Imaging Pearls ❯ Cardiac ❯ AI and Cardiac Imaging
|
-- OR -- |
|
- “The 12-lead electrocardiogram (ECG) is an integral component to the diagnosis of a multitude of cardiovascular conditions. It is performed using a complex set of skin surface electrodes, limiting its use outside traditional clinical settings. We developed an artificial intelligence algorithm, trained over 600,000 clinically acquired ECGs, to explore whether fewer leads as input are sufficient to reconstruct a 12-lead ECG. Two limb leads (I and II) and one precordial lead (V3) were required to generate a reconstructed 12-lead ECG highly correlated with the original ECG. An automatic algorithm for detection of ECG features consistent with acute myocardial infarction (MI) performed similarly for original and reconstructed ECGs (AUC = 0.95). When interpreted by cardiologists, reconstructed ECGs achieved an accuracy of 81.4 ± 5.0% in identifying ECG features of ST-segment elevation MI, comparable with the original 12-lead ECGs (accuracy 84.6 ± 4.6%). These results will impact development efforts to innovate ECG acquisition methods with simplified tools in non-specialized settings.”
AI-enhanced reconstruction of the 12-lead electrocardiogram via 3-leads with accurate clinical assessment
Federico Mason, Amitabh C. Pandey, Matteo Gadaleta , Eric J. Topol Evan D. Mus, Giorgio Quer - “With over 300 million being performed worldwide on an annual basis1, the 12-lead electrocardiogram (ECG) has established itself as a bedrock diagnostic in the assessment of cardiovascular disease2–5. Using an array of 10 individual skin-surface electrodes, a series of 12 different electrical signals is arranged to assist in the diagnosis of multiple cardiopulmonary diseases. Despite the advancements provided by vectorcardiography and other recording techniques, including the Mason-Likar system, most clinical diagnoses still rely on the standard 12-lead ECG, whose acquisition process has not iterated to great degrees from its initial inception. It can be cumbersome, requiring special equipment available only at a hospital or clinic, and specially trained individuals to perform and interpret the ECG.”
AI-enhanced reconstruction of the 12-lead electrocardiogram via 3-leads with accurate clinical assessment
Federico Mason, Amitabh C. Pandey, Matteo Gadaleta , Eric J. Topol Evan D. Mus, Giorgio Quer - “Using our reconstruction algorithm, we generated three versions for each of the selected ECGs. The first version was the original 12-lead ECG (Original), the second version was the 12-lead ECG synthetized by our reconstruction algorithm considering two limb leads (I+II) as input, and the third version was the synthetized 12-lead ECG considering limb leads and precordial lead (I+II+V3) as input. For each ECG, each of the three versions (I+II, I+II+V3, and Original)was randomly assigned to one of the three cardiologists, so that each cardiologist was evaluating 238 ECGs, without knowing which of them were original 12-lead ECG, and which were synthetized.”
AI-enhanced reconstruction of the 12-lead electrocardiogram via 3-leads with accurate clinical assessment
Federico Mason, Amitabh C. Pandey, Matteo Gadaleta , Eric J. Topol Evan D. Mus, Giorgio Quer - “Comparing the answers of the cardiologists with the original data labels, we estimated the detection accuracy, sensitivity, and specificity, associated with each input configuration of the reconstruction algorithm (I+II, I+II+V3, and Original). We then proved the non-inferiority of the I+II+V3 system with respect to the original ECG using an unpooled z-test and considering 10% as margin of error. We considered the z-test’s outcome statistically significant if associated with a p-value < 0.05. The p-value is the probability of observing a given event given that the null hypothesis is true; in our work, the null hypothesis is that the accuracy of the I+II+V3 system is >10% lower than that obtained when using the original ECGs as input. Hence, a smaller the p-value corresponds to stronger evidence that I+II+V3 does not lead to relevant performance degradation.”
AI-enhanced reconstruction of the 12-lead electrocardiogram via 3-leads with accurate clinical assessment
Federico Mason, Amitabh C. Pandey, Matteo Gadaleta , Eric J. Topol Evan D. Mus, Giorgio Quer
- “The significant number of imaging markers of CV risk, such as CAC score, plaque features, adipose tissue, and radiomics, are being incorporated to traditional clinical risk factors, in order to create predictive models that can increase the performance of current risk score and prognostic models. AI plays an important role in this process, enabling the consideration of clinical and imaging data together while including a larger number of clinical parameters. Several fusion models have been developed to combine multiple resources, and they have been successfully applied to CVD risk and severity assessment, acute CVD detection, and CVD phenotyping.”
The Role of Artificial Intelligence in Cardiac Imaging
Carlotta Onnis et al.
Radiol Clin N Am 62 (2024) 473–488 - “Even though CCTA mainly provides anatomic information, thanks to CT-FFR, it can provide functional assessment as well. However, it requires complex computer fluid dynamics computations, which are time-consuming and costly. ML has recently been applied to a CT-FFR calculation (FFRML) as opposed to a computational fluid dynamics (FFRCFD) -based approach in order to shorten execution times. As shown by Tesche and colleagues, FFRML required significantly shorter processing time when compared with FFRCFD, while performing equally in detecting ischemia. Moreover, FFRML closely reproduces FFRCFD calculations, assesses the hemodynamic severity of coronary stenosis, correlating with invasive FFR results, and improves diagnostic accuracy and positive-predictive value of CCTA on a per-vessel and per-patient level.”
