Mechanical Characterization of the Human Liver - Computer Models in Biomechanics - page 363

increase in the stiffness parameter relative to the peripheral parenchyma. For the cur-
rent database, the median increase is 2.4-fold, where the lower significance bound
is located at 1 . 7.
The absolute values of the aspiration parameters and corresponding elastic mod-
uli derived from the aspiration data, 'normal', 'desmoplasia', and 'connective tis-
sue', are in good agreement with literature data and agree with the finding from
elastography (Sandrin et al., 2003 ). It is interesting to note that the diagnostic per-
formance of tissue aspiration and digital palpation were similar (Hollenstein, 2011 ).
Both reached 100 % empirical sensitivity for desmoplastic lesions at specificities
of 63 % and 85 %, respectively. Other cases are not likely to be detected with both
techniques. Observe that aspiration is probably more limited to superficial lesions,
where palpation might as well detect deeper lesions. Also it should be noted that dig-
ital palpation was non-blinded and always conducted by the same surgeon, which
might have biased the outcome.
26.4.4 Mechanical Behavior of Human Livers
The solution of the inverse problem enabled determining model parameters for cap-
sule and parenchyma which are considered as representative of the behavior ob-
served in aspiration experiments on normal liver tissue. From the FE-analysis of
the aspiration experiment and despite the shielding-effect of the capsule, it was
surprising to learn that for the proposed macrostructural two-layer model of the
liver, the constitutive parameters for both, the capsule and the parenchyma, are well-
determined for the implied Rubin-Bodner and 2nd order reduced polynomial form,
respectively. This was quantitatively evaluated in Hollenstein ( 2011 ). The degree of
determination will rapidly decrease for higher-order models, especially regarding
the parenchyma. The inverse FE-analysis based solely on the apex-displacement in
the objective function cannot possess a unique solution: for instance, the optimiza-
tion kernel could drive the elasticity of the parenchyma down to zero and adjust the
capsule properties to capture the measured course of the apex-displacement by itself.
Thus it takes a good initial guess for the parameters in order to find a meaningful
local solution to the optimization problem.
26.5 Conclusions
The present study improved with respect to existing models of liver tissue in that
(i) parenchyma and capsule were separately modeled, (ii) a wide range mechani-
cal characterization of liver capsule was provided, (iii) data were used from intra-
operative in vivo mechanical measurements.
The aspiration technique was applied in an extensive clinical study with over 200
measurements during open abdominal surgery. All procedures related to this exper-
iment were optimized leading to an excellent repeatability of the intra-operative
measurements.