition and prevention of plant diseases. Early diagnosis of plant disease is essential for maximizing the level of agricultural yield as well as saving costs and reducing crop loss. In addition, the computerization of the whole process makes it simple for implementation. In this paper, an intelligent method based on deep learning is presented to recognize nine common tomato diseases. To this end, a residual neural network algorithm is presented to recognize tomato diseases. This research is carried out on four levels of diversity including depth size, discriminative learning rates, training and validation data split ratios, and batch sizes. For the experimental analysis, five network depths are used to measure the accuracy of the network. Based on the experimental results, the proposed method achieved the highest F1 score of 99.5%, which outperformed most previous competing methods in tomato leaf disease recognition. Further testing of our method on the Flavia leaf image dataset resulted in a 99.23% F1 score. However, the method had a drawback that some of the false predictions were of tomato early light and tomato late blight, which are two classes of fine-grained distinction.

Emergence of Deep Learning in Knee Osteoarthritis Diagnosis

Osteoarthritis (OA), especially knee OA, is the most common form of arthritis, causing significant disability in patients worldwide. Manual diagnosis, segmentation, and annotations of knee joints remain as the popular method to diagnose OA in clinical practices, although they are tedious and greatly subject to user variation. Therefore, to overcome the limitations of the commonly used method as above, numerous deep learning approaches, especially the convolutional neural network (CNN), have been developed to improve the clinical workflow efficiency. Medical imaging processes, especially those that produce 3-dimensional (3D) images such as MRI, possess ability to reveal hidden structures in a volumetric view. Acknowledging that changes in a knee joint is a 3D complexity, 3D CNN has been employed to analyse the joint problem for a more accurate diagnosis in the recent years. In this review, we provide a broad overview on the current 2D and 3D CNN approaches in the OA research field. We reviewed 74 studies related to classification and segmentation of knee osteoarthritis from the Web of Science database and discussed the various state-of-the-art deep learning approaches proposed. We highlighted the potential and possibility of 3D CNN in the knee osteoarthritis field. We concluded by discussing the possible challenges faced as well as the potential advancements in adopting 3D CNNs in this field.