Qualitative Classification of Proximal Femoral Bone Using Geometric Features and Texture Analysis in Collected MRI Images for Bone Density Evaluation DOI Creative Commons

Mojtaba Najafi,

Tohid Yousefi Rezaii,

Sebelan Danishvar

et al.

Sensors, Journal Year: 2023, Volume and Issue: 23(17), P. 7612 - 7612

Published: Sept. 2, 2023

The aim of this study was to use geometric features and texture analysis discriminate between healthy unhealthy femurs identify the most influential features. We scanned proximal femoral bone (PFB) 284 Iranian cases (21 83 years old) using different dual-energy X-ray absorptiometry (DEXA) scanners magnetic resonance imaging (MRI) machines. Subjects were labeled as “healthy” (T-score > −0.9) “unhealthy” based on results DEXA scans. Based geometry PFB in MRI, 204 retrieved. used support vector machine (SVM) with kernels, decision tree, logistic regression algorithms classifiers Genetic algorithm (GA) select best set maximize accuracy. There 185 participants classified 99 unhealthy. SVM radial basis function kernels had performance (89.08%) geometrical ones. Even though our findings show high model, further investigation more subjects is suggested. To knowledge, first that investigates qualitative classification PFBs MRI reference scans learning methods GA.

Language: Английский

Oncologic Applications of Artificial Intelligence and Deep Learning Methods in CT Spine Imaging—A Systematic Review DOI Open Access

Wilson Ong,

Aric Lee,

Wei Chuan Tan

et al.

Cancers, Journal Year: 2024, Volume and Issue: 16(17), P. 2988 - 2988

Published: Aug. 28, 2024

In spinal oncology, integrating deep learning with computed tomography (CT) imaging has shown promise in enhancing diagnostic accuracy, treatment planning, and patient outcomes. This systematic review synthesizes evidence on artificial intelligence (AI) applications CT for tumors. A PRISMA-guided search identified 33 studies: 12 (36.4%) focused detecting malignancies, 11 (33.3%) classification, 6 (18.2%) prognostication, 3 (9.1%) 1 (3.0%) both detection classification. Of the classification studies, 7 (21.2%) used machine to distinguish between benign malignant lesions, evaluated tumor stage or grade, 2 (6.1%) employed radiomics biomarker Prognostic studies included three that predicted complications such as pathological fractures AI's potential improving workflow efficiency, aiding decision-making, reducing is discussed, along its limitations generalizability, interpretability, clinical integration. Future directions AI oncology are also explored. conclusion, while technologies promising, further research necessary validate their effectiveness optimize integration into routine practice.

Language: Английский

Citations

3
A deep learning model to enhance the classification of primary bone tumors based on incomplete multimodal images in X-ray, CT, and MRI DOI Creative Commons
Liwen Song,

Chuanpu Li,

Tan Lilian

et al.

Cancer Imaging, Journal Year: 2024, Volume and Issue: 24(1)

Published: Oct. 10, 2024

Abstract Background Accurately classifying primary bone tumors is crucial for guiding therapeutic decisions. The National Comprehensive Cancer Network guidelines recommend multimodal images to provide different perspectives the comprehensive evaluation of tumors. However, in clinical practice, most patients’ medical are often incomplete. This study aimed build a deep learning model using incomplete from X-ray, CT, and MRI alongside characteristics classify as benign, intermediate, or malignant. Methods In this retrospective study, total 1305 patients with histopathologically confirmed (internal dataset, n = 1043; external 262) were included two centers between January 2010 December 2022. We proposed Primary Bone Tumor Classification Transformer (PBTC-TransNet) fusion Areas under receiver operating characteristic curve (AUC), accuracy, sensitivity, specificity calculated evaluate model’s classification performance. Results PBTC-TransNet achieved satisfactory micro-average AUCs 0.847 (95% CI: 0.832, 0.862) 0.782 0.749, 0.817) on internal test sets. For malignant tumors, respectively 0.827/0.727, 0.740/0.662, 0.815/0.745 internal/external Furthermore, across all patient subgroups stratified by distribution imaging modalities, gained ranging 0.700 0.909 0.640 sets, respectively. showed highest AUC 0.909, accuracy 84.3%, sensitivity 92.1% those only X-rays set. On set, X-ray + CT. Conclusions successfully developed externally validated transformer-based PBTC-Transnet effective model, rooted characteristics, effectively mirrors real-life scenarios, thus enhancing its strong practicability.

Language: Английский

Citations

3
A comprehensive exploration of artificial intelligence in orthopaedics within lower-middle-income countries: a narrative review DOI Creative Commons

Umm E Salma Shabbar Banatwala,

Muhammad Ibrahim,

Reyan Hussain Shaikh

et al.

Journal of the Pakistan Medical Association, Journal Year: 2024, Volume and Issue: 74(4)

Published: May 3, 2024

Integrating Artificial Intelligence (AI) in orthopaedic within lower-middle-income countries (LMICs) promises landmark improvement patient care. Delving into specific use cases—fracture detection, spine imaging, bone tumour classification, and joint surgery optimisation—the review illuminates the areas where AI can significantly enhance practices. could play a pivotal role improving diagnoses, enabling early ultimately enhancing outcomes— crucial regions with constrained healthcare services. Challenges to integration of include financial constraints, shortage skilled professionals, data limitations, cultural ethical considerations. Emphasising AI's collaborative role, it act as complementary tool working tandem physicians, aiming address gaps access education. We need continued research conscientious approach, envisioning catalyst for equitable, efficient, accessible patients LMICs. Keywords: Intelligence, Orthopaedics, Health Services, Patient Care, Bone Neoplasms, Physicians, precision medicine; predictive analysis

Language: Английский

Citations

2
The diagnostic value of machine learning for the classification of malignant bone tumor: a systematic evaluation and meta-analysis DOI Creative Commons
Yue Li, Bo Dong, P. Yuan

et al.

