Figure 1 - uploaded by Eni Halilaj
Content may be subject to copyright.
Kellgren-Lawrence grading scale. The scale ranges from 0 to 4. A score of 0 indicates that there is no evidence of osteoarthritis (OA; green); a score of 1 indicates the possibility of joint space narrowing (orange) and osteophyte formation (blue); a score of 2 indicates definite osteophyte formation and possible joint space narrowing; a score of 3 indicates multiple osteophytes, definite joint space narrowing, sclerosis (purple), and possibly bone deformity (pink); a score of 4 indicates end-stage OA, marked by severe sclerosis, joint space narrowing (sometimes bone-on-bone contact), and large osteophytes (5).
Source publication
Purpose:
To develop an automated model for staging knee osteoarthritis severity from radiographs and to compare its performance to that of musculoskeletal radiologists.
Materials and methods:
Radiographs from the Osteoarthritis Initiative staged by a radiologist committee using the Kellgren-Lawrence (KL) system were used. Before using the images...
Similar publications
This paper focused on creating an interpretable model for automatic rapid eye movement (REM) and non-REM sleep stage scoring for a single-channel electroencephalogram (EEG). Many methods attempt to extract meaningful information to provide to a learning algorithm. This method attempts to let the model extract the meaningful interpretable informatio...
This tutorial demonstrates workflows to incorporate text data into actuarial classification and regression tasks. The main focus is on methods employing transformer-based models. A dataset of car accident descriptions with an average length of 400 words, available in English and German, and a dataset with short property insurance claims description...
Multi-omic single-cell technologies, which simultaneously measure the transcriptional and epigenomic state of the same cell, enable understanding epigenetic mechanisms of gene regulation. However, noisy and sparse data pose fundamental statistical challenges to extract biological knowledge from complex datasets. SHARE-Topic, a Bayesian generative m...
Several recent studies have established that seismic full-waveform inversion (FWI) can be used to generate interpretable models of acoustic reflectivity from practically raw seismic data. Owing to their
use of the full wavefield and an iterative least-squares approach to optimisation, these models, referred to as FWI images, offer an improvement in...
In recent years, fake news detection and its characteristics have attracted a number of researchers. However, most detection algorithms are driven by data rather than theories, which causes the existing approaches to only perform well on specific datasets. To the extreme, several features only perform well on specific datasets. In this study, we fi...
Citations
... Encoder-decoder architectures like the UNet and nnUNet have been effective in identifying osseous regions and soft tissues in 2D and 3D scans [15], [16], [18]- [21]. CNNs have also been studied for OA staging and treatment prediction [22]- [24]. In this context, core research gaps may be synthesized. ...
... Their encoder-decoder design, with skip connections, captures image features at different resolutions enhancing the anatomical identification. These architectures were extensively studied for the segmentation of knee, ankle, and shoulder bones in CT and MRI scans, and vertebral bodies in CTs [22], [30], [31]. Reduced segmentation accuracy and model generalization ability were assessed when training networks with small datasets [32], [33]. ...
... However, the same drawbacks might persist even after increasing the dataset numerosity [35]. Likewise, CNNs were proposed to evaluate cartilage osteonecrosis in knee X-ray images using the Kellgren-Lawrence scoring, achieving results comparable to expert operators [17], [22]. In [36], the authors tested various pre-trained models, such as ResNet and DenseNet, for discriminating fracture/nonfracture conditions in X-ray images. ...
Goal: Effective preoperative planning for shoulder joint replacement requires accurate glenohumeral joint (GH) digital surfaces and reliable clinical staging. Methods: xCELUNet was designed as a dual-task deep network for humerus and scapula bone reconstruction in CT scans, and assessment of three GH joint clinical conditions, namely osteophyte size (OS), joint space reduction (JS), and humeroscapular alignment (HSA). Results: Trained on a dataset of 571 patients, the model optimized segmentation and classification through transfer learning. It achieved median root mean squared errors of 0.31 and 0.24 mm, and Hausdorff distances of 2.35 and 3.28 mm for the humerus and scapula, respectively. Classification accuracy was 91 for OS, 93 for JS, and 85% for HSA. GradCAM-based activation maps validated the network's interpretability. Conclusions: this framework delivers accurate 3D bone surface reconstructions and dependable clinical assessments of the GH joint, offering robust support for therapeutic decision-making in shoulder arthroplasty.
