Atomic force microscope (AFM) is a tool that allows micro and nano scale imaging of samples ranging from solid state physics to biology. AFM uses mechanical forces to sense the sample and recreate a topography image with high spatial resolution. The biggest disadvantage of the standard AFMs is their scanning speed, as it typically takes up to several tens of minutes to capture an image. A lot of research was conducted to increase AFM scanning speed, which resulted in the development of high-speed AFMs (HS-AFMs), that can obtain an image in matter of seconds. Such increase in scanning speed enabled the study of various processes, ranging from functional mechanisms of proteins to cellular biology dynamics. Increasing the speed further, towards several tens of images per second would highly benefit many applications, from both material and life sciences. The imaging speed of an AFM is limited by the speed of its components. While scanners and electronic systems are constantly being improved, there exists a certain hold-up in the development of cantilevers and deflection sensing techniques. The mechanical bandwidth of the cantilever can be increased by decreasing its size. While it is possible to fabricate sub-micron sized cantilevers it becomes very challenging to sense their deflection. Standard AFMs rely on the optical beam deflection (OBD) readout, which can sense cantilevers down to 2 µm in width. Novel sensing techniques are needed to increase AFM imaging speed further. Strain-sensing techniques are particularly interesting as they offer many advantages over OBD readout, like the ability to sense sub-micron sized cantilevers. We investigated nanogranular tunneling resistors (NTRs) as strain-sensors for cantilever deflection sensing. With NTR ability to be deposited on various substrates and in arbitrary geometries, with lateral dimensions down to tens of nm and having reasonably high gauge factors, they are an interesting candidate for cantilever deflection sensing. We applied NTRs in AFM imaging for the first time, showing that their sensitivity is well suited for imaging of both solid state and biological samples. We also demonstrated that NTRs can be used for sensing of 500 nm wide cantilevers. We performed a study of doped Si piezoresistive strain sensors and of an unexploited potential which can be reached with the miniaturization of the cantilever dimensions. We demonstrated both theoretically and experimentally that by decreasing the size of the piezoresistive cantilevers, one can reach the AFM imaging noise performance equal or better than the noise performance of the OBD readout. We showed that piezoresistive cantilevers are very well suited for nm and Å scale imaging of both solid state and biological samples in air. In addition, we performed a research on an advancement of the AFM feedback controller. Most AFMs use digital signal processor (DSP) based feedback controllers. Digital implementation of the controller has some disadvantages, as it necessitates data converters which introduce additional delays in the feedback loop. We developed a fast digitally controlled analog proportional-integral-derivative (PID) controller. We successfully used this PID controller in AFM imaging, realizing several hundreds of Hz line rates. While the analog implementation of the controller provided large amplification and frequency bandwidth, digital control provided precise control of the system and reproducibility of parameter values.