Crystallite size is the mean size of crystalline domains in a material (crystalline domains are inevitably interrupted by interfaces with other crystalline domains of different orientation or phase boundaries). For very small nanoparticles (e.g. < 5 nm), crystallite size and particle size will generally be the same.
Larger nanoparticles may be single crystals or polycrystalline and composed of multiple crystallites.
Thus, there is no specific relation between nanoparticle size and crystallite size.
As mentionned by Geoffrey the relations between crystalline domain size and nanoparticles size is a very tricky problem allowing misfit. The best way, to my mind, to solve the question is to face TEM tby using Brigth Field (BF), Dark Field (DF) , Selective Area Electron Diffraction (SAED, Lattice Fringes(LF) with Fourrier Transform (TF) modes and tools. BF allows you ti image single or not NP, DF atomic family planes (in fact coherent domains in Bragg conditions) dimension and distribution, SAED of one NP or several but without any information about size, LF + TF atomic plane measurment + d hkl identification. By combining all these data you can get information about the texture and structure of NP. In the case of Quantum Dots these modes associated with X-EDS is a way to precise exact dimension and composition of the QD.