... Generally, the MRI plays a vital part in detecting the biochemical and morphologic features that provide an in-depth understanding of patterns. So, the suggested study [23] has MRI-based studies to conduct the identification of lesion severity in the ACL, meniscus, bone marrow, and cartilage. A three-dimensional CNN has been developed to identify the Region of Interest (ROI) and then grade the abnormalities. ...
Bone fractures are common in diabetic patients and can result in several musculoskeletal conditions. Both type 1 and type 2 diabetes substantially increase the risk and severity of bone fractures. Prompt treatment and management of diabetes and its complications are crucial to mitigate this serious complication. Detection and diagnosis in its early stage can reduce the challenging conditions in treatment. Traditional image processing techniques like digital-geometric analysis, entropy measures, and gray-level co-occurrence matrices have been used for automated bone fracture detection. However, these detection methods rely neither on healthy controls nor diabetic-affected patients. Only few studies focused on detecting fractures in diabetic patients. The rising prevalence of diabetic ankle fractures made the study emphasize the development of a fracture detection model based on the Meta Magnify (MetaMag) efficiency model. The proposed model involves the Lower Extremity Radiographs (LERA) dataset, which consists of image samples of normal and abnormal lower extremities of the body, such as the hip, ankle, knee, and foot. Pre-processing involves a one-hot encoding method that handles the missing data and represents categorical variables as numerical values. Further, the classification is performed using the MetaMag efficiency model, incorporated with MetaMag scaling and unified normalization. Further, the efficiency of the proposed model is analyzed by comparing it with conventional EfficientNet and another model. Finally, the proposed work's performance is analyzed using evaluation measures such as accuracy, precision, recall and F1-score. The results indicate the improved efficiency of the model.
... posing challenges in diagnosis and treatment due to its progressive nature [2]. Early detection and accurate classification are pivotal for effective management and improved patient outcomes [4]. In this context, leveraging advancements in deep learning techniques, particularly Convolutional Neural Networks (CNNs), offers promising avenues for enhancing knee OA classification [5]. ...
... The Data Flow Diagram (DFD) for the knee osteoarthritis classification project illustrates the flow of data and processes involved in the system [4]. At the highest level, the DFD captures the interactions between different modules and components, including data preprocessing, CNN model training, Streamlit interface, evaluation, and deployment. ...
... The core functionality of this project revolves around leveraging state-of-the-art machine learning techniques, specifically Convolutional Neural Networks (CNNs), to accurately classify knee osteoarthritis from X-ray images. Through a user-friendly web interface powered by Streamlit, users can effortlessly upload knee X-ray images for analysis [4]- [5]. The uploaded images undergo preprocessing to ensure optimal input for the CNN model. ...
Knee osteoarthritis is a degenerative joint disease that affects millions worldwide. Early detection and classification are crucial for effective treatment and management. This study proposes a computer-aided diagnosis system using X-ray images to detect and classify knee osteoarthritis. The system employs deep learning techniques to analyze X-ray images and classify the severity of osteoarthritis based on standardized radiographic criteria. The results show high accuracy in detecting osteoarthritis and classifying its severity, demonstrating the potential of this system to assist clinicians in early diagnosis and treatment planning. A new reality of transforming diagnostic medicine. An aggregated-based deep learning method for leukemic B-lymphoblast classification. Classification using deep-neural-network-based features. Automated classification of radiographic knee osteoarthritis severity using deep neural networks. Enabling early detection of osteoarthritis from presymptomatic cartilage texture maps via transport-based learning.
... Thomas et al. [11] proposed an ordinal regression module (ORM) with neural networks, to classify the KOA by using the KL grading system. They compared the outcomes of their proposed model with neural network approaches. ...
... This layer aimed to alleviate training time constraints and model complexity. Equation (11) is used for employing the dropout process in the proposed model. ...
