Open Journal of Medical Imaging, 2013, 3, 144-155
Published Online December 2013 (
Open Access OJMI
Segmentation of Sinusoids in Hematoxylin and Eosin
Stained Liver Specimens Using an
Orientation-Selective Filter
Masahiro Ishikawa1,2, Sercan Taha Ahi1, Fumikazu Kimura1, Masahiro Yamaguchi1,
Hiroshi Nagahashi1, Akinori Hashiguchi3, Michiie Sakamoto3
1Global Scientific Information and Computing Center, Tokyo Institute of Technology, Meguroku, Japan
2Faculty of Health & Medical Care, Saitama Medical University, Hidaka-shi, Japan
3Department of Pathology, Graduate School of Medicine, Keio University, Shinjuku-ku, Japan
Received November 10, 2013; revised December 10, 2013; accepted December 17, 2013
Copyright © 2013 Masahiro Ishikawa et al. This is an open access article distributed under the Creative Commons Attribution Li-
cense, which permits unrestricted use, distribution, and reproduction in any medium, provided the original work is properly cited.
The liver comprises cell layers of hepatocytes called trabeculae, which are separated by vascular sinusoids. Under-
standing the structure of hepatic trabeculae and liver sinusoids in hematoxylin and eosin (HE)-stained liver specimens is
important for the differential diagnosis of liver diseases. In this study, we develop an approach to extracting liver sinu-
soids from HE-stained images. The proposed approach involves: 1) a new orientation-selective filter (OS filter) for edge
enhancement and image denoising, 2) the clustering of image pixels to identify candidate sinusoids, and 3) a classifica-
tion procedure that discards unlikely candidates and selects the final sinusoid areas. Experimental studies using a data-
base of 16 images with a resolution of 512 × 512 pixels showed that the proposed approach could segment liver sinu-
soid pixels with 81% of specificity and 94% of sensitivity. A comparison with a method based on bilateral filters
showed that this method improved the sensitivity for all images with an average improvement of 4% and no difference
in specificity. The results were presented to a group of pathologists and they confirmed that the images were highly
representative of the tissue morphology features.
Keywords: Histopathological Tissue Images; Orientation-Selective Filter; Segmentation of Sinusoids
1. Introduction
During conventional histopathology, samples of abnor-
mal tissues are removed from body, placed in a fixative
to prevent decay, and stained with dyes to highlight cer-
tain features of interest. Later, the highlighted features
are inspected by pathologists using microscopes. The
goal of visual inspection is to identify the manifestations
of disease, such as cell atypia and malignant neoplasm.
The results of these subjective inspections are used to
identify appropriate therapeutic procedures.
Recent advances in whole-slide imaging technology
mean that high resolution images of tissues can be ac-
quired using specialized hardware and these images are
used at conferences, for educational purposes, and in
telepathology. Computer-aided diagnosis is also expected
to make a major contribution to the quantification of
digital images.
In recent years, there have been many reports of patho-
logical image analysis methods. Many of these previous
studies were focused on the automatic grading of gastric
cancer [1] and prostate cancer [2-4]. In the area of
pathological image segmentation, a cell nucleus extrac-
tion competition is held each year at the International
Conference on Pattern Recognition (ICPR) [5]. However,
there have been few reports of the automatic grading of
hepatocellular carcinoma (HCC). In [6] and [7], auto-
matic grading was conducted according to the Edmond-
son classification, where results were reported with good
precision based on characteristic parameters related to
the cell nucleus. The histopathological features of cell
nuclei have been used in most conventional image analy-
sis methods. However, other characteristics are also im-
portant for the diagnosis of HCC, including the morpho-
logical features of tissues, such as sinusoids and hepatic
trabeculae; cellular features such as the cytoplasm and
nuclei; as well as lymphocytes and red blood cells [8].
Thus, the more accurate evaluation of hepatic lesions
demands the extraction and quantification of structures
such as trabeculae and fibrous cells. In the present study,
we developed a method for extracting sinusoids, which is
a basic approach to structure recognition.
Hepatic cells are normally arranged in a radial pattern
that originates from a central vein. This radial structure is
called a hepatic trabecula. Any irregularities in the he-
patic trabeculae are important for the histological classi-
fication of HCC. Trabeculae are normally made of a sin-
gle row of cells, but sometimes they exhibit increased
thickness and the formation of multiple rows of cells
during the development of structural atypia in carcinoma.
The specific mixture of thick and thin hepatic trabeculae
indicates the stage of progression in cancer.
