Elucidating Early Radiation-Induced Cardiotoxicity Markers in Preclinical Genetic Models Through Advanced Machine Learning and Cardiac MRI

Q: What is the clinical evidence from this study?

Population: Radiation-induced heart disease (RIHD) in a preclinical rat model (n=20). Intervention: Localized whole-heart radiation therapy evaluated with cardiac MRI strain analysis and machine learning algorithms vs. Sham irradiation. Primary outcome: Classification performance of machine learning algorithms in differentiating sham-treated and irradiated rats based on cardiac MRI regional strain and motion parameters (F1 score 0.94, ROC 0.95 (Gradient Boosting); Accuracy 0.88, ROC 1.00 (Random Forest)).

Q: What does this research mean for the field?

Machine learning algorithms applied to cardiac MRI regional strain measurements accurately identify early radiation-induced cardiotoxicity before global ejection fraction declines. Novelty: ClaimNovelty.METHODOLOGICAL. Consensus alignment: ConsensusAlignment.NEUTRAL.

Dayeong An; El‐Sayed H. Ibrahim

doi:10.3390/jimaging10120308

December 1, 2024Journal of ImagingOpen Access

Elucidating Early Radiation-Induced Cardiotoxicity Markers in Preclinical Genetic Models Through Advanced Machine Learning and Cardiac MRI

View Full Paper

Ask AI

Bookmark

Share

Key result

Machine learning on cardiac MRI strain detects radiation cardiotoxicity with ~88% accuracy in rats.

F1 score 0.94, ROC 0.95 (Gradient Boosting); Accuracy 0.88, ROC 1.00 (Random Forest)
n=20

Figures

Why the study?

Radiation therapy carries a risk of radiation-induced heart disease, motivating the detection of early markers using machine learning and cardiac MRI in a preclinical model.

Does machine learning applied to cardiac MRI myocardial strain parameters accurately detect early preclinical radiation-induced heart dysfunction in SS.BN3 consomic rats before declines in global ejection fraction?

Key Points

To detect early markers of radiation-induced heart disease using cardiac MRI and machine learning algorithms in a preclinical rat model.
Treated 10-week-old SS.BN3 consomic rats with localized cardiac radiation therapy or sham exposure.
Assessed global and regional myocardial function using cardiac MRI at 8 and 10 weeks post-treatment.
Trained machine learning algorithms, including gradient boosting and random forest, across multiple feature selection sets to distinguish sham from irradiated hearts.
Left ventricular ejection fraction remained preserved or increased post-radiation (72 ± 1.2% sham vs. 80 ± 1.7% at 8 weeks, p = 0.003, and 76 ± 1.1% at 10 weeks, p = 0.013), alongside increased left ventricular mass (0.36 ± 0.02 g sham vs. 0.46 ± 0.03 g at 8 weeks, p = 0.02, and 0.44 ± 0.02 g at 10 weeks, p = 0.003).
Regional myocardial strain showed statistically significant reductions specifically in the anteroseptal and anterolateral segments of irradiated rats compared to sham controls.
Gradient boosting achieved an F1 score of 0.94 and an ROC of 0.95 using Lasso features, while random forest achieved an accuracy of 88% and an ROC of 1.00.

Structured PICO

Does machine learning applied to cardiac MRI myocardial strain parameters accurately detect early preclinical radiation-induced heart dysfunction in SS.BN3 consomic rats before declines in global ejection fraction?

P

Population

20 adult female SS.BN3 consomic rats aged 10 weeks treated with localized whole-heart radiation therapy or sham and imaged with cardiac MRI at 8 and 10 weeks post-treatment.

I

Intervention

Localized whole-heart radiation therapy (24 Gy in 1 fraction using a 1.5 cm collimator, 225 kVp, 13 mA, 0.32 mm Cu, 2.69 Gy/min via anterior-posterior and two lateral beams 1:1:1) evaluated at 8 and 10 weeks post-treatment using 9.4T cardiac MRI (cine and tagged sequences) coupled with machine learning algorithms (logistic regression, support vector machine, random forest, gradient boosting, and multi-layer perceptron neural networks).

