Computer assisted image analysis in the diagnosis of gynaecological lesions A quantitative and comparative investigation of haematoxylin eosin with special dyes on tissue lable at ScienceDirect Journa[.]
Trang 1Original Article
Computer-assisted image analysis in the diagnosis of gynaecological
lesions: A quantitative and comparative investigation of
haematoxylin-eosin with special dyes on tissue
F.M Onyijea,b,*, A.A Ngokereb, A.E Lighac, O.O Mgbered, O.G Avwioroe
a Department of Medical Laboratory Science, Faculty of Basic Medical Sciences, College of Health Sciences, Niger Delta University, Wilberforce Island,
Bayelsa State, Nigeria
b Department of Medical Laboratory Science, Faculty of Health Science and Technology, Nnamdi Azikiwe University, Awka, Anambra State, Nigeria
c Department of Human Anatomy, Faculty of Basic Medical Sciences, College of Health Sciences, Niger Delta University, Wilberforce Island, Bayelsa State,
Nigeria
d Institute of Community Health, University of Houston, Texas Medical Center, Houston, TX, USA
e Faculty of Science, Delta State University, Abraka, Nigeria
a r t i c l e i n f o
Article history:
Received 24 July 2016
Accepted 1 November 2016
Available online xxx
Keywords:
Gynaecological lesion
Dye
Morphometry
Computerized Quantitative Analysis
Image analysis
a b s t r a c t
Background: Morphometry and quantitative image analysis have shown considerable potential for diagnostic applications because they eliminate subjectivity and increase the measurement reproduc-ibility In our study, we investigated the use of morphometric quantitative image analysis as a tool to aid diagnosis of lesion tissues stained with seven staining techniques
Material and methods: Thirteen already established lesions were examined in this study Tissues were sectioned at 5mm thickness using the rotary microtome and prepared for staining according to standard methods before staining them with seven different histochemical dyes: Haematoxylin and Eosin (H&E), Masson's Trichrome (MT), Periodic Acid Schiff (PAS), Phosphotungstic Acid Haematoxylin (PTAH), Southgate Mucincamine (SGM), Alcian Blue (AB) and Verhoeff Van Gieson (VVG) Photomicrographs were analyzed using imageJ for percentage area and intensity measurement
Results: The average % area and intensity measurement based on the seven staining techniques used in this study were 47.88% and 122.23 points (pts) The lesions, adenomyosis in MT (80%), squamous cell carcinoma in PAS (80%) andfibroma in PTAH (77%) stained above the average % area The squamous cell carcinoma stained with MT was the most intense tissue (55 pts), followed by Brenner tumour in PTAH (61 pts) and squamous cell carcinoma in PAS (69 pts) There were significant correlations between the % area of tissues covered by PAS and H&E (r ¼ 0.599, p ¼ 0.030), and PAS and MT (r ¼ 0.572, p ¼ 0.041) as well as in the intensity of tissues in PAS and H&E (r¼0.615, p¼0.025), VVG and H&E (r ¼ 0.707, p ¼ 0.007), and VVG and PTAH (r¼ 0.577, p ¼ 0.039)
Conclusion: MT staining technique produced the best results in the % area of tissues covered and the intensity measurements, and therefore was recommended for routine use alongside with H&E in diagnostic histopathology
© 2016 The Chinese Oncology Society Production and hosting by Elsevier B.V This is an open access article under the CC BY-NC-ND license (http://creativecommons.org/licenses/by-nc-nd/4.0/)
1 Introduction Histopathology involves the use of dyes for microscopic exam-ination of tissues in part to spot signs and characteristic features of diseases Recently, computer assisted diagnosis (CAD) has been widely used in developed countries.1e3 In general, it is well-recognised that the importance of quantitative analysis in histo-pathology cannot be overemphasized, particularly because most pathology diagnoses are based on the subjective opinion of
* Corresponding author Department of Medical Laboratory Science, Faculty of
Basic Medical Sciences, College of Health Sciences, Niger Delta University,
Wilber-force Island, Bayelsa State, Nigeria.
E-mail address: onyijefelix@yahoo.com (F.M Onyije).
Peer review under responsibility of The Chinese Oncology Society.
Contents lists available atScienceDirect Journal of Cancer Research and Practice
j o u r n a l h o m e p a g e : h t t p : / / w w w j o u r n a l s e l s e v i e r c o m / j o u r n a l o f c a n c e r
-r e s e a -r c h - a n d - p -r a c t i c e
http://dx.doi.org/10.1016/j.jcrpr.2016.11.002
2311-3006/© 2016 The Chinese Oncology Society Production and hosting by Elsevier B.V This is an open access article under the CC BY-NC-ND license ( http:// creativecommons.org/licenses/by-nc-nd/4.0/ ).