In the present study, we attempted to extract sinusoidal
regions. Sinusoids run parallel to the trabecula and they
provide nutrition to cells. Therefore, analyzing the struc-
ture of sinusoids is crucial for the extraction of morpho-
logical features from hepatic trabeculae. The direct ex-
traction of morphological features from hepatic trabecu-
lae is a very difficult task because the trabecula structure
includes various objects such as the cell nucleus, cyto-
plasm, and cell membrane.
2. Liver Sinusoids
As shown in Figure 1, liver tissues mainly comprise a
hepatic lobule and Glisson’s sheath. The lobule contains
hepatic cells, sinusoids, and blood capillaries. Most of
the hepatic cells are arranged in trabeculae and the blood
capillaries run between the cells. The sinusoidal wall
between the sinusoids and hepatic cells comprises endo-
thelial cells. The boundary region is relatively distinct,
but substances are exchanged between hepatic cells, as
Figure 1. Seven major structures in a hematoxylin and
eosin-stained hepatic histological specimen: (1) interhepatic
bile duct, (2) hepatic artery, (3) hepatic portal vein, (4) fiber,
(5) nucleus, (6) sinusoid, and (7) hepatic trabecula.
well as sinusoidal blood through the holes or gaps be-
tween endothelial cells. The sinusoids have a lumen
structure and hematoxylin and eosin (HE)-stained sinu-
soid specimens are only weakly colored. However, it
does not form a full cavity because it contains secretions,
blood, and endothelial cells. In some cases, the sinusoids
are squeezed and their lumens are invisible in the corre-
sponding pathological specimen. In particular, the lumen
is often obscure in moderately or poorly differentiated
lesions. In the present study, we extracted sinusoids with
a lumen structure [9].
3. Methodology
In general, sinusoids appear white in color because they
are not stained by hematoxylin or eosin, but they still
include endothelial cells, red blood cells, bodily secre-
tions, and other substances. The boundary between the
sinusoid and the hepatic cytoplasm is sometimes obscure
and a newly developed orientation-selective (OS) filter is
used as a pretreatment to highlight the structures in the
sinusoid and the cytoplasm. The flowchart shown in
Figure 2 illustrates our methodology. After enhancing
the margins of sinusoids using the OS filter, the candi-
date sinusoidal regions are extracted by the expectation-
maximization algorithm (EM algorithm). Next, we cal-
culate the histogram features of the candidate regions and
classify them using a previously trained linear support
vector machine (SVM).
3.1. Orientation-Selective Filter
The sinusoid wall cells, such as the endothelial cells on
the boundary between the sinusoid and the hepatic cells,
appear to be arranged in smooth curves but they have
interruptions and voids, as described above. Therefore, if
a sinusoid region is divided along this boundary during
image processing, it will be difficult to determine the
correct boundary between the sinusoid and the hepatic
cell in the interrupted section. Thus, an OS filter is used
to connect the interrupted section to the interrupted
boundary in a smooth manner, thereby defining the bor-
derline. Another possible method for image segmentation
is a bilateral filter [10], which is a powerful method for
smoothing while preserving edges. However, it has no
effect on connecting broken boundaries. A suitable filter
should smooth the boundary in the direction of the boun-
dary, while retaining the edges in the direction perpen-
dicular to the boundary. Thus, an OS filter was designed
specifically for this purpose because no other filters have
this functionality.
The proposed OS filter is a selective version of a bar
filter [11] that performs an affine transformation in the
orientation of the brightness gradient. If an original im-
age is assumed to be I(x,y), the brightness gradient ori-
Open Access OJMI
Open Access OJMI
Figure 2. Flowchart of the proposed method: Section 3.1 is the pre-processing stage, Section 4.1 is the cluster ing process, and
Section 4.4 is the discarding process.
entation of each pixel is determined using Equation (1):
 
cossin 0
cijbijh ij
,tan ,
1214, 8
bxy  
 
 
 
xxyIxy Ixy
fyxyI xyI xy
mxyfxxy fyxy
In the present study, a filter was used with Gaussian
smoothing of 11 × 1 lines, where NS is the image size.
The length of the filter is determined by the actual gaps
being connected. In the present study, the filter was de-
signed so it was two pixels shorter on either side com-
pared to the size used to calculate the intensity gradient.