The Role of Artificial Intelligence in Cardiac Imaging
Carlotta Onnis et al.
Radiol Clin N Am 62 (2024) 473–488 - “CMR, thanks to phase-contrast sequences and the possibility of obtaining specific anatomic planes, has been used to evaluate valves’ anatomy. AI has been applied in this setting to classify and grade valve diseases. Fries and colleagues created a DL model that satisfactorily classified aortic valve malformations from phase-contrast CMR images. They used weak supervision to train a DL model and used it to classify bicuspid aortic valve in unlabeled MR imaging sequences from the UK Biobank. Using health outcome data, they found that the model identified individuals at increased risk of MACE. In addition, ML models have been used to identify different phenotypes of bicuspid valve-associated aortopathy (root, ascending, and arch) and their association with specific clinical findings.”
The Role of Artificial Intelligence in Cardiac Imaging
Carlotta Onnis et al.
Radiol Clin N Am 62 (2024) 473–488 - “The significant number of imaging markers of CV risk, such as CAC score, plaque features, adipose tissue, and radiomics, are being incorporated to traditional clinical risk factors, in order to create predictive models that can increase the performance of current risk score and prognostic models. AI plays an important role in this process, enabling the consideration of clinical and imaging data together while including a larger number of clinical parameters. Several fusion models have been developed to combine multiple resources, and they have been successfully applied to CVD risk and severity assessment, acute CVD detection, and CVD phenotyping.”
The Role of Artificial Intelligence in Cardiac Imaging
Carlotta Onnis et al.
Radiol Clin N Am 62 (2024) 473–488 - “In summary, AI has been successfully used to perform time-consuming tasks, such as segmentation and postprocessing, optimization of data acquisition and reconstruction, and grading of disease severity. AI has been proven to show an improvement, in terms of time and accuracy, of human work, and therefore, serves as an aid to physicians in better understanding of the patient’s cardiac health. Several AI applications in cardiac imaging demonstrate human-level performance, and it is likely that, in the near future, these applications will be further improved to be integrated into clinical workflow; this will have a great impact on costs, wider usability, and optimization of workflow efficiency.”
The Role of Artificial Intelligence in Cardiac Imaging
Carlotta Onnis et al.
Radiol Clin N Am 62 (2024) 473–488
- "KD is associated with mucocutaneous lymph node syndrome and predominantly affects medium and small arteries in infants and children less than 5 years of age. It is more prevalent in Asian populations and has a male dominance.”
Radiologic Imaging in Large and Medium Vessel Vasculitis
Weinrich JM et al.
Radiol Clin N Am 58 (2020) 765–779 - “The coronary arteries are often involved in KD and coronary artery aneurysms develop as a result of coronary vasculitis in about 15% to 25% of untreated patients. Coronary artery aneurysms can be classified according to their size (small, <5 mm; medium, 5–8 mm; and large, >8 mm) and shape (saccular or fusiform). Large coronary artery aneurysms are associated with a higher risk of complications such as rupture, thrombosis, and stenosis, which possibly lead to myocardial infarction and death."
Radiologic Imaging in Large and Medium Vessel Vasculitis
Weinrich JM et al.
Radiol Clin N Am 58 (2020) 765–779
Conclusion: Compared with Coronary Artery Disease Reporting and Data System and other scores, machine learning methods better discriminated patients who subsequently experienced an adverse event from those who did not.- Background: Coronary CT angiography contains prognostic information but the best method to extract these data remains unknown.
Purpose: To use machine learning to develop a model of vessel features to discriminate between patients with and without subsequent death or cardiovascular events. Performance was compared with that of conventional scores.
Conclusion: Compared with Coronary Artery Disease Reporting and Data System and other scores, machine learning methods better discriminated patients who subsequently experienced an adverse event from those who did not.
Scoring of Coronary Artery Disease Characteristics on Coronary CT Angiograms by Using Machine Learning
Johnson KM et al.
Radiology 2019; 00:1–9 - Key Points
* For prediction of all-cause mortality on the basis of coronary CT angiography, the area under the receiver operating characteristic curve (AUC) for a machine learning score was higher than for Coronary Artery Disease Reporting and Data System (CAD- RADS; 0.77 vs 0.72, respectively; P , .001).
* For prediction of coronary artery deaths on the basis of coronary CT angiography, the AUC was higher for a machine learning score than for CAD-RADS (0.85 vs 0.79, respectively; P , .001).
* When deciding whether to start statins, a machine learning score ensures 93% of patients with events will be administered the drug; if CAD-RADS is used instead, only 69% will be treated.
Scoring of Coronary Artery Disease Characteristics on Coronary CT Angiograms by Using Machine Learning
Johnson KM et al.
Radiology 2019; 00:1–9 - “In conclusion, machine learning can improve the use of vessel features to discriminate between patients who will have an event and those who will not.”
Scoring of Coronary Artery Disease Characteristics on Coronary CT Angiograms by Using Machine Learning
Johnson KM et al.
Radiology 2019; 00:1–9