... This output is produced by the dense layer. The final output of the KOC_Net model is generated by a dense layer comprising five neurons, which employs the SoftMax activation function [11]. SoftMax is an activation function that operates based on probability, where the quantity of neurons is equivalent to the total number of classes. ...
One of the most common diseases afflicting humans is knee osteoarthritis (KOA). KOA occurs when the knee joint cartilage breaks down, and knee bones start rubbing together. The diagnosis of KOA is a lengthy process, and missed diagnosis can have serious consequences. Therefore, the diagnosis of KOA at an initial stage is crucial which prevents the patients from Severe complications. KOA identification using deep learning (DL) algorithms has gained popularity during the past few years. By applying knee X-ray images and the Kellgren–Lawrence (KL) grading system, the objective of this study was to develop a DL model for detecting KOA. This study proposes a novel model based on CNN called knee osteoarthritis classification network (KOC_Net). The KOC_Net model contains 05 convolutional blocks, and each convolutional block has three components such as Convlotuioanl2D, ReLU, and MaxPooling 2D. The KOC_Net model is evaluated on two publicly available benchmark datasets which consist of X-ray images of KOA based on the KL grading system. Additionally, we applied contrast-limited adaptive histogram equalization (CLAHE) methods to enhance the contrast of the images and utilized SMOTE Tomek to deal with the problem of minority classes. For the diagnosis of KOA, the classification performance of the proposed KOC_Net model is compared with baseline deep networks, namely Dense Net-169, Vgg-19, Xception, and Inception-V3. The proposed KOC_Net was able to classify KOA into 5 distinct groups (including Moderate, Minimal, Severe, Doubtful, and Healthy), with an AUC of 96.71%, accuracy of 96.51%, recall of 91.95%, precision of 90.25%, and F1-Score of 96.70%. Dense Net-169, Vgg-19, Xception, and Inception-V3 have relative accuracy rates of 84.97%, 81.08%, 87.06%, and 83.62%. As demonstrated by the results, the KOC_Net model provides great assistance to orthopedics in making diagnoses of KOA.
... Thomas et al 21 . used a convolutional Neural Network model from assessing knee OA severity through x-ray images. ...
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%.
... For example, Christos Kokkotis et al [6] incorporated Support Vector Machines in their research. Specific studies utilized Convolutional Neural Network (CNN) models referenced by [3], [7], and [2]. Meanwhile, Joseph Humberto Cueva et al. [8] and Bofei Zhang et al. [9] [10] employed the EfcientNet-B0 model, with achieving classification rates of 72.00%. ...
... Our model, while being less complex, is advanced and user-friendly, proving its effectiveness. Article Accuracy Precision Recall F1-score Specificity [6] 74.07 ---- [3] 69.70 ---- [7] 71.00 ---- [9] 74.81 ---- [2] 66.71 ---- [8] 61.00 ---- [10] 72 VII. THREATS TO VALIDITY Although our research uses augmentation to address dataset imbalance, it is not without limits. ...
Abstract—Knee osteoarthritis (OA) significantly limits activity
and causes physical disability in older adults. Early classification
of OA is crucial to slowing its progression. This paper intro�duces a method for OA classification using a transfer learning
fusion network with hyperparameter tuning and a Multi-Layer
Ensemble approach. The process begins by balancing the dataset
through data augmentation and then integrates a pre-trained
model with a customized CNN model. The architecture includes
four convolutional blocks with varying filter sizes (32, 64, 128, and
256), dropout layers (0.1, 0.3, 0.5, and 0.7), and Mish activation
functions. The flattened output of the last MaxPool2D layer is fed
into three fully connected layers (256, 128, and 5 neurons) with
Softmax activation for multi-class prediction. The performance
of various models is combined through a Multi-Layer Ensemble,
resulting in superior performance with 75.73% accuracy. Our
proposed method achieves a well-performing model for OA
classification and overcomes prior limitations, emphasizing the
importance of automated knee OA classification and providing
an effective solution.