If the filter is too large, intensity information other than
that related to the sinusoid boundary would be included
in the calculation. If there are two or more gradient ori-
entations, these gradient orientations are calculated and
integrated. The convolution-like operation in Equation (3)
is performed using all of the RGB color channels.
where θ(x,y) is the gradient orientation and m(x,y) is the
gradient magnitude. The pixel values in the green chan-
nel of the RGB color space are used because the differ-
ence between the sinusoids and cytoplasm is relatively
high [12]. Computing the orientation of a single pixel
using only its four-connected neighborhood might be
unstable, so we compute the pixel-wise orientation val-
ues in three steps. First, we calculate the gradient of each
pixel using centered 1D point derivatives with [1, 0, +1]
masks. Next, we quantize the gradient magnitudes into
18 equally spaced values between 0 and 170. Finally, we
calculate the weighted gradient orientation histogram
around each pixel in a neighborhood of 15 × 15 pixels.
The size of the window is determined by the size of the
object for which the intensity gradient is obtained. In the
present study, this was slightly larger than the thickness
of the sinusoid. The weights are the gradient magnitudes.
The largest value in the weighted histogram is assigned
to the center pixel. If multiple peaks in the histogram
exceed 80% of the maximum, they are all assumed to be
valid. This protects the structures of the edge crossings.
This threshold is a parameter and when it is high, it gives
greater priority to local edge enhancement, but more
protection to edge crossings when it is low. Next, a bar
filter is affine-transformed in the gradient orientation of
each pixel and applied using Equation (3). This filter is
similar to convolution, but the kernel depends on the
weighted gradient orientation (θ) histogram described
3.2. Effectiveness of the Orientation-Selective
The filter used in the present study changes its convolu-
tion kernel using spatial information, so the effects are
also regarded as varying with the image properties. Thus,
we examined whether the filter properties were suitable
for sinusoid extraction from the hepatic histopathological
tissue images considered in this study. First, the results of
filtering are shown in Figure 3. Figures 3(a) and (b)
show the original image and the results obtained by con-
volution with the OS filter, respectively. Figure 4(b)
shows the variation in a sinusoid where the gray level
was obscure or a segment where favorable results were
obtained compared with the original image (Figure 4(a)).
Figure 4(c) shows the results obtained with the bilateral
filter compared with the OS filter, where the cytoplasm
around the sinusoid is clear, although the boundary be-
tween the sinusoid and the cytoplasm is obscure. In the
 
ijIij Fcij
Figure 3. Original and filtered images. (a) Original image;
(b) Orientation-selective filtered image.
Figure 4. Original and filtered images. (a) Original; (b) Ori-
entation-selective filtered image; (c) Bilateral filtered image.
results obtained after the application of the OS filter
(Figure 4(b)), the boundary with the clear cytoplasm
exhibits deeper pink eosin staining compared with the
surrounding cytoplasm while the boundary between the
sinusoid and the hepatic cell is more distinct because the
sinusoid’s interior secretion is whiter as a consequence of
smoothing. In addition, information related to the varia-
tions in concentration from the sinusoid to the cytoplasm
is conserved.
4. Extraction of Candidate Sinusoidal
HE-stained specimens mainly contain cell nuclei that
appear blue due to hematoxylin, cytoplasm that appears
red due to eosin, and sinusoids that appear white due to a
lack of chemical reaction. We assumed that the signal
intensities of the nuclei, cytoplasm, and sinusoids had
Gaussian distributions in the RGB color space and we
used the EM algorithm to estimate the corresponding
class means and variances [13].
4.1. Clustering with the EM Algorithm
The EM algorithm is a well-established method that is
often used often to estimate the parameters for mixed
distribution models. In the present study, a mixed normal
distribution was applied in the RGB color space to esti-
mate the white region. An RGB feature vector is defined
as, where N is the number of pixels in the image. The
Gaussian mixture distribution can be defined using Equa-
tion (4):
kkk kk
0, 1
exp 2
kk k
 
 
where k
is the average, k
is the covariance matrix,
k is the distribution weight, and F is the number of
classes. In this case, the maximum likelihood estimate is
represented by Equation (5).
 
log ,logmax
,, ;,0,
kkk kkk
 
 
 
Open Access OJMI
The estimation process involves an expectation step
and a maximization step. The expectation step is repre-
sented by Equation (6) using Bayes’ rule, where i is the
pixel number and k is the class number of each distribu-
kki kk
The maximization step is represented using Equation
ik kk
In a test conducted as part of this study, an initial value
was determined using the k-means method. Sinusoids are
basically white regions so they should be extracted if an
appropriate threshold value can be determined. HE-
stained images have color densities that vary among im-
ages, however, which means that it is necessary to de-
termine the threshold as a relative value. In the present
study, the EM algorithm was used to determine the rela-
tive threshold with good precision.
