A detailed theoretical and empirical investigation of additive noise for in
direct detection, active matrix hat-panel imagers (AMFPIs) has been perform
ed. Such imagers comprise a pixelated array, incorporating photodiodes and
thin-film transistors (TFTs), and an associated electronic acquisition syst
em. A theoretical model of additive noise, defined as the noise of an imagi
ng system in the absence of radiation, has been developed. This model is ba
sed upon an equivalent-noise-circuit representation of an AMFPI. The model
contains a number of uncorrelated noise components which have been designat
ed as pixel noise, data line thermal noise, externally coupled noise, pream
plifier noise and digitization noise. Pixel noise is further divided into t
he following components: TFT thermal noise, shot and 1/f noise associated w
ith the TFT and photodiode leakage currents, and TFT transient noise. Measu
rements of various additive noise components were carried out on a prototyp
e imaging system based on a 508 mu m pitch, 26x26 cm(2) array. Other measur
ements were performed in the absence of the array, involving discrete compo
nents connected to the preamplifier input. Overall, model predictions of to
tal additive noise as well as of pixel, preamplifier, and data line thermal
noise components were in agreement with results of their measured counterp
arts. For the imaging system examined, the model predicts that pixel noise
is dominated by shot and 1/f noise components of the photodiode and TFT at
frame times above similar to 1 s. As frame time decreases, pixel noise is i
ncreasingly dominated by TFT thermal noise. Under these conditions, the rea
sonable degree of agreement observed between measurements and model predict
ions provides strong evidence that the role of TFT thermal noise has been p
roperly incorporated :into the model. Finally, the role of the resistance a
nd capacitance of array data lines in the model was investigated using disc
rete component circuits at the preamplifier input. Measurements of preampli
fier noise and data line thermal noise components as a function of input ca
pacitance and resistance were found to be in reasonable agreement with mode
l predictions. (C) 2000 American Association of Physicists in Medicine. [S0
094-2405(00)01408-5].