... Classification results for KL grades 0 and 4 were excellent; KL grade 1 was good, while KL grade 2 and KL grade 3 were failures. [14,[44][45][46] Region of Interest(ROI) segmentation identifies a sub-frame or sub-region of interest within a given frame or image Histogram Equalization [14,47] Histogram Equalization is an image processing technique that adjusts the contrast of an image by stretching out the intensity range of an image [48] CLAHE [21,49,50] Contrast Limited Adaptive Histogram Equalization is an adaptive histogram equalization that prevents contrast over-amplification and is applied for noise reduction HOG [13,46,51,52] A Histogram of Oriented Gradients is a feature descriptor of an image's gradients or edges Random Forest [44,53] A popular Machine Learning algorithm that decides following multiple decision trees CNN [49,52,[54][55][56] Convolutional Neural Networks are commonly used to extract features from images Siamese Network [47,[57][58][59] A Siamese neural network is a network with two or more identical subnetworks with the same weights but work on different input vectors FasterRCNN [60,61] Faster-RCNN is a deep CNN used for Object detection Autoencoder [62,63] Autoencoders are neural networks that can be used to remove noise from data DenseNet [60,62,[64][65][66][67][68][69] DenseNet is a type of CNN that connects each layer to each layer through dense blocks that promote learning and reduce parameters ResNet [30,41,47,60,64,[70][71][72][73][74][75][76][77] A type of deep learning model that works with residual connections for better information flow and prevents gradients from vanishing Region Proposal Network [61,78] Region Proposal Network (RPN) is a CNN that efficiently identifies multiple objects in an image using a wide range of scale and aspect ratios VGG [21,30,60,70,74,[79][80][81] Visual Geometric Group Net is a standard deep CNN with multiple layers for localization and classification tasks Attention-based [82][83][84] Attention-based models are deep learning models that focus on specific inputs. Convolutional Block Attention Module (CBAM) is an attention module for conventional neural networks Ordinal Loss [30,60,80] KL grading can be treated as an ordinal classification task Since KL grades are in an order (0-4). ...
... Convolutional Block Attention Module (CBAM) is an attention module for conventional neural networks Ordinal Loss [30,60,80] KL grading can be treated as an ordinal classification task Since KL grades are in an order (0-4). An ordinal loss is used to train the DL model GradCAM [60,80] Gradient-Weighted Class Activation Maps used to reflect features learned by the model Saliency Map [68,69] To visualize key features learned by the model EigenCAM [30,66] EigenCAM visualizes the key features learned by the convolutional layer Gornale et al. [53] used a semi-automated method to segment a knee X-ray image. Various features are extracted and classified using a Random Forest classifier. ...
... Thomas et al. [69] developed an automated model to assess the severity of OA using knee radiographs. The model's performance is compared with the musculoskeletal radiologists. ...
Osteoarthritis (OA) is a musculoskeletal disorder that afects weight-bearing joints like the hip, knee, spine, feet, and fn-
gers. It is a chronic disorder that causes joint stifness and leads to functional impairment. Knee osteoarthritis (KOA) is a
degenerative knee joint disease that is a signifcant disability for over 60 years old, with the most prevalent symptom of knee
pain. Radiography is the gold standard for the evaluation of KOA. These radiographs are evaluated using diferent clas-
sifcation systems. Kellgren and Lawrence’s (KL) classifcation system is used to classify X-rays into fve classes (Normal
= 0 to Severe = 4) based on osteoarthritis severity levels. In recent years, with the advent of artifcial intelligence, machine
learning, and deep learning, more emphasis has been given to automated medical diagnostic systems or decision support
systems. Computer-aided diagnosis is needed for the improvement of health-related information systems. This survey aims
to review the latest advances in automated radiographic classifcation and detection of KOA using the KL system. A total
of 85 articles are reviewed as original research or survey articles. This survey will beneft researchers, practitioners, and
medical experts interested in X-rays-based KOA diagnosis and prediction.
... Thomas et al. [13] developed an automatic CNN-based model for assessing severity of knee osteoarthritis from x-ray images, achieving 0.70 as F1 score and 0.71 as the accuracy. The author of [14] used a dataset of 201 randomly chosen osteophytes to compare the performance of their classifier to an extrinsic method. ...