4.2. Eliminating Cell Nuclei from Sinusoids
Endothelial cell nuclei and blood cells are present within
the extracted sinusoids. Thus, it is necessary to eliminate
them before determining the structures of sinusoids. The
cell nuclei are often isolated in the interiors of the sinu-
soids because they do not include cytoplasm. Thus, it is
necessary to eliminate all of the isolated areas within the
sinusoidal regions. The interior sinusoids are labeled and
small regions less than or equal to a given size are treated
as sinusoids.
4.3. Discarding Unlikely Sinusoid Candidates
Images of hepatic tissue contain various white structures
that resemble sinusoids. Examples of these structures are
cells undergoing ballooning degeneration, clear cells, and
fat droplets. Ballooning degeneration is a form of cell
death, which is characterized by a thin cell cytoplasm.
Clear cells indicate a specific type of carcinoma in the
liver and these cells lack most of their cytoplasmic con-
tent, thereby leaving nuclei surrounded by large white
areas. Fat droplets indicate liver disease and they can be
observed as round white areas in HE-stained specimens.
Although the causes and underlying mechanisms of
these structures are different, their appearance is quite
similar to sinusoids. Therefore, a procedure that labels all
of the white regions as sinusoids may produce erroneous
results. We propose a two-step approach to correct these
errors and to improve the specificity of the sinusoid seg-
mentation procedure. After the first step described above,
the second step involves the classification of sinusoids
and sinusoid-like structures using a supervised learning
approach, which is explained in the following subsec-
4.4. Extracting Gray-Level Histogram Features
The contrast between sinusoids and tissue cytoplasm is
apparent in the middle regions of the visible spectrum
because the absorption spectrum of eosin has a peak in
that region. Therefore, given an 8-bit RGB image I_rgb,
we initially select the green channel of the image and
extract four different features from the gray-level histo-
gram, i.e., the mean, variance, skewness, and kurtosis.
The contrast between the sinusoid and cytoplasm needs
to be enhanced to extract the sinusoids. HE staining dyes
the cytoplasm with eosin and the nucleus with hematoxy-
lin. A sinusoid contains few cellular tissues and its prop-
erties are similar to glass areas, which generate the re-
sults for hematoxylin absorption spectrum shown in Fig-
ure 5. Figure 5 shows the absorbance levels of the he-
matoxylin and eosin simple stains, respectively. There
are major differences between the absorption spectra of
the sinusoid (glass) and eosin (cytoplasm) at 475 - 580
nm. Thus, using the green channel helps to distinguish
eosin sufficiently well to obtain dark areas, which en-
hances the contrast compared with the sinusoid. This is
why the green channel is used in this method. If we as-
sume that the set represents the pixel values of the Green
channel between 0 and 255 nm, ni represents the number
of pixels in the gray level i, and N represents the total
number of pixels in the image, then the probability of a
pixel occurring at level i is as follows (Equation (8)).
pI ii
 
Given p(i), the mean (β),variance (v2), skewness (ι),
and kurtosis(ϖ) of the histogram can be computed as fol-
ip i
255 2
Open Access OJMI
Open Access OJMI
Figure 5. Absorbance of HE-stained samples.
255 4
As shown in Figure 6, ballooning regions exhibit a
smooth variation in the pink color between the interior
and exterior regions. During experiments, we dilated the
initial sinusoid candidates five times using a 3 × 3 struc-
turing element and we extracted separate features for
pixels in the original region and pixels in the expanded
region. Figure 6(b) shows the gray-level histogram fea-
tures derived from both regions inside the red and green
lines. Therefore, the number of features was eight. A
linear SVM was used to reduce the number of candidate
sinusoid regions based on the gray-level histogram fea-
tures [14,15]. The classifier was trained using 300 sinu-
soids and 300 ballooning regions, each of which was
extracted from another set of HE-stained liver images
obtained at a magnification of 20×.
(a) (b)
Figure 6. Difference between a sinusoid and ballooning. (a)
Original image; (b) Sinusoid and ballooning.
soid and hepatic cell regions are indicated as white and
black, respectively. The results were shown to a group of
pathologists and the expert assessments indicated that the
images obtained were good representations of the tissue
morphology features in benign Edmondson grade 1 (well
differentiated) and grade 2 (moderately differentiated)
tissues. However, non-sinusoidal regions were still ex-
tracted, such as fatty cells and portal veins. In future
work, we will aim to eliminate these errors. The eva-
luation was not performed for grades 3 and 4 because the
appearances of the sinusoids are completely different in
these grades.