C

Comparator

Sham irradiation at 10 weeks of age, imaged at 8 weeks post-sham on the same 9.4T cardiac MRI protocol.

O

Outcome

Classification performance (accuracy, F1 score, sensitivity, specificity, and area under the ROC curve) of machine learning models in differentiating sham vs. irradiated rats based on regional myocardial strain (circumferential, radial, longitudinal), rotation angle, and tissue motion parameters.surrogate

Main Result

Effect estimate: F1 score 0.94, ROC 0.95 (Gradient Boosting); Accuracy 0.88, ROC 1.00 (Random Forest)

Combining cardiac MRI regional strain imaging with machine learning algorithms enables accurate preclinical detection of radiation-induced cardiotoxicity before reductions in global left ventricular ejection fraction occur.

Limitations

Limited number of studied animals (small sample size of 20 rats)
Potential risk of overfitting in machine learning models due to small sample size, affecting robustness and generalizability
Limited number of study timepoints
Limited number of studied animals (n=20)
Potential risk of overfitting in machine learning models due to small sample size, limiting generalizability
Limited study timepoints and lack of assessment across different heart substructures or additional genetic variations

Implication

ML on cardiac MRI strain detects early radiation cardiotoxicity in rats with preserved LVEF; leaves open human validation.

Abstract

Radiation therapy (RT) is widely used to treat thoracic cancers but carries a risk of radiation-induced heart disease (RIHD). This study aimed to detect early markers of RIHD using machine learning (ML) techniques and cardiac MRI in a rat model. SS.BN3 consomic rats, which have a more subtle RIHD phenotype compared to Dahl salt-sensitive (SS) rats, were treated with localized cardiac RT or sham at 10 weeks of age. Cardiac MRI was performed 8 and 10 weeks post-treatment to assess global and regional cardiac function. ML algorithms were applied to differentiate sham-treated and irradiated rats based on early changes in myocardial function. Despite normal global left ventricular ejection fraction in both groups, strain analysis showed significant reductions in the anteroseptal and anterolateral segments of irradiated rats. Gradient boosting achieved an F1 score of 0.94 and an ROC value of 0.95, while random forest showed an accuracy of 88%. These findings suggest that ML, combined with cardiac MRI, can effectively detect early preclinical changes in RIHD, particularly alterations in regional myocardial contractility, highlighting the potential of these techniques for early detection and monitoring of radiation-induced cardiac dysfunction.

Discussion

No takes yet. Share an insight, caveat, or question.

Cite This Study

An et al. (2024) studied Radiation-induced heart disease (RIHD) in a preclinical rat model (n=20). Localized whole-heart radiation therapy evaluated with cardiac MRI strain analysis and machine learning algorithms vs. Sham irradiation was evaluated on Classification performance of machine learning algorithms in differentiating sham-treated and irradiated rats based on cardiac MRI regional strain and motion parameters (F1 score 0.94, ROC 0.95 (Gradient Boosting); Accuracy 0.88, ROC 1.00 (Random Forest)). Gradient boosting and random forest machine learning algorithms applied to cardiac MRI strain differentiated irradiated from sham rats with an F1 score of 0.94 (ROC 0.95) and 88% accuracy.

synapsesocial.com/papers/6a7af7138d250a4360f443c7 https://doi.org/10.3390/jimaging10120308

View Full Paper

Ask AI

Bookmark

Share

Elucidating Early Radiation-Induced Cardiotoxicity Markers in Preclinical Genetic Models Through Advanced Machine Learning and Cardiac MRI

Figures

Key Points

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Elucidating Early Radiation-Induced Cardiotoxicity Markers in Preclinical Genetic Models Through Advanced Machine Learning and Cardiac MRI

Figures

Key Points

Structured PICO

Main Result

Limitations

Implication

Abstract

Discussion

Cite This Study

Keep reading

Keep reading