Super Resolution Deep Learning Reconstruction: Benefits of Contrast Reduction in Cardiac Tomography Angiography, Combining the Power of Artificial Intelligence

• DOI: https://doi.org/10.22533/at.ed.1594672409072

• Palavras-chave: CT angiography; coronary arteries; contrast; iodine

• Keywords: CT angiography; coronary arteries; contrast; iodine

• Abstract:

  Nowadays, iodinated contrast represents the most common drug administered to patients. A wide range of radiological examinations benefit from the opacification and enhancement characteristics of iodine. Recently, the lack of MCI was observed worldwide, resulting in a problematic scenario for health services. Due to this scarcity, several scientific works have been published discussing means and ways to reduce and save contrast. In the cardiological setting, coronary artery CT angiography (CTA) represents a well-established diagnostic method in non-invasive coronary assessment and its indications are increasing. Recently, a Super-Resolution Deep Learning reconstruction algorithm (SR-DLR) trained using data acquired on an Ultra High-Resolution Tomography (UHRTC) system was introduced worldwide in the clinical setting. It has the potential to accurately diagnose arteries and stent structures by combining exceptional spatial resolution, noise reduction and increased high contrast resolution, which can benefit the use of smaller volumes of iodine. The objective of this work was to evaluate the cardiac image with the PIQE Deep Learning reconstruction algorithm, using smaller volumes of iodinated contrast. CTA was performed on a 320-channel tomograph with 640 slices. Cardiac acquisition used prospective volumetric acquisition with cardiac synchronization through ECG monitoring, in addition to 100 KVp and non-ionic contrast (370mg/ml), with a flow of 5 mL/s. Results: In cases of ATC performed, a reduction greater than 35% was possible compared to the standard used by the institution. The use of PIQE with smaller MCI volume improved image resolution when compared to hybrid iterative reconstruction. The physical properties present in the PIQE algorithm, such as noise reduction and greater high-contrast resolution, resulted in excellent opacification of iodinated contrast in the evaluated exams.