Journal of Cancer Research and Practice xxx (2016) 1e9
Trang 2pathologists.4The utilization of CAD has the potential to improve
the practice of histopathology and other related discipline in
different ways.5With the influx of different imaging technologies,
histological images of diverse staining techniques including
immunehistochistry are produced for either manual or automated
analysis Generally, such images are better obtained at low
magnification.6ImageJ, which is a public domain, Java-based image
processing program developed at the National Institute of Health,7,8
is used in quantitative analysis of histological stains on tissues,
among other functions.9
Although the human eye is sensitive to a number of factors,
including luminosity and variation in contrast and brightness,
critical visual analysis is open to subjective interpretation The
essence of morphometry is to eliminate subjectivity and increase
the reproducibility of measurements.10The quantification of
per-centage area covered by different stains on different tissues as well
as the intensity measurement have not been given due attention, a
situation that has resulted in the lack of improvement in routine
staining procedures Advances in computer image analysis
tech-niques allow more accurate quantification of histopathological
analysis.11
The mechanisms of tissue staining are histochemical in nature,
which is the application of chemical substances onto a tissue in
order to produce a visible outcome To achieve this reaction, the
tissue takes up the stain either through a chemical or physical
method (adsorption, absorption, solubility, osmotic pressure or
capillary attraction) The type and nature of this uptake may differ from one tissue to another, as well as by the technique employed There are two groups of dyes commonly used: 1) the natural group
of dyes, which are derived from plants and insects, 2) the synthetic dyes, which are derived from coal tar These groups of dyes can be applied regressively (differentiation), progressively (without dif-ferentiation) or directly (without the use of mordant).12 The importance of histopathological staining cannot be over-emphasized, especially as the process allows for adequate and reliable histological results to be obtained Hence, in this study, we investigated the use of morphometric quantitative image analysis
as a tool to aid in the diagnosis of stained lesions using seven staining techniques
2 Materials and methods 2.1 Tissue
Ninety-one (91) cut sections from thirteen (13) different established lesions retrieved from archives of the Histopathology Laboratory at Braithwaite Memorial Specialist Hospital, Port Har-court, Nigeria were used for this analysis The tissues were sectioned at 5mm thickness using the rotary microtome and pre-pared for staining, using the methods described by Suvarna et al.13
Fig 1 Chart showing the % area measurement of lesions within stain methods Haematoxylin and Eosin (H&E), Masson's Trichrome (MT), Periodic Acid Schiff (PAS), Phospho-tungstic Acid Haematoxylin (PTAH), Southgate Mucincamine (SGM), Alcian Blue (AB) and Verhoeff Van Gieson (VVG).
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 2
Trang 32.2 Staining methods
The staining techniques used in this study were; Haematoxylin
and Eosin,14 Masson's Trichrome,15 Periodic Acid Schiff,16
Phos-photungstic Acid Haematoxylin,17Southgate Mucincamine,18Alcian
Blue19and Verhoeff Van Gieson20
2.3 Principle of H&E
This is the oxidation of haematoxylin to haematein by natural or
chemical agent and characteristically requires a mordant for its
effectiveness The haematein dye binds to acidic components of the
tissue that results in a blue coloration On the other hand, eosin
stains pink and has the ability to bind to basic components due to
its acidic nature
2.4 Principle of PTAH
The phosphotungstic acid binds to the available haematein to
form a blue lake pigment, which stains the muscle cross striations,
fibrin, nuclei, and other tissue elements blue, Phosphotungstic acid
stains collagenfibres red to brown
2.5 Principle of PAS
Periodic acid oxidizes 1:2 glycol groups to yield aldehydes,
which thereafter recolor Schiff's reagent The principle of the Schiff’’s reaction depends on the conversion or loss of the quinoid structure and the masking of chromophores This forms a colorless compound called leuco-fuchsin which can change to the classic red reaction when exposed to aldehydes Excess Schiff reagent is removed by potassium metabisulfite thus preventing a false positive by oxidation of the reagent in the tissue
2.6 Principle of AB Alcian blue consists of a group of polyvalent basic dyes that are characteristically water-soluble, whose blue color can
be attributed to the presence of copper in the molecule
It stains sulphated mucoploysaccharides at pH 1, and stains both sulphated and carboxylated mucopolysaccharides at
pH 2.5
2.7 Principle of MT
As the name implies, three dyes are employed selectively staining muscle, collagenfibers, fibrin, and erythrocytes The gen-eral rule in trichrome staining is that less porous tissues are col-oured by the smallest dye molecule; thus, whenever a dye of large molecular size penetrates, it will always do so at the expense of the smaller molecule Others suggest that the tissue is stainedfirst with
Fig 2 Chart showing the intensity measurement of lesions within stain methods Haematoxylin and Eosin (H&E), Masson's Trichrome (MT), Periodic Acid Schiff (PAS), Phos-photungstic Acid Haematoxylin (PTAH), Southgate Mucincamine (SGM), Alcian Blue (AB) and Verhoeff Van Gieson (VVG).
