Comparing various AI approaches to traditional quantitative assessment of the myocardial perfusion in [82Rb] PET for MACE prediction

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: April 26, 2024

Abstract Assessing the individual risk of Major Adverse Cardiac Events (MACE) is major importance as cardiovascular diseases remain leading cause death worldwide. Quantitative Myocardial Perfusion Imaging (MPI) parameters such stress Blood Flow (sMBF) or Reserve (MFR) constitutes gold standard for prognosis assessment. We propose a systematic investigation value Artificial Intelligence (AI) to leverage [ $$^{82}$$ 82 Rb] Silicon PhotoMultiplier (SiPM) PET MPI MACE prediction. establish general pipeline AI model validation assess and compare performance global (i.e. average entire signal), regional (17 segments), radiomics Convolutional Neural Network (CNN) models leveraging various signals on dataset 234 patients. Results showed that all significantly outperformed ( $$p<0.001$$ p < 0.001 ), where best AUC 73.9% (CI 72.5–75.3) was obtained with CNN model. A based MBF averages from 17 segments fed Logistic Regression (LR) constituted an excellent trade-off between simplicity performance, achieving 73.4% 72.3–74.7). intensity features revealed least important feature when compared other aggregations signal over myocardium. conclude can allow better personalized assessment MACE.

Language: Английский

---

Myocardial perfusion SPECT radiomic features reproducibility assessment: Impact of image reconstruction and harmonization

Medical Physics, Journal Year: 2024, Volume and Issue: 52(2), P. 965 - 977

Published: Oct. 29, 2024

Coronary artery disease (CAD) has one of the highest mortality rates in humans worldwide. Single-photon emission computed tomography (SPECT) myocardial perfusion imaging (MPI) provides clinicians with metabolic information non-invasively. However, there are some limitations to interpreting SPECT images performed by physicians or automatic quantitative approaches. Radiomics analyzes objectively extracting features and can potentially reveal biological characteristics that human eye cannot detect. reproducibility repeatability radiomic be highly susceptible segmentation conditions. We aimed assess extracted from uncorrected MPI-SPECT reconstructed 15 different settings before after ComBat harmonization, along evaluating effectiveness realigning feature distributions. A total 200 patients (50% normal 50% abnormal) including rest stress (without attenuation scatter corrections) were included. Images using combinations filter cut-off frequencies, orders, types, reconstruction algorithms, number iterations subsets resulting 6000 images. Image was on left ventricle first for each patient applied 14 others. 93 segmented area, used harmonize them. The intraclass correlation coefficient (ICC) overall concordance (OCCC) tests examine impact parameter robustness harmonization efficiency. ANOVA Kruskal-Wallis evaluate correcting In addition, Student's t-test, Wilcoxon rank-sum, signed-rank implemented significance level impacts made batches groups (normal vs. features. Before applying ComBat, majority (ICC: 82, OCCC: 61) achieved high (ICC/OCCC ≥ 0.900) under every batch except Reconstruction. largest smallest poor < 0.500) obtained IterationSubset Order batches, respectively. most reliable first-order (FO) gray-level co-occurrence matrix (GLCM) families. Following minimum robust increased 84, 78). Applying showed Reconstruction least responsive families, a descending order, found FO, neighborhood gray-tone difference (NGTDM), GLCM, run length (GLRLM), size zone (GLSZM), dependence (GLDM) Cut-off, Filter, batches. rank-sum test significantly differed Normal Abnormal groups. show levels across OSEM parameters MPI-SPECT. is effective distributions enhancing reproducibility.

Language: Английский

---

Deep learning automatically distinguishes myocarditis patients from normal subjects based on MRI

The International Journal of Cardiovascular Imaging, Journal Year: 2024, Volume and Issue: unknown

Published: Nov. 7, 2024

Myocarditis, characterized by inflammation of the myocardial tissue, presents substantial risks to cardiovascular functionality, potentially precipitating critical outcomes including heart failure and arrhythmias. This investigation primarily aims identify optimal magnetic resonance imaging (CMRI) views for distinguishing between normal myocarditis cases, using deep learning (DL) methodologies. Analyzing CMRI data from a cohort 269 individuals, with 231 confirmed cases 38 as control participants, we implemented an innovative DL framework facilitate automated detection myocarditis. Our approach was divided into single-frame multi-frame analyses evaluate different types acquisitions diagnostic accuracy. The results demonstrated weighted accuracy 96.9%, highest achieved late gadolinium enhancement (LGE) 2-chamber view, underscoring potential in on data.

Language: Английский

---

Dual-Centre Harmonised Multimodal Positron Emission Tomography/Computed Tomography Image Radiomic Features and Machine Learning Algorithms for Non-small Cell Lung Cancer Histopathological Subtype Phenotype Decoding

Clinical Oncology, Journal Year: 2023, Volume and Issue: 35(11), P. 713 - 725

Published: Aug. 8, 2023

Language: Английский

---

Artificial Intelligence Empowered Nuclear Medicine and Molecular Imaging in Cardiology: A State-of-the-Art Review

Phenomics, Journal Year: 2023, Volume and Issue: 3(6), P. 586 - 596

Published: Dec. 1, 2023

Language: Английский

---

Robust versus Non-Robust Radiomic features: Machine Learning Based Models for NSCLC Lymphovascular Invasion

2021 IEEE Nuclear Science Symposium and Medical Imaging Conference (NSS/MIC), Journal Year: 2022, Volume and Issue: unknown, P. 1 - 3

Published: Nov. 5, 2022

The application of radiomic features for predicting and diagnosing lung cancer has been established in previous studies. However, might be weak terms reproducibility. This study aimed to determine robust against motion a multicenter study, followed by choosing bold using various feature selection methods comparing them with the standard method results without considering robustness features. To this end, an in-house developed phantom was used. A specific motor manufactured simulate 2 orthogonal directions. Lesions tumors were delineated semi-automatic segmentation. In total, 105 extracted. Three different selections five machine learning classifiers implanted study. First, Intraclass Correlation Coefficient (ICC) calculated show variability select ICC more than 90%. Next, selected went through methods. Finally, compared outcomes regular multiple imbalanced clinical data set. Our result demonstrated that although minor negative impact on prediction accuracy, it significant productive sensitivity.

