This study evaluates the use of deep learning (DL) to automatically measure epicardial adipose tissue (EAT) volume and attenuation from low-dose, ungated computed tomography (CT) scans in patients undergoing hybrid myocardial perfusion imaging (MPI). The goal is to assess whether these automated measurements can improve cardiovascular risk prediction. The study included 8781 patients from four sites, with 500 patients used for model training and validation, and the remaining 8281 patients used for testing. DL-based EAT measurements were obtained in less than 2 seconds, compared to 15 minutes for manual annotations. There was excellent agreement between DL and expert annotations for both EAT volume and attenuation across all three sites. During a median follow-up of 2.7 years, 565 patients experienced death or myocardial infarction (MI). Elevated EAT volume and attenuation were independently associated with an increased risk of death or MI after adjusting for confounders. DL-based EAT measurements showed strong correlations with expert annotations and provided additional prognostic insights within the context of hybrid MPI. The study demonstrates that DL can automatically measure EAT volume and attenuation from low-dose, ungated CT with excellent agreement with expert annotations, but in a fraction of the time. These measurements offer independent prognostic information regarding inflammation and metabolism, complementing the functional and anatomic information available through hybrid perfusion imaging. The study also highlights the potential of EAT measurements for therapeutic interventions in patients with elevated EAT volume. However, the study has limitations, including the use of ungated CT scans and the lack of information on race, ethnicity, and inflammatory markers. Overall, the study shows that DL-based EAT measurements can improve cardiovascular risk prediction in patients undergoing MPI.This study evaluates the use of deep learning (DL) to automatically measure epicardial adipose tissue (EAT) volume and attenuation from low-dose, ungated computed tomography (CT) scans in patients undergoing hybrid myocardial perfusion imaging (MPI). The goal is to assess whether these automated measurements can improve cardiovascular risk prediction. The study included 8781 patients from four sites, with 500 patients used for model training and validation, and the remaining 8281 patients used for testing. DL-based EAT measurements were obtained in less than 2 seconds, compared to 15 minutes for manual annotations. There was excellent agreement between DL and expert annotations for both EAT volume and attenuation across all three sites. During a median follow-up of 2.7 years, 565 patients experienced death or myocardial infarction (MI). Elevated EAT volume and attenuation were independently associated with an increased risk of death or MI after adjusting for confounders. DL-based EAT measurements showed strong correlations with expert annotations and provided additional prognostic insights within the context of hybrid MPI. The study demonstrates that DL can automatically measure EAT volume and attenuation from low-dose, ungated CT with excellent agreement with expert annotations, but in a fraction of the time. These measurements offer independent prognostic information regarding inflammation and metabolism, complementing the functional and anatomic information available through hybrid perfusion imaging. The study also highlights the potential of EAT measurements for therapeutic interventions in patients with elevated EAT volume. However, the study has limitations, including the use of ungated CT scans and the lack of information on race, ethnicity, and inflammatory markers. Overall, the study shows that DL-based EAT measurements can improve cardiovascular risk prediction in patients undergoing MPI.