Yes this is possible and has been tested by the TAOS (Transneptunian Automated Occultation Survey) group among others. That group is in the process of building the TAOS-II project which will use several telescopes and fast readout CMOS cameras (since the occultations only last a few seconds) to search for occultations caused by unknown Kuiper Belt Objects. Multiple telescopes are very useful to build confidence that you have a real event and not atmospheric fluctuations or other instrumental effect. A detailed workthrough and calculations of how many objects you would see, what is needed to detect them and resolve the Fresnel diffraction and other issues is given in a section by Roques et al. in the Solar System Beyond Neptune book and which is available here.
In general for surveys such as occultations by unknown KBOs, it is etendue (product of telescope collecting area and amount of sky covered in a single exposure) that matters (but it is often more complicated as survey depth doesn't just scale with telescope area) in that higher etendue value equals "better" survey. So for the camera(s) in a survey, it is sensor area that matters, which is a product of pixel size and no. of pixels (and number of sensors and size of gaps between if you are tiling them). The telescope area/diameter tends to be set by how bright your targets of your survey are and practicalities such as cost and manufacturability tend to influence the f/ratio and therefore focal length and pixel scale at your detector. Rarity of your targets and how many you want to find, drive how much area you need to cover. (The telescope area also influences this since in general you get more targets per area if you use a bigger telescope and go fainter). So you end up trading all of these things off to get your best survey for your particular science and type of targets, normally constrained by available or potential budget.
In the case of trying to time-resolve occultations by KBOs, your exposure time is set by the event timescale which depends on the velocity of the KBO across the line of sight and the diameter of the KBO. For typical KBO velocities of 20 km/s, this means that a Pluto (approx 2000km diameter) occultation lasts $2000\,km/20\, kms^{-1}=100\,s$. If you wanted to find small 20 km KBOs similar to 2014 MU69/Ultima Thule, then the events are $20\,km/20\, kms^{-1}=1\,s$ long (both of these calculations assume the occultation goes over the largest part of the KBO and isn't grazing).
Given that you will want several points during the occultation, both to convince yourself it's real and to measure diameters etc, you will need at least 5 points across the event. For the 20km KBOs/1s events, this means at least 5 frames per second, ideally more. This means that you need to get enough signal-to-noise on your target stars in a 0.2 or 0.1s exposure.
As discussed in the reply comments to the question author, scientific CMOS sensors such as the Andor Marana (a commercially available closeish match to the custom detectors used by TAOS-II with ~$31k pricetag; Jan 2019) will be better at both the 'signal' part (due to higher quantum efficiency) and the 'noise' parts (larger, less noisy pixels, cooling and temperature stabilization) than a consumer DSLR sensor. Looking at a study of DSLR dark current by Photonics.com, the dark current is ~50 counts/pixel/second (the Photonics.com figure is for 30s exposures and assuming 1500DN) rising to ~100 counts/pixel/second (averaging at ~3000DN for 3 of the pixels) after 100 frames as the sensor heats. For the Marana, the figure is 0.26 counts/pixel/second (taking datasheet value of 0.2 electrons/pix/s and assuming a gain of 1.3 to match the full well capacity of 85k electrons to the 16 bit ADC (65536 levels)), an almost 400x lower value. (The ADC data range is another area of difference; the Marana can do 16bit digitization whereas DSLRs are typically 12 or 14bit, giving only 4096 or 16,384 levels to record all the signal levels in the image)
The other use of occultations is to monitor predicted occultations by known asteroids and Kuiper Belt Objects. This can be used to get a better idea on the size and shape of the object (as you note for Ultima Thule/2014 MU69), search for additional satellites or rings around the object (as has been done for the KBO Haumea (Nat Geo article) and the Centaur Chariklo (Space.com article)) and look for and monitor changes in the atmosphere (as has been done for Pluto since the first occultation in 1988).