Language: Английский

---

Analysis of Machine and Deep Learning Algorithms for Pattern Recognition in Medical Data

Published: Feb. 23, 2024

Pattern recognition is a data analysis technique that utilizes various algorithms for the automatic of patterns and regularities. Recently, problems scenes need pattern quick resolution difficult issues, particularly those can't resolved by multiple dimensional data, because involved in spectral information. In this study, different machine learning (ML) deep (DL) techniques are analyzed which implemented using medical data. This study discussed significant assumptions, advantages, drawbacks ML DL techniques. Different Artificial Neural Networks (ANN), Machine Learning Regression (MLR), so on. Various Convolutional (CNN), EfficientNet performance measures like accuracy, precision, recall, f1-score error rates used previous studies evaluation study. The concludes have potential to overcome every drawback there option integrating method developing an ensemble technique.

Language: Английский

---

Radiomics Nomogram Derived from Gated Myocardial Perfusion SPECT for Identifying Ischemic Cardiomyopathy

Deleted Journal, Journal Year: 2024, Volume and Issue: 37(6), P. 2784 - 2793

Published: May 28, 2024

Personalized management involving heart failure (HF) etiology is crucial for better prognoses. We aim to evaluate the utility of a radiomics nomogram based on gated myocardial perfusion imaging (GMPI) in distinguishing ischemic from non-ischemic origins HF. A total 172 patients with reduced left ventricular ejection fraction (HFrEF) who underwent GMPI scan were divided into training (n = 122) and validation sets 50) chronological order scans. Radiomics features extracted resting GMPI. Four machine learning algorithms used construct models, model best performances selected calculate Radscore. was constructed Radscore independent clinical factors. Finally, performance validated using operating characteristic curves, calibration curve, decision curve analysis, integrated discrimination improvement values (IDI), net reclassification index (NRI). Three optimal build model. Total deficit (TPD) identified as factors conventional metrics building In set, integrating Radscore, age, systolic blood pressure, TPD significantly outperformed cardiomyopathy (ICM) (NICM) (AUC 0.853 vs. 0.707, p 0.038). IDI analysis indicated that improved diagnostic accuracy by 28.3% compared set. By combining signatures indicators, we developed GMPI-based helps identify HFrEF.

Language: Английский

---

A Novel Approach to Identifying Hibernating Myocardium Using Radiomics-Based Machine Learning

Cureus, Journal Year: 2024, Volume and Issue: unknown

Published: Sept. 16, 2024

Background To assess the feasibility of a machine learning (ML) approach using radiomics features perfusion defects on rest myocardial imaging (MPI) to detect presence hibernating myocardium. Methodology Data patients who underwent 99mTc-sestamibi MPI and 18F-FDG PET/CT for viability assessment were retrieved. Rest data processed ECToolbox, polar maps saved NFile PMap tool. The reference standard defining myocardium was mismatched perfusion-metabolism defect with impaired contractility at rest. Perfusion delineated regions interest (ROIs) after spatial resampling intensity discretization. Replicable random sampling allocated 80% (257) from January 2017 September 2022 training set remaining 20% (64) validation set. An independent dataset 29 consecutive October 2023 used as testing model evaluation. One hundred ten first second-order texture extracted each ROI. After feature normalization imputation, 14 best-ranked selected multistep selection process including Logistic Regression Fast Correlation-Based Filter. Thirteen supervised ML algorithms trained stratified five-fold cross-validation validated Log Loss <0.688 <0.672 in steps evaluated Performance matrices assessed included area under curve (AUC), classification accuracy (CA), F1 score, precision, recall, specificity. provide transparency interpretability, SHapley Additive exPlanations (SHAP) values depicted beeswarm plots. Results Two thirty-nine (214 males; mean age 56 ± 11 years) enrolled study. There 371 (321 sets; 50 set). Based standard, 168 had (139 On cross-validation, six AUC >0.800. validation, 10 value <0.672, among which evaluation models unseen set, nine >0.800 Gradient Boosting Random Forest (xgboost) [GB RF (xgboost)] achieving highest 0.860 could 21/29 (72.4%) precision 87.5% (21/24), specificity 85.7% (18/21), CA 78.0% (39/50) Score 0.792. Four clear pattern interpretability based SHAP These GB (xgboost), (scikit-learn), Forest. Conclusion Our study demonstrates potential detecting images. This proof-of-concept underscores notion that capture nuanced information beyond what is perceptible human eye, offering promising avenues improved assessment.

Language: Английский

---

Comparison of Machine Learning Algorithms Using Manual/Automated Features on 12-Lead Signal Electrocardiogram Classification: A Large Cohort Study on Students Aged Between 6 to 18 Years Old

Cardiovascular Engineering and Technology, Journal Year: 2023, Volume and Issue: 14(6), P. 786 - 800

Published: Oct. 17, 2023

Language: Английский

---