Medical imaging plays a key role in the diagnosis of disease, with Magnetic Resonance Imaging (MRI) providing non-invasive and non-ionizing visualization of human anatomy and function. In cardiovascular medicine, the measurement of volumetric flow over time, also known as 4D flow MRI, has become particularly important as it allows the derivation of hemodynamic parameters associated with various cardiovascular diseases. However, clinical applicability is often limited by noise, resolution, and long scan times. Machine learning has shown promising results in overcoming these limitations by reducing artifacts and improving image resolution beyond clinical practice. Our team focuses on the development of spatiotemporal super-resolution and denoising methods for 4D flow MRI data. We are investigating state-of-the-art deep learning approaches to handle computationally demanding three-directional volumetric flow data. Unlike previous methods that address spatial and temporal resolution separately, we aim to increase both simultaneously using convolutional neural networks that maintain spatiotemporal coherence. While this approach has shown promising results, it requires significant computational resources, which we aim to optimize by proposing and comparing different neural networks.