... During the past few years, there have been several efforts [3][4][5][6][7][8][9][10] to automatically classify radiographic severity of a knee with the aid of convolutional neural network (CNN), and the results have been promising. Deep learning (DL) methods are commonly used in this automatic KLG grading task since large scale data are utilized to improve model accuracy. ...
... Therefore, the purpose of this study was to develop a prediction model that automatically assesses the KLG of a knee image applying PIM. The authors tried to develop a model with better accuracy and better generalization in classifying the KLG of a knee than the models provided in the previous literature [3][4][5][6][7][8][9][10]. ...
... The most important basis of our model development was that classifying KOA severity using KLG is a fine-grained classification tasks. The previous 14 models [3][4][5][6][7][8][9][10][24][25][26][27][28][29] were all CNN-based and have been successful in discerning KLGs 3 or 4 since these grades have distinct features: joint space narrowing. However, these previous models showed relatively low accuracy in lower grades (KLGs 0, 1, and 2). ...
Background
Fine-grained classification deals with data with a large degree of similarity, such as cat or bird species, and similarly, knee osteoarthritis severity classification [Kellgren–Lawrence (KL) grading] is one such fine-grained classification task. Recently, a plug-in module (PIM) that can be integrated into convolutional neural-network-based or transformer-based networks has been shown to provide strong discriminative regions for fine-grained classification, with results that outperformed the previous deep learning models. PIM utilizes each pixel of an image as an independent feature and can subsequently better classify images with minor differences. It was hypothesized that, as a fine-grained classification task, knee osteoarthritis severity may be classified well using PIMs. The aim of the study was to develop this automated knee osteoarthritis classification model.
Methods
A deep learning model that classifies knee osteoarthritis severity of a radiograph was developed utilizing PIMs. A retrospective analysis on prospectively collected data was performed. The model was trained and developed using the Osteoarthritis Initiative dataset and was subsequently tested on an independent dataset, the Multicenter Osteoarthritis Study (test set size: 17,040). The final deep learning model was designed through an ensemble of four different PIMs.
Results
The accuracy of the model was 84%, 43%, 70%, 81%, and 96% for KL grade 0, 1, 2, 3, and 4, respectively, with an overall accuracy of 75.7%.
Conclusions
The ensemble of PIMs could classify knee osteoarthritis severity using simple radiographs with a fine accuracy. Although improvements will be needed in the future, the model has been proven to have the potential to be clinically useful.
... Thomas et al. [38] built DenseNet169 from scratch with 169 layers to diagnose knee osteoarthritis severity according to the KL classification system using radiographic images. In this model, each pair of layers was connected in such a way that contours and primitive shapes could be used directly in the fully connected layer. ...
... Secondly, the models that obtained low predictions did not indicate whether the training of the models used was saturated or not, bearing in mind that in the event of non-saturation, increasing the number of epochs used could improve their predictions. Finally, although the majority of evaluation methods used were based on mean accuracy, several studies used other evaluation criteria such as mean F1 score, mean recall, and mean precision [7,38], making the results incomparable. Lastly, the Scopus index was not considered in the selection of reviewed papers, which would have enhanced the findings of this literature review. ...
Knee osteoarthritis is a chronic, progressive disease that rapidly progresses to severe stages. Reliable and accurate diagnosis, combined with the implementation of preventive lifestyle modifications before irreversible damage occurs, can effectively protect patients from becoming an inactive population. Artificial intelligence continues to play a pivotal role in computer-aided diagnosis with increasingly convincing accuracy, particularly in identifying the severity of knee osteoarthritis according to the Kellgren–Lawrence (KL) grading scale. The primary objective of this literature review is twofold. Firstly, it aims to provide a systematic analysis of the current literature on the main artificial intelligence models used recently to predict the severity of knee osteoarthritis from radiographic images. Secondly, it constitutes a critical review of the different methodologies employed and the key elements that have improved diagnostic performance. Ultimately, this study demonstrates that the considerable success of artificial intelligence systems will reinforce healthcare professionals’ confidence in the reliability of machine learning algorithms, facilitating more effective and faster treatment for patients afflicted with knee osteoarthritis. In order to achieve these objectives, a qualitative and quantitative analysis was conducted on 60 original research articles published between 1 January 2018 and 15 May 2024.