One of the major disadvantages of Nuclear Magnetic Resonance (NMR) is its low sensitivity, mainly due to a very low spin polarization. Since 2003, Dissolution Dynamic Nuclear Polarization (D-DNP) provides a way of overcoming this drawback in solution by increasing polarization by factors exceeding 10'000-fold. In a D-DNP experiment, the high electron spin polarization at low temperatures is transferred to other nuclei (like 1H and 13C) by microwave irradiation, and the frozen sample is subsequently dissolved to room temperature while preserving part of the polarization. This thesis presents novel conceptual and instrumental developments in D-DNP that enable a better hyperpolarization of several nuclear spins. More specifically, after a brief introduction to NMR and D-DNP in Part A, we will discuss in Part B new strategies for improving microwave irradiation and implementing Longitudinal Detection Electron Spectroscopy Resonance (LODESR) in situ, c.a. at low temperature and high magnetic field. We will see how Cross Polarization can be made fully compatible with D-DNP with an original double resonance circuit developed in our laboratory (1H and another nucleus like 13C, 15N, 29Si or 129Xe). Finally, we will show how the problem of low field transfer of hyperpolarized solutions can be overcome by the use of a novel concept of a magnetic tunnel that sustains the magnetic field and prevent excessive losses of our hyperpolarized "ambitions". Finally, new applications in D-DNP that were either enabled or better realized by these new strategies are briefly presented in Chapter C.