5. Experimental
5.1. Materials
5.3. Evaluation of the Extraction Accuracy
Whole-slide images of surgical liver specimens were
captured using a 20 microscope objective (Nano Zoomer
2.0; pixel width = 460 nm) and 967 different regions-of-
interest (ROIs) with dimensions of 1 × 1 mm were
selected by a pathologist for further analysis. A patholo-
gist classified the images according to Edmondson grad-
ing [16]. Of the 967 ROIs, 551 were labeled as back-
ground, 80 as grade 1, 247 as grade 2, 64 as grade 3, and
25 as grade 4.
The sinusoidal regions extracted using the proposed app-
roach were compared with the manually extracted sinu-
soids to determine the accuracy of the automated ex-
traction method.
5.3.1. Test Images Used in the Experiment
To compare the accuracy of the extraction results, the
manually extracted sinusoidal regions were used in this
study. Sixteen representative images were selected to
assess their staining and ballooning features. In HE-
stained images, the densities of hematoxylin and eosin
vary according to the state of fixation and staining. The
5.2. Experimental Results
Figure 7 shows some of the results obtained from the
automatically segmented sinusoid images, where the sinu-
Figure 7. Results obtained using the propose d method. (a) Background; (b) Edmondson grade 1; (c) Edmondson gr ade 2.
cytoplasm also has a clear appearance in chronic liver
disease cells because of glycogen. The manually ex-
tracted results are shown in Figure 8 and the results ob-
tained using the proposed method are shown in Figure 9.
5.3.2. C omparison of the Metho d s
We used three metrics to evaluate the pixel-wise segmen-
tation accuracy, i.e., the sensitivity, specificity, and over-
lap, which were determined based on the relationships
shown in Table 1. The sensitivity was calculated as
TP/(TP + FN), the specificity as TN/(TN + FP), and the
overlap as TP/(TP + FP + FN), where TP is the number
of positive cases of sinusoids that were identified correctly,
TN is the number of negative cases of sinusoids that
were identified correctly, FP is the number of positive
cases of sinusoids that were classified incorrectly, and
FN is the number of negative cases of sinusoids that were
classified incorrectly.
5.3.3. Experimental Results
The evaluation results are shown in Table 2. Using the
proposed approach, the average sensitivity was 81% and
the specificity was 94% with the 16 test images. To
verify the effectiveness of the filtering and candidate
reduction procedures, we compared the use of these
treatments and direct clustering with the original images.
Our results showed that filtering increased the sensitivity
by 5% and the specificity by 2%. In particular, we ob-
served the effective acquisition of continuous, smooth
results on the boundary of the sinusoid (Figure 10(d)).
Figure 10(c) shows the results obtained by extraction
using the bilateral filter compared with the proposal me-
thod. Figure 10(a) shows the original image and Figure
10(b) shows the results extracted from the original image.
The effects of feature-based candidate reduction on the
sensitivity and specificity were as follows. The sensiti-
vity did not decline because the incorrect deletion of re-
gions did not occur in the experiment, even after narrow-
ing. The specificity was 2% higher for the results that
included incorrectly extracted regions. There was 3%
higher specificity compared with that before narrowing.
Thus, narrowing based on the gray-level histogram fea-
tures provided favorable results, because it increased the
specificity by 3% without decreasing the sensitivity.
5.4. Comparison with Other Methods
Image processing using bilateral filters is widely re-
cognized as a data smoothing technique that considers
edges. Bilateral filters are used for preprocessing during
pathological image segmentation [4]. Thus, we compared
the proposed method with a bilateral filter. The results
obtained with the bilateral filter were based on the ex-
traction of sinusoids with the EM algorithm, which re-
placed the OS filter during preprocessing. The parame-
ters of the bilateral filter were standard deviation values
of σ1 = 3 for the geometric spread, σ2 = 0.1 for the
photometric spread, and the window size = 10 pixels,
which supported the identification of edges that were
sufficiently solid to maintain useful images. Table 3
shows the evaluation results based on the area ratios. The
ground truth data were the results obtained by manual
extraction, which were the same as those used in the
experimental precision evaluation. Seven images with
good color conditions were used in the experiment where
the image size was 2174 × 2174 pixels. Furthermore, the
area ratios were calculated by weighting the vicinity of
the edges because there were differences among indivi-
duals in the manual extraction results. Thus, the vicinity
Open Access OJMI
(a) (b) (c) (d)
(e) (f) (g) (h)
(i) (j) (k) (l)
(m) (n) (o) (p)
Figure 8. Sixteen different HE-stained liver specimens, which were captured using a 20× objective lens. The sinusoids in the
images were segmented manually and the segmented images were used to evaluate the proposed algorithm. The percentages
represent the ratio of the sinusoid area relative to the image area. The ratios in the images varied between 8% and 37%. (a)
19%; (b) 8%; (c) 24%; (d) 16%; (e) 17%; (f) 8%; (g) 10%; (h) 30%; (i) 37%; (j) 25%; (k) 27%; (l) 22%; (m) 24%; (n) 32%;
(o) 37%; (p) 27%.
