In this paper, the three-dimensional geometry of the human cochlea is model
ed by the helico-spiral seashell model. The 3-D helico-spiral model, the ge
neralized representation of the Archimedian spiral model, provides a framew
ork for measuring cochlear features based on consistent estimation of model
parameters. Nonlinear least square minimization based algorithms are devel
oped for the identification of rotation, center and intrinsic parameters of
the helico-spiral representation. Two algorithms are designed for the rota
tion axis aligned to the modiolar axis: one is more susceptible in the pres
ence of noise, while the other allows applicability to two-dimensional data
sets. The estimated center and intrinsic parameters allow the calculation
of length, height and angular positions needed for frequency mapping of mul
tichannel cochlear implant electrodes. Model performance is evaluated with
numerically synthesized curves with different levels of added random noise,
histologic data and real human cochlear spiral computed tomography data.