The transition from the second to the third generation of IR algorithms resulted in enhancements across all aspects of image quality. The more advanced IMR algorithm demonstrated superior noise suppression in comparison to iDOSE, suggesting the potential for dose reduction, enhanced spatial resolution, and higher d' scores.Conversely, the third generation IR ADMIRE did not demonstrate equivalent improvements to the second generation IR SAFIRE. The ASIR-V exhibited comparable enhancements to the second generation algorithm. However, it is important to note that these improvements are strongly influenced by contrast and CTDI values, which makes the enhancement of image quality dependent on the specific  task.

In contrast, the deep learning reconstruction algorithm demonstrated a more consistent enhancement, with minimal dependence on acquisition and reconstruction parameters, facilitating its implementation in clinical practice.