edges were counted as 0.5 pixels. The voids in the fi-
brous regions were not considered during this proce-
dure, so fibrous regions were excluded from the calcula-
tions. The results showed that the proposed method im-
proved the sensitivity for all images with an average im-
provement of 4% while there was no major difference in
the specificity. There was a 1% decrease in the specific-
ity with the proposed method for some images compared
with the bilateral filter. We also performed an experiment
where a SVM was used to eliminate unlikely sinusoid
candidates. The results showed that there was a 1% - 3%
improvement in the specificity with images that had
rather clear photographic conditions compared with the
the OS filter and using a SVM to eliminate unlikely can-
didates. The green pixels indicate negative results that
were classified correctly by the SVM, while the red pix-
els indicate positive results that were classified incor-
rectly as negative. The proposed method improved the
average overlapping by 2%.
6. Conclusions
original images. Figure 11 shows images obtained with
oids were extracted to facilitate the In this study, sinus
structural analysis of hepatic histopathological tissue
images. Our results showed that the area ratio was in-
creased by 6% using the proposed pretreatment, i.e., the
OS filter, and the boundary of the sinusoid was extracted
in a natural manner. Furthermore, narrowing was con-
Open Access OJMI
(a) (b) (c) (d)
(e) (f) (g) (h)
(i) (j) (k) (l)
(m) (n) (o) (p)
Figure 9. Sixteen differens.he results were
extracted using the propo t HE-stainever specimens, where captured usin20× objective len
sed method.
d liich wg a T
(a) (b)
(c) (d)
Figure 10. Comparison of the original image iginal image; (b) Results extracted from the
original image; (c) Result after extraction from) Result after extraction from the OS-filtered
with the filtages. (a) Or
the bilateral-filtered image; (d
ered im
Open Access OJMI
Table 1. Confusion matrix.
Ground truth
Positive Negativ
Algorithm Positive T TP + FN P FN
Output Negative FP
Total + F + TNN
Tavaf theenef thed
sinusoid segmentation apsin imagThe
ecificity and sensitivity values were computed by com-
+ Discarding
ble 2. Eluation o effectivss o propose
proach ug 16 testes.
paring the pixels in the segmentation outputs with the pixels
in the ground truth data shown in Figure 8. The segmen-
tation outputs were obtained using three different app-
roaches: 1) dividing unfiltered images into three clusters; 2)
dividing filtered images into three clusters; 3) dividing fil-
tered image into three clusters and discarding unlikely can-
didates, which were predicted using a support vector ma-
chine. The values highlighted with a dark gray background
have the highest sensitivities, the values highlighted with a
light gray background have the highest speci ficities, and the
values highlighted with a mid-light gray background have
the highest overlaps.
1) Clustering 2) Filtering
+ Clustering
3) Filtering
+ Clustering
Sen Spe Ove Sen Spe SenveOve s SpeO
A 0.73 0.86 0.46 0.778 0.83 0.45 0.7 0.84 0.46
B 0.75 0.85 0.26 0.78 0.83 0.25 0.78 0.910.37
C 0.61 0.93
0.50 0.64 0.91 0.50 0.57 0.94 0.47
D 0.65 0.94 0.49
0.66 0.96 0.53 0.63 0.97 0.54
E 0.84 0.77 0.39
0.88 0.73 0.38 0.87 0.86 0.52
F 0.66 0.92 0.35
0.70 0.92 0.37 0.70 0.930.40
G 0.72
0.95 0.48 0.78 0.92 0.46 0.78 0.92 0.46
H 0.82
0.95 0.74 0.87 0.94 0.76 0.87 0.94 0.76
I 0.83
0.98 0.80 0.90 0.97 0.85 0.90 0.97 0.85
J 0.75 0.95 0.66
0.77 0.95 0.67 0.76 0.96 0.68
K 0.81
0.96 0.73 0.92 0.95 0.80 0.92 0.95 0.81
L 0.84 0.94 0.70
0.95 0.92 0.74 0.95 0.960.82
M 0.88
0.97 0.82 0.95 0.97 0.87 0.95 0.970.87
N 0.70
0.96 0.65 0.85 0.95 0.77 0.84 0.95 0.76
O 0.80 0.93 0.72
0.86 0.95 0.79 0.82 0.96 0.77
P 0.88 0.95 0.78
0.90 0.96 0.81 0.88 0.98 0.84
A ve 0.76 0.92 0.59 0.82 0.91 0.62 0.81 0.94 0.65
Sen = Sensitivity, Spe = Specificity, Ove = Overlap.
