The Role of Artificial Intelligence and Machine Learning in Cardiovascular Imaging and Diagnosis

Cureus, Год журнала: 2024, Номер unknown

Опубликована: Сен. 2, 2024

Cardiovascular diseases remain the leading cause of global mortality, underscoring critical need for accurate and timely diagnosis. This narrative review examines current applications future potential artificial intelligence (AI) machine learning (ML) in cardiovascular imaging. We discuss integration these technologies across various imaging modalities, including echocardiography, computed tomography, magnetic resonance imaging, nuclear techniques. The explores AI-assisted diagnosis key areas such as coronary artery disease detection, valve disorders assessment, cardiomyopathy classification, arrhythmia prediction events. AI demonstrates promise improving diagnostic accuracy, efficiency, personalized care. However, significant challenges persist, data quality standardization, model interpretability, regulatory considerations, clinical workflow integration. also address limitations ethical implications their implementation practice. Future directions point towards advanced architectures, multimodal integration, precision medicine population health management. emphasizes ongoing collaboration between clinicians, scientists, policymakers to realize full while ensuring equitable implementation. As field continues evolve, addressing will be crucial successful into care, potentially revolutionizing capabilities patient outcomes.

Язык: Английский

---

Evaluation of 18F-FDG absorbed dose ratios in percent in adult and pediatric reference phantoms using DoseCalcs Monte Carlo platform

Applied Radiation and Isotopes, Год журнала: 2025, Номер 218, С. 111705 - 111705

Опубликована: Фев. 8, 2025

Язык: Английский

---

The Evolution of Artificial Intelligence in Nuclear Medicine

Seminars in Nuclear Medicine, Год журнала: 2025, Номер unknown

Опубликована: Фев. 1, 2025

Nuclear medicine has continuously evolved since its beginnings, constantly improving the diagnosis and treatment of various diseases. The integration artificial intelligence (AI) is one latest revolutionizing chapters, promising significant advancements in diagnosis, prognosis, segmentation, image quality enhancement, theranostics. Early AI applications nuclear focused on diagnostic accuracy, leveraging machine learning algorithms for disease classification outcome prediction. Advances deep learning, including convolutional more recently transformer-based neural networks, have further enabled precise segmentation as well low-dose imaging, patient-specific dosimetry personalized treatment. Generative AI, driven by large language models diffusion techniques, now allowing process, interpretation, generation complex medical images. Despite these achievements, challenges such data scarcity, heterogeneity, ethical concerns remain barriers to clinical translation. Addressing issues through interdisciplinary collaboration will pave way a broader adoption medicine, potentially enhancing patient care optimizing therapeutic outcomes.

Язык: Английский

---

Deep learning-aided respiratory motion compensation in PET/CT: addressing motion induced resolution loss, attenuation correction artifacts and PET-CT misalignment

European Journal of Nuclear Medicine and Molecular Imaging, Год журнала: 2024, Номер 52(1), С. 62 - 73

Опубликована: Авг. 13, 2024

Abstract Purpose Respiratory motion (RM) significantly impacts image quality in thoracoabdominal PET/CT imaging. This study introduces a unified data-driven respiratory correction (uRMC) method, utilizing deep learning neural networks, to solve all the major issues caused by RM, i.e., PET resolution loss, attenuation artifacts, and PET-CT misalignment. Methods In retrospective study, 737 patients underwent [ 18 F]FDG scans using uMI Panorama scanner. Ninety-nine patients, who also had respiration monitoring device (VSM), formed validation set. The remaining data of 638 were used train networks uRMC. uRMC primarily consists three key components: (1) signal extraction, (2) map generation, (3) alignment. SUV metrics calculated within 906 lesions for approaches, (proposed), VSM-based uRMC, OSEM without (NMC). RM magnitude organs estimated. Results enhanced diagnostic capabilities revealing previously undetected lesions, sharpening lesion contours, increasing values, improving Compared NMC, showed increases 10% 17% max mean across lesions. Sub-group analysis significant small medium-sized with Minor differences found between methods, was statistically marginal or insignificant two methods. observed varied amplitudes organs, typically ranging from 10 20 mm. Conclusion A solution has been developed, validated evaluated. To best our knowledge, this is first that compensates blur PET, mismatch artifacts misalignment, misalignment CT.

Язык: Английский

---

ReconU-Net: a direct PET image reconstruction using U-Net architecture with back projection-induced skip connection

Physics in Medicine and Biology, Год журнала: 2024, Номер 69(10), С. 105022 - 105022

Опубликована: Апрель 19, 2024

This study aims to introduce a novel back projection-induced U-Net-shaped architecture, called ReconU-Net, based on the original U-Net architecture for deep learning-based direct positron emission tomography (PET) image reconstruction. Additionally, our objective is visualize behavior of PET reconstruction by comparing proposed ReconU-Net with and existing DeepPET encoder-decoder without skip connections.