Frontiers in Oncology, Journal Year: 2023, Volume and Issue: 13

Published: Sept. 7, 2023

Background Malignant bone tumors are a type of cancer with varying malignancy and prognosis. Accurate diagnosis classification crucial for treatment prognosis assessment. Machine learning has been introduced early differential malignant tumors, but its performance is controversial. This systematic review meta-analysis aims to explore the diagnostic value machine tumors. Methods PubMed, Embase, Cochrane Library, Web Science were searched literature on in up October 31, 2022. The risk bias assessment was conducted using QUADAS-2. A bivariate mixed-effects model used meta-analysis, subgroup analyses by methods modeling approaches. Results inclusion comprised 31 publications 382,371 patients, including 141,315 Meta-analysis results showed sensitivity specificity 0.87 [95% CI: 0.81,0.91] 0.91 0.86,0.94] training set, 0.83 0.74,0.89] 0.79,0.92] validation set. Subgroup analysis revealed MRI-based radiomics most common approach, 0.85 0.74,0.91] 0.79 0.70,0.86] Convolutional neural networks type, 0.86 0.72,0.94] 0.92 0.82,0.97] 0.51,0.98] 0.69,0.96] Conclusion mainly applied diagnosing showing desirable performance. can be an adjunctive method requires further research determine practical efficiency clinical application prospects. Systematic registration https://www.crd.york.ac.uk/prospero/ , identifier CRD42023387057.

Language: Английский

Citations

4
Machine Learning–Assisted Decision Making in Orthopaedic Oncology DOI
Paul A. Rizk, Marcos R. Gonzalez, Bishoy Galoaa

et al.

JBJS Reviews, Journal Year: 2024, Volume and Issue: 12(7)

Published: July 1, 2024

» Artificial intelligence is an umbrella term for computational calculations that are designed to mimic human and problem-solving capabilities, although in the future, this may become incomplete definition. Machine learning (ML) encompasses development of algorithms or predictive models generate outputs without explicit instructions, assisting clinical predictions based on large data sets. Deep a subset ML utilizes layers networks use various inter-relational connections define generalize data. can enhance radiomics techniques improved image evaluation diagnosis. While shows promise with advent radiomics, there still obstacles overcome. Several calculators leveraging have been developed predict survival primary sarcomas metastatic bone disease utilizing patient-specific these often report exceptionally accurate performance, it crucial evaluate their robustness using standardized guidelines. increased computing power suggests continuous improvement algorithms, advancements must be balanced against challenges such as diversifying data, addressing ethical concerns, enhancing model interpretability.

Language: Английский

Citations

1
Differential diagnostic value of radiomics models in benign versus malignant vertebral compression fractures: A systematic review and meta-analysis DOI
Jiayuan Zheng,

W. Y. Liu,

Jianan Chen

et al.

European Journal of Radiology, Journal Year: 2024, Volume and Issue: 178, P. 111621 - 111621

Published: July 14, 2024

Language: Английский

Citations

1
Conventional radiography for the assessment of focal bone lesions of the appendicular skeleton: fundamental concepts in the modern imaging era DOI Creative Commons
George R. Matcuk, Leah E. Waldman, Brandon K.K. Fields

et al.

Skeletal Radiology, Journal Year: 2024, Volume and Issue: unknown

Published: Dec. 24, 2024

Abstract Bone lesions of the appendicular skeleton can be caused by primary benign or malignant tumors, metastases, osteomyelitis, pseudotumors. Conventional radiography plays a crucial role in initial assessment osseous and should not underestimated even this era modern complex advanced imaging technologies. Combined with patient age, clinical symptoms biology, lesion features including location, solitary versus multiplicity, density, margin (transitional zone evaluated Lodwick-Madewell grading score), and, if present, type periosteal reaction matrix mineralization narrow differential diagnosis offer likely diagnosis. These radiographic help guide further follow-up management.

Language: Английский

Citations

1
Qualitative Classification of Proximal Femoral Bone Using Geometric Features and Texture Analysis in Collected MRI Images for Bone Density Evaluation DOI Creative Commons

Mojtaba Najafi,

Tohid Yousefi Rezaii,

Sebelan Danishvar

et al.

Sensors, Journal Year: 2023, Volume and Issue: 23(17), P. 7612 - 7612

Published: Sept. 2, 2023

The aim of this study was to use geometric features and texture analysis discriminate between healthy unhealthy femurs identify the most influential features. We scanned proximal femoral bone (PFB) 284 Iranian cases (21 83 years old) using different dual-energy X-ray absorptiometry (DEXA) scanners magnetic resonance imaging (MRI) machines. Subjects were labeled as “healthy” (T-score > −0.9) “unhealthy” based on results DEXA scans. Based geometry PFB in MRI, 204 retrieved. used support vector machine (SVM) with kernels, decision tree, logistic regression algorithms classifiers Genetic algorithm (GA) select best set maximize accuracy. There 185 participants classified 99 unhealthy. SVM radial basis function kernels had performance (89.08%) geometrical ones. Even though our findings show high model, further investigation more subjects is suggested. To knowledge, first that investigates qualitative classification PFBs MRI reference scans learning methods GA.

Language: Английский

Citations

0