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 3
Trang 4the acid dye Biebrich Scarlet, which binds with acidophilic tissue
components Thereafter, when treated with phospho amino acids,
the less permeable components retain the red stain, while the red
stain is extracted from the collagen Simultaneously, this causes a
link to which collagen can bind
2.8 Principle of SGM
Aluminum is believed to form a chelation complex with
carmine, changing the molecule to a positive charge and allowing it
to bind with the acid substrates of low density (such as mucins)
2.9 Principle of VVG
For VVG, the tissue is stained with a regressive haematoxylin,
consisting of ferric chloride and iodine The differentiation is
accomplished by using excess mordant (ferric chloride) to break the
tissue-mordant dye complex The dye is attracted to the larger
amount of mordant in the differentiating solution and is removed
from the tissue Elastic tissue has the strongest affinity of the
iron-haematoxylin complex, and retains the dye longer than other tissue
elements
2.10 Microscopy and data acquisition from photomicrographs
The stained tissue slides were viewed using OMAX 40X-2000X
built-in 3.0 MP digital camera compound LED Binocular Micro-scope The histologic image characteristics of the stains were analyzed and the stain intensity and percentage (%) area stained were measured using ImageJ 1.48 version (National Institute of Health, USA)
2.11 Percentage (%) area and intensity measurement Imported RGB images are converted to gray scale images on ImageJ The software quantifies the staining intensity by measuring the pixel value of each pixel in grayscale images following a threshold of areas of staining activity, and converting the pixel value to brightness value or gray value, in a scale of
0e255 from the less bright (that is lower points (pts) and greater intensity) to brighter (that is higher points and reduced intensity)
2.12 Statistical analysis 2.12.1 Analytical package used Measurement system analyses were performed to compare the staining methods by lesions using JMP statistical discovery™ soft-ware, version 12.0 (SAS Institute, Cary, NC, USA) For all tests per-formed, the probability value of 0.05 was used as the threshold for determining statistical significance level
Fig 3 Partition flowchart for % area by stain methods and lesions group.
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 4
Trang 52.13 Ethical approval
Ethical approval for this study was obtained from the Hospital
Management Board of Rivers State through the ethics committee of
BMSH
3 Results
3.1 Percentage (%) area measurement
Fig 1, shows the average % area (48%) of the lesions covered
by stains Adenomyosis in MT (80%) followed by squamous cell
carcinoma in PAS (80%) and fibroma in PTAH (77%) were
above the average percentage area However, the % area of
con-dyloma in AB (17%), adenocarcinoma in PAS (18%) andfibroma in
SGM (20%) were generally lower than the average % area in our
study
3.2 Intensity measurement
Fig 2, shows the mean intensity measurement (122.23 pts) for
the different lesions impacted by the different stains Lesions in MT,
PTAH and H&E were highly intense, with squamous cell carcinoma
in MT (55 pts) and Brenner tumour in PTAH (61 pts) being the most
intense lesions Contrarily, mucinous cyst in PAS (201 pts) and
endometrial polyp in AB (181 pts) were the least intense lesions observed in our study
3.3 Partitionflowchart
Fig 3, shows the partitionflow chart of the staining technique and the lesion group The lesion group stained with VVG, PATH and MT had a highest mean % area of 57%, while those stained with SGM, PAS and H&E recorded a lower mean % area of 41% The group with the higher average % area recorded 61% when stained in the right split (fibroma, condyloma, adenomyosis, squamous cell carcinoma, mucinous cyst, leiomyoma, Brenner tumour and adenoma) The left split consisting of demoid cyst, choriocarcinoma, adenocarcinoma and endometrial hyperplasia
of the same group recorded 50% Based on the SGM, PAS and H&E staining techniques, the following lesions; endometrial polyp, squamous cell carcinoma, leiomyoma, Brenner tumour and ade-noma recorded an average % area of 49% on the right split Thirty two sections of the tissue samples in the left split of the flow-chart recorded the mean % area of 35%.Fig 4, shows the partition flowchart for intensity measurement by stain methods and le-sions group Based on our analysis two main groups were
iden-tified as follows; MT and PTAH (group 1), which recorded overall average intensity measurement of 84 pts compared to H&E, VVG, SGM, PAS and AB (group 2), which produced an average intensity
Fig 4 Partition flowchart for intensity measurement by stain methods and lesions.