Язык: Английский

---

Total-Body Parametric Imaging Using Relative Patlak Plot

Journal of Nuclear Medicine, Год журнала: 2025, Номер unknown, С. jnumed.124.268496 - jnumed.124.268496

Опубликована: Фев. 27, 2025

The standard Patlak plot, a simple yet efficient model, is widely used to describe irreversible tracer kinetics for dynamic PET imaging. Its widespread application whole-body parametric imaging remains constrained because of the need full-time-course input function (e.g., 1 h). In this paper, we demonstrate relative (RP) which eliminates early-time function, total-body and its 20-min clinical scans acquired in list mode. Methods: We conducted theoretic analysis indicate that RP intercept b' equivalent ratio SUV plasma concentration, whereas slope Ki ' equal (net influx rate) multiplied by global scaling factor each subject. One challenge applying short scan duration 20 min) resulting high noise images. applied self-supervised deep-kernel method reduction. Using plot as reference, was evaluated lesion quantification, lesion-to-background contrast, myocardial visualization 22 human subjects (12 healthy 10 cancer patients) who underwent 1-h 18F-FDG scan. also data reconstructed from list-mode either at or 2 h after injection patients. Results: demonstrated it feasible obtain high-quality images using with noise-reduction strategy. highly correlated lesions major organs, demonstrating quantitative potential across subjects. Compared conventional SUVs, significantly improved contrast enabled myocardium cardiac assessment. showed similar benefits. Conclusion: approach, generate

Язык: Английский

---

Robust whole-body PET image denoising using 3D diffusion models: evaluation across various scanners, tracers, and dose levels

European Journal of Nuclear Medicine and Molecular Imaging, Год журнала: 2025, Номер unknown

Опубликована: Фев. 6, 2025

Язык: Английский

---

Deep learning image enhancement algorithms in PET/CT imaging: a phantom and sarcoma patient radiomic evaluation

European Journal of Nuclear Medicine and Molecular Imaging, Год журнала: 2025, Номер unknown

Опубликована: Фев. 27, 2025

PET/CT imaging data contains a wealth of quantitative information that can provide valuable contributions to characterising tumours. A growing body work focuses on the use deep-learning (DL) techniques for denoising PET data. These models are clinically evaluated prior use, however, image assessment provides potential further evaluation. This uses radiomic features compare two manufacturer enhancement algorithms, one which has been commercialised, against 'gold-standard' reconstruction in phantom and sarcoma patient set (N=20). All studies retrospective clinical [ 18 F]FDG dataset were acquired either GE Discovery 690 or 710 scanner with volumes segmented by an experienced nuclear medicine radiologist. The modular heterogeneous used this was filled F]FDG, five repeat acquisitions scanner. DL-enhanced images compared algorithms trained emulate input images. difference between sets tested significance 93 international biomarker standardisation initiative (IBSI) standardised features. Comparing 'gold-standard', 4.0% 9.7% measured significantly different (pcritical < 0.0005) respectively (averaged over DL algorithms). Larger differences observed comparing algorithm 29.8% 43.0% measuring found be similar generated using target method more than 80% not all comparisons across unseen result offers insight into performance demonstrate applications harmonisation radiomics evaluation algorithms.

Язык: Английский

---

Arrival time mapping with 15O-gas PET for cerebrovascular steno-occlusive diseases: a comparative study with CT perfusion

EJNMMI Research, Год журнала: 2025, Номер 15(1)

Опубликована: Март 7, 2025

Positron emission tomography (PET) with 15O-gas for quantifying cerebral blood flow (CBF) and oxygen metabolism is the gold standard assessing hemodynamics in ischemic cerebrovascular disease. However, conventional PET methods do not provide information on regional arrival timing, a hemodynamic parameter typically measured using computed (CT) perfusion contrast media. This study demonstrated that state-of-the-art clinical scanner optimized analysis can generate time maps. In this retrospective of ten patients unilateral stenosis or occlusion major arteries, we compared PET-derived maps CT Tmax short inhalation [15O]-CO2 gases, dynamic images were reconstructed 2-sec temporal resolution, followed by weighted least-squares fitting one-tissue compartment models, without contributions from vascular components. visually comparable to regarding spatial extent delayed brain regions, less noise higher image quality when model Region-of-interest analyses showed good correlations between two modalities: correlation coefficients 0.834 absolute values 0.718 ipsilateral-to-contralateral differences, respectively, indicating quantitatively measure reasonable accuracy. The present method generates arrival-time applying kinetic acquired state-of-the-art, high-sensitivity scanner. Additional parameters CBF may facilitate more comprehensive understanding status steno-occlusive diseases.

Язык: Английский

---

Preliminary exploratory study on differential diagnosis between benign and malignant peripheral lung tumors: based on deep learning networks

Frontiers in Medicine, Год журнала: 2025, Номер 12

Опубликована: Март 27, 2025

Background Deep learning has shown considerable promise in the differential diagnosis of lung lesions. However, majority previous studies have focused primarily on X-ray, computed tomography (CT), and magnetic resonance imaging (MRI), with relatively few investigations exploring predictive value ultrasound imaging. Objective This study aims to develop a deep model based differentiate between benign malignant peripheral tumors. Methods A retrospective analysis was conducted cohort 371 patients who underwent ultrasound-guided percutaneous tumor procedures across two centers. The dataset divided into training set ( n = 296) test 75) an 8:2 ratio for further evaluation. Five distinct models were developed using ResNet152, ResNet101, ResNet50, ResNet34, ResNet18 algorithms. Receiver Operating Characteristic (ROC) curves generated, Area Under Curve (AUC) calculated assess diagnostic performance each model. DeLong’s employed compare differences groups. Results Among five models, one algorithm demonstrated highest performance. It exhibited statistically significant advantages accuracy p &lt; 0.05) compared ResNet34 Specifically, showed superior discriminatory power. Quantitative evaluation through Net Reclassification Improvement (NRI) revealed that NRI values model, when 0.180, 0.240, 0.186, 0.221, respectively. All corresponding -values less than 0.05 comparison), confirming significantly outperformed other four reclassification ability. Moreover, its outcomes led marked improvements risk stratification classification accuracy. Conclusion ResNet18-based distinguishing tumors, providing effective non-invasive tool early detection cancer.

Язык: Английский

---