10. Acoustic Models of Sound Production
415
(on a 1.5-mm cubic grid) or half resolution (on a 3.0-mm cubic grid).
For the GE CT9800 scanner used to collect the D. de/phis data, a linear
mapping between X-ray attenuation values in Hounsfield Units (HU) and
human tissue density has been shown to be accurate to within ±5% for densities ranging from soft tissue to cortical bone (Henson, Ackland, and Fox
1987). The slope of this correspondence varies with the beam energy of the
scanner. Application of more precise medical techniques for tissue HU-todensity mapping could potentially reduce these errors to well below 1%
(Erlichman 1986; Cann 1988; Hunter 1988). The present study assumed that
the scan image pixel values were linearly related to Hounsfield Units, and
mapped tissue densities from HU values using the linear model shown in
Figure 10.2a. Three calibration points confirming this map were provided
by the known density (0.90 g/cc) of the inner melon (Varanasi et al. 1975),
the known density (1.18g/cc) of the plexiglass specimen registration frame,
and the maximum density (roughly 2.7 g/cc) of delphinid periotic bone
(Lees et al. 1996). The image scale ranges from -1,024 to +2,519HU. The
low end of the scale (below -138 HU) was below all soft tissue structures
(except air sacs, which were handled separately). Since it was necessary to
map the air surrounding the scanned specimen to seawater, this lower range
is mapped to the density of seawater (1.03 g/cc) in Figure lO.2a.
Tissue velocity was then deduced from the above tissue density information and the following experimental data. Figure lO.2b plots average
experimental density and velocity values and approximate ranges for
several types of normal (fresh) terrestrial mammalian soft tissues and
delphinid melon lipids at 37°C from literature sources. Terrestrial mammalian tissue data was taken from Parry and Chivers (1979), Goss et al.
(1980), Lele and Sleefe (1985), and Miles (1996). Bottlenose dolphin melon
lipid velocity data was taken primarily from Norris and Harvey (1974) and
secondarily from Litchfield et al. (1979), since the latter study measured the
velocities of lipid samples after extraction from melon and blubber tissues.
Extraction of the lipids from connective tissue components probably
lowered the values reported for the velocities of the "outer" and "under"
melon. Both papers agree approximately on the velocity range for the inner
melon tissues. Melon and lower jaw lipid densities (down to .90glcc) are
taken from Varanasi, Feldman, and Malins (1975) assuming a 60% to 40%
by weight triglyceride-isovalerate wax ester melon lipid composition, which
is similar across the family Delphinidae (Litchfield et al. 1975). Although
few measurements of intact tissue velocity, density, and temperature are
available, an attempt was made to define the ranges for the normal mammalian tissues in Figure lO.2b as I-sigma deviations in the tabulated values.
The density and velocity ranges for delphinid melon lipids shown in
Figure 10.2b correspond to the ranges that exist within the layered melon
tissues.
It is important to note that the velocity of terrestrial mammalian soft
tissues is correlated approximately linearly with density in the range from
415
(on a 1.5-mm cubic grid) or half resolution (on a 3.0-mm cubic grid).
For the GE CT9800 scanner used to collect the D. de/phis data, a linear
mapping between X-ray attenuation values in Hounsfield Units (HU) and
human tissue density has been shown to be accurate to within ±5% for densities ranging from soft tissue to cortical bone (Henson, Ackland, and Fox
1987). The slope of this correspondence varies with the beam energy of the
scanner. Application of more precise medical techniques for tissue HU-todensity mapping could potentially reduce these errors to well below 1%
(Erlichman 1986; Cann 1988; Hunter 1988). The present study assumed that
the scan image pixel values were linearly related to Hounsfield Units, and
mapped tissue densities from HU values using the linear model shown in
Figure 10.2a. Three calibration points confirming this map were provided
by the known density (0.90 g/cc) of the inner melon (Varanasi et al. 1975),
the known density (1.18g/cc) of the plexiglass specimen registration frame,
and the maximum density (roughly 2.7 g/cc) of delphinid periotic bone
(Lees et al. 1996). The image scale ranges from -1,024 to +2,519HU. The
low end of the scale (below -138 HU) was below all soft tissue structures
(except air sacs, which were handled separately). Since it was necessary to
map the air surrounding the scanned specimen to seawater, this lower range
is mapped to the density of seawater (1.03 g/cc) in Figure lO.2a.
Tissue velocity was then deduced from the above tissue density information and the following experimental data. Figure lO.2b plots average
experimental density and velocity values and approximate ranges for
several types of normal (fresh) terrestrial mammalian soft tissues and
delphinid melon lipids at 37°C from literature sources. Terrestrial mammalian tissue data was taken from Parry and Chivers (1979), Goss et al.
(1980), Lele and Sleefe (1985), and Miles (1996). Bottlenose dolphin melon
lipid velocity data was taken primarily from Norris and Harvey (1974) and
secondarily from Litchfield et al. (1979), since the latter study measured the
velocities of lipid samples after extraction from melon and blubber tissues.
Extraction of the lipids from connective tissue components probably
lowered the values reported for the velocities of the "outer" and "under"
melon. Both papers agree approximately on the velocity range for the inner
melon tissues. Melon and lower jaw lipid densities (down to .90glcc) are
taken from Varanasi, Feldman, and Malins (1975) assuming a 60% to 40%
by weight triglyceride-isovalerate wax ester melon lipid composition, which
is similar across the family Delphinidae (Litchfield et al. 1975). Although
few measurements of intact tissue velocity, density, and temperature are
available, an attempt was made to define the ranges for the normal mammalian tissues in Figure lO.2b as I-sigma deviations in the tabulated values.
The density and velocity ranges for delphinid melon lipids shown in
Figure 10.2b correspond to the ranges that exist within the layered melon
tissues.
It is important to note that the velocity of terrestrial mammalian soft
tissues is correlated approximately linearly with density in the range from