Optical imaging of ultrafast dynamics within scattering biological tissue is invaluable to many biomedical fields such as neuroscience, immunology, and cancer biology. Among all imaging modalities available, multiphoton microscopy continues to be the best technology for deep-tissue biological imaging at microscopic resolution. However, the maximum permissible excitation power in tissue fundamentally limits the imaging speed, depth, and field-of-view that can be achieved simultaneously in multiphoton microscopy. This thesis focuses on leveraging recent advances to facilitate ultrafast multiphoton imaging at substantial depths. To this end, we built upon a previously proposed concept, the adaptive excitation source (AES), which mitigates the constraint imposed by the maximum permissible excitation power in biological tissue. By illuminating only the regions of interest in the biological specimen, AES significantly reduces the excitation power without compromising the imaging signal-to-noise ratio (SNR). By integrating AES into custom-built, fast-scanning microscopes, we demonstrated unprecedented performance across multiple in vivo biological imaging experiments. First, we report a simple yet improved implementation of AES in a high-speed microscope based on a polygon scanner. This system enabled us to perform in vivo two-photon (2P) imaging of voltage activity in multiple neurons at high frame rates (500-1650 Hz) across two planes simultaneously at depths of 500-640 µm in awake mouse brain, while still operating below the thermal damage threshold. As of 2025, these results highlight the deepest non-invasive imaging of multiple neurons with the frame rate and SNR necessary to record neuronal voltage activity, as well as the first demonstration of dual-plane voltage imaging in layer 5 (L5) of the mouse brain. Next, with an electrically tunable lens integrated to the previous system, we present a high-speed volumetric 2P imaging method, which is tailored for in vivo recording of cerebral hemodynamics in 3D vascular networks. Together with the advantages of AES, this system enabled us to directly visualize the blood flow in capillary networks deep (> 900 µm) in the mouse brain. This simultaneous combination of imaging speed (100 Hz), depth, and volume (365 x 288 x 50 μm3) has not been achieved by any other existing volumetric imaging technique. Our technology sets the groundwork for future studies of cerebral hemodynamics in Alzheimer's model mice, which will help advance our understanding of brain pathologies. Finally, we extended the concept of AES to imaging moving targets in biological samples. We present adaptive cell tracking (ACT), which combines AES with real-time cell tracking during volumetric imaging of moving objects. We apply ACT to intravital 2P and three-photon (3P) imaging of lymphocyte migration in the mouse lymph node (LN), achieving up to a 10-fold excitation power reduction. The extra photon budget allowed us to track the motile lymphocytes over significantly larger volumes and at substantially greater depths compared with previous reports using 2P/3P imaging methods.