Abdominal inguinal mammary gland chains were excised at necropsy and prepared as whole mounts as described by Thompson et al [8
]. The work followed ethical guidelines approved by the Colorado State University Animal Care and Use Committee. The left chain was fixed in methacarn (24 hrs) and the right chain fixed in 10% neutral buffered formalin (24 hrs). Fixed mammary whole mounts were dehydrated using a series of graded ethanols, cleared in xylene, hydrated using a series of graded ethanols and stained in modified Mayer's carmalum (0.4% carmine in 1.0% aluminum potassium sulfate) for 3 days. Whole mounts were rinsed in tap water to remove excess carmine stain, dehydrated using a series of graded ethanols and cleared in xylene.
Cleared whole mounts were prepared using a modified protocol described by Wellings et al [9
]. Whole mounts were placed in a 4 × 6 inch 4.5 mil thick kapak®
(3 M, St. Paul, MN) heat seal bags, filled with 20 ml of methyl salicylate (Sigma. St Louis, MO) and the bags crimped at the top using a kapak®
(3 M) heat sealer. All alcohols and solvents were used under a chemical fume hood. In addition, a NIOSH approved air-purifying unit (3 M) was worn during dispensing and handling of methyl salicylate to avoid unnecessary exposure.
Air bubbles were removed by placing the bagged whole mount on a flat surface and forcing the air bubbles to the periphery of the whole mount using slight finger pressure. The bagged whole mounts were placed vertically in the heat sealer. Residual air and excess methyl salicylate was displaced to the top by applying slight pressure to the bottom half of the bag containing the whole mount. The bag was crimped in the middle with the whole mount in the bottom of the bag and the residual air and excess methyl salicylate at the top. The bag was cut along the midline to separate the two halves.
Digital images of the mammary gland whole mounts were captured using a semi-automated image acquisition system (North Central Instruments, Plymouth, MN). The components of this system included a 3.0 megapixel CMOS digital camera (Clemex Technologies, Inc. Longueuil, Canada) mounted on a Leica Z16 APO monocular zoom lens 16:1 with a magnification range of 0.57 - 9.2x. The camera and lens were mounted on a Leica Z motor attached to a transmitted light base with a 100 × 100 mm motorized stage (Clemex Technologies, Inc.). An X-Y control box and joystick (Clemex Technologies, Inc.) in conjunction with a Pentium 4 desktop PC (Dell, Round Rock, TX) and Captiva v4.0 software (Clemex Technologies, Inc.) were used for image capture.
Whole mounts were placed on a 6 mm thick sheet of white acrylic plastic (Gagne, Inc. Johnson City, NY) mounted on top of the motorized stage to act as a diffuser. Specimens were trans-illuminated using a 20 V/150 W halogen lamp light source (Volpi, Auburn, NY) with daylight filter mounted at the rear of the base. A series of Z-stack images were automatically captured at 10&3215; magnification using the motorized stage in conjunction with the Captiva 4.0 software (Clemex Technologies, Inc.) and × Y controller. The software seamlessly merged tiled Z-stack images together to form a single uniformly focused composite image based on a best contrast algorithm. Resulting images were saved as TIF files.
plus 4.5 (Mediacybernetics. Silver Spring, MD) image analysis software was used to quantify images according to a modified version of a protocol described by Thompson et al [10
]. A composite image of a stage micrometer was captured under the same conditions as the whole mounts. This image was used to calibrate the area measurements in units of cm2
. A standard optical density curve relating gray level (0-255) to intensity (0 - 2.4) was used to measure ΣOD. An Image Pro macro was written to simplify image analysis. This not only increased the speed of analysis, but also eliminated potential errors.
The macro converted the original color images (Figure ) to 8 bit gray scale. Lymph nodes were selected using the irregular area of interest (AOI) tool wand option (Figure ). A mask of the selected lymph node areas was created and an "or" image operation process was applied to extract the lymph nodes from the gray scale images. Mammary gland epithelium was manually circumscribed using the irregular AOI tool trace option (green outline). Folds, wrinkles, residual muscle and nipples were excluded from the AOI (Figure ). The AOI was removed and the image was "flattened", a digital filtering process that decreases the variation in the intensity of background pixels, which is essential when attempting to properly threshold areas within an image that contain similar intensities, e.g. differentiating mammary epithelia from interductal fat pad within a whole mount (Figure ). High contrast was applied to the flattened image to enhance visualization of dense areas and the AOI was reapplied (Figure ). A 0-254 gray level selection threshold, marked in red was applied to the AOI (Figure ), from which two binary masks were created, one mask containing non-dense elements (interductal fat pad) within the AOI and other inverted mask containing dense elements (mammary epithelia) within the whole mount AOI (Figure )
An "or" image operation was performed using the gray scale image (Figure ) and the inverted dense area mask (Figure ) to extract only the dense areas (mammary epithelia) from the whole mount (Figure ), from which area and volume estimates of the mammary epithelia were obtained (Figure ). A similar image process was applied to the gray scale image (Figure ) and non-dense area mask to extract only non-dense areas (interductal fat pad) from the whole mount (Figure ), which in turn was used to estimate area and volume of the interductal fat pad (Figure ). The circumscribed AOI in the original gray scale image (Figure ) was then used to obtain total area and volume estimates, thus representing combined epithelial and interductal fad pad measurements of the mammary chain (Figure ). All measurements were exported to an Excel spreadsheet (Microsoft. Redmond, WA) via dynamic data exchange. Data were then imported into Systat 13 (Systat Software, Inc., Chicago, IL) for statistical analysis.