specificity and sensitivity values were computed by com-
paring the pixels in the segmentation outputs with the pixels
in the gr ntationere
ob ineddifferches: 1) dividing bi-
lateral-filtered images into three clusters; 2)S-
filteredgteus il
image into three clusters and discarding unlikely candidates,
wh cs e
values highlighted k
hsie valuesl t
s, and the values
Table 3. Evaluation of the effectiveness of the proposed sin-
usoid segmentation approach using seven test images. The
ound truth data.
using three The segme outputs w
taent approa dividing O
imaes ino thre clters;3) dividing OS-ftered
hicwerepredited uing asupport vector machin. The
with a dark gr
s, theay
high bac
ighted ground have
with the
grayighet senstivitia ligh
background have the highest specificitie
highlighted with a mid-light gray background have the
highest overlap.
1) Bilateral
Filtering 2) OS Filtering 3) OS Filtering +
SenSpeOveSenSpe Ove Sens SpeOve
A0.800.95 0.620.83 0.95 0.62 0.82 0.960.66
B0.860.97 0.77 0.88 0.97 0.78 0.88 0.990.84
0.99 0.770.88 0.98 0.80 0.88 0.990.82
D0.870.96 0.780.91 0.96 0.81 0.91 0.970.83
0.980.840.910.98 0.86 0.91 0.980.86
0.980.790.900.97 0.81 0.90 0.97 0.81
G0.870.99 0.860.91 0.99 0.89 0.91 1.000.90
Ave0.88 0.
50.97 0.789 0.97 0.80 0.89 0.980.82
Figure 11. Comparison between the images obtained using
the orientation-selective filter and with a support vector
machine to eliminate unlikely candidates. (a) Result of ex-
traction from filtered image; (b) Result of extraction from
discarding and filtered image.
Open Access OJMI
ducted based on the gray-level histogram features, or
primary statistics. Our results showed that this increased
the specificity by 6% without any reduction in the sensi-
tivity. Another experiment compared the results obtained
using an edge-preserving bilateral filter with the conven-
tional method, which showed that the average sensitivity
of the proposed method was 3% higher than that with the
bilateral filter, while the difference in the specificity was
less than 1%. In addition, the proposed method yielded
2% of higher overlaps. We confirmed that better results
were obtained with the proposed method cmpared with
the bilateral filter based one experimental area ratios
Tabesh, M. Teverovskiy, H.-Y. Pang, V. P. Kumar, D.
. Saidi, “Multifeature Prostate
obtained, which was attributable to the effect of linking
edges along a boundary. Future improvements will in-
clude handling liver biopsy images, as well as the surgi-
cal specimens considered in the present study. Liver bi-
opsies involve collecting cells from the liver by tapping a
needle into it, which means that the sinusoids tend to be
distorted by the compression of cells and they are less
visible. Liver biopsy images are very important because
they are prepared in greater numbers than surgical
The proposed method is considered to be effective
from a perspective of computer diagnostic support in pa-
thological applications. However, a high level of judg-
ment is required to determine the morphological proper-
ties of sinusoids because their diagnosis is based on rela-
tive differences compared with normal liver cells. Thus,
this method could provide diagnostic support by making
quantitative information available to physicians based on
the structural analysis of sinusoids. In addition, the struc-
ture of sinusoids is important from an image processing
perspective. Cells, sinusoids, and stroma are dominant in
hepatic pathological images, and the extraction of sinu-
soids provides important information that facilitates the
recognition of other structures, such as the cells and
stroma. Furthermore, it may be possible to quantify the
thickness and structure of cells by extracting the remain-
ing areas after the sinusoids and stroma have been elimi-
7. Acknowledgements
The authors would like to thank Dr Yuri Murakami for
her helpful guidance and Dr Tokiya Abe for arranging
the experiments. This work was supported by grants from
the New Energy and Industrial Technology Development
Organization (NEDO) of Japan.
[1] M. Ogura, A. Saito, H. Graf, E. Cosatto, C. Malon, A.
Marugame, T. Kiyuna, Y. Yamashita and M. Fukumoto,
“The E-Pathologist Cancer Diagnosis Assistance System
for Gastric Biopsy Tissues,” Analytical Cellular Pathol-
ogy, Vol. 4, 2011, p. 34.
[2] A.
Verbel, A. Kotsianti and O
Cancer Diagnosis and Gleason Grading of Histological
Images,” IEEE Transactions on Medical Imaging, Vol.
26, No. 10, 2007, pp. 1366-1378.
[3] S. Naik, S. Doyle, M. Feldman, J. Tomaszewski and A.
Madabhushi, “Gland Segmentation and Computerized
Gleason Grading of Prostate Histology by Integrating
Low-, High-Level andrmation,”
N. Rajpoot, “Pattern Recog-
Pattern Recog-
Domain Specific Info
[4] N. R. Muhammad Arif, “Classification of Potential Nu-
clei in Prostate Histology Images Using Shape Manifold
Learning,” International Conference on Machine Vision,
2007, pp. 113-118.
[5] N. Metin, A. M. Gurcan and
nition in Histopathological Images: An ICPR 2010 Con-
test,” International Conference on ICPR, Islamabad, 28-
29 December 2010, pp. 226-234.
[6] P.-W. Huang and Y.-H. Lai, “Effective Segmentation and
Classification for HCC Biopsy Images,”
nition, Vol. 43, No. 4, 2010, pp. 1550-1563.
[7] C. Atupelage, H. Nagahashi, M. Yamaguchi, T. Abe, A.
Hashiguchi and M. Sakamoto, “Computational Grading
of Hepatocellular Carcinoma Using Multifractal Feature
Description,” Journal of Computerized Medical Imaging
and Graphics, Vol. 37, No. 1, 2012, pp. 61-71.
[8] F. T. Bosman, F. Carneiro, R. H. Hruban and N. D.
Theise, “WHO Classification of Tumours of the Digestive
ateral Filtering for Gray
Chen, S.
d Assess-
System,” World Health Organization, 4th Edition, 2010.
[9] P. J. Scheuer and J. H. Lefkowitch, “Scheuer’s Liver Bio-
psy Interpretation,” Saunders Elsevier, 8th Edition, 2010.
[10] C. Tomasi and R. Manduchi, “Bil
and Color Images,” IEEE International Conference on
Computer Vision, Bombay, 1998.
[11] B. H. Hall, M. Ianosi-Irimie, P. Javidian, W.
Ganesan and D. J. Foran, “Computer-Assiste
ment of the Human Epidermal Growth Factor Receptor 2
Immunohistochemical Assay in Imaged Histologic Sec-
tions Using a Membrane Isolation Algorithm and Quanti-
tative Analysis of Positive Controls,” BMC Medical Im-
aging, Vol. 8, 2008, p. 11.
[12] W. Wang, J. Ozolek, D. Slepcev, A. Lee, C. Chen and G
Rohde, “An Optimal Transportation Approach for Nu-
clear Structure-Based Pathology,” IEEE Transactions on
Medical Imaging, Vol. 99, 2011, p. 1.
[13] J. A. Bilmes, “A Gentle Tutorial of the EM Algorithm
and Its Application to Parameter Estimation for Gaussian
Mixture and Hidden Markov Models,” Tech. Rep. TR-97-
021, International Computer Science Institute, Berkeley,
[14] K.-R. Müller, S. Mika, G. Rätsch, K. Tsuda and B.
Schölkopf, “An Introduction to Kernel-Based Learning
Algorithms,” IEEE Transactions on Neural Networks,
Vol. 12, No. 2, 2001, pp. 181-201.
[15] C.-C. Chang and C.-J. Lin, “Libsvm: A Library for Sup-
Open Access OJMI
Open Access OJMI
M Transactions on Intelligentport Vector Machines,” AC
Systems and Technology, Vol. 2, No. 3, 2011, pp. 1-7.
[16] H. A. Edmondson and P. E. Steiner, “Primary Carcinoma
of the Liver: A Study of 100 Case
cropsies,” Cancer, Vol. 7, No. 3, 1954
s among 48,900 Ne-
, pp. 462-503.<462::AI