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 5
Trang 6measurement of 137 pts However, subsequent assessment
in-dicates that squamous cell carcinoma, Brenner tumour, mucinous
cyst, condyloma, fibroma, endometrial polyp and
choriocarci-noma when stained using techniques in group 1 produced a
more intense measurement (74 pts) compared to 96 pts obtained
when demoid cyst, adenoma leiomyoma, endometrial
hyperpla-sia, adenomyosis and adenocarcinoma were stained using
methods in group 2
3.4 Matched pairs analysis of percentage area of stains
Fig 5, presents the matched pairs analysis of % area
measure-ments by staining methods Using H&E as the standard (mean ¼
42.6%), we compared the six staining methods Results from this
analysis indicate that the mean % areas of tissues covered were
significantly improved by 50% with MT (64.4%, p ¼ 0.0002), 27%
with PTAH (54.2%, p¼ 0.021) and 26% with VVG (53.7%, p ¼ 0.004)
when compared to H&E However, SGM, PAS and AB performed
poorly by producing mean % areas that were below or failed to
match that of H&E (p > 0.005)
3.5 Matched pairs analysis of stains intensity
Fig 6, shows the matched pairs analysis of the intensity
mea-surements by staining methods Tissues stained using MT and PTAH
techniques produced the most intense mean measurements that
were 34% (80.9 pts, p¼ 0.0001) and 29% (87.3 pts, p ¼ 0.0001)
better than those produced from tissues stained with H&E, which
served as the standard On the other hand, intensity measurement for tissues stained with AB and PAS performed poorly being 33% (162.9 pts, p¼ 0.0001) and 18% (144.7 pts, p ¼ 0.009) less intense than those stained with H&E
3.6 Pair-wise correlation of staining methods
Tables 1 and 2, show the pair-wise correlation of % area and intensity measurement by staining methods With reference
to % area, except for significant correlations noted between PAS and H&E (r ¼ 0.599, p ¼ 0.030) and PAS and MT (r ¼ 0.572, p
¼ 0.041), all other staining methods were not statistically significant (p > 0.05) Similarly, our study noted significant correlations in intensity measurements by staining methods as -follows; PAS and H&E (r ¼ 0.615, p ¼ 0.025), VVG and
H&E (r ¼ 0.707, p ¼ 0.007) and, VVG and PTAH (r ¼ 0.577, p ¼ 0.038)
4 Discussion The % area of the tissue covered by stain as well as the in-tensity is dependent on the type of stain and tissue Stain uptake
is often due to dye-tissue or reagent-tissue affinities, perme-ability and pore size MT covered a greater % area with a high intensity measurement for most tissues compared to other type
of staining methods used in our study This was reflected in the outcome of theflow chart (Fig 3), where it was observed that VVG, PTAH and MT occupied a larger surface area than SGM, PAS Fig 5 Matched pairs analysis of percentage area of stains.
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 6
Trang 7and H&E MT was originally used for the demonstration of
con-nective and mesenchymal tissues,13with the ability to impact up
to three colours to tissue and bind with collagenfibres.24Suvarna
et al.13 stated that smaller molecule dyes will stain any of the
basic type of tissues This is similar to the rule of trichrome
staining, which states that a smaller dye molecule will penetrate
and stain tissue element, but whenever a larger dye molecule can
penetrate the same element, the smaller molecule will be
dis-placed by it.21Other factors that contribute to dye-tissue affinity
include Coulombic attraction, which is also referred to as salt
links or electrostatic bonds Also, Van der Waals forces including
intermolecular attractions, hydrogen bonding, covalent bonding,
and the hydrophobic effect are among other contributing
factors.22,23,13
In our study we observed that MT produced a 50% improvement
in the % area of tissues covered over that of H&E It is possible that
its three-in-one (trichrome) combination of dyes may be
respon-sible for the large % area covered and high intensity measurement
This is in line with the report of Street et al.,25who stated that MT is
a widely used stain in Histology MT is primarily used to enhance
the visibility of collagen and elasticfibres and most tissues in the
body contain thesefibres
Adenomyosis refers to the growth of the basal layer of the
endometrium into the myometrium.26The elevated % area covered
by MT in adenomyosis is an indication that adenomyosis muscle
cells have the capacity to absorb stain and has a large surface area,
as a result of its dense fascicles with elongated cytoplasm and nuclei having afinely dispersed chromatin.27
Squamous cell carcinoma was the most covered and intensified tissue in our study, which may be due to the numerous malignant cells and their metachromatic nature Our observation is in line with the publication of the British Association for the Advancement
of Science where it was noted that malignant cells take up stains two to three times more than the normal cells.28This affirms why malignant lesions had a higher intensity and % area measurement than the benign lesions with MT Thesefindings support the work
of Ouyang et al.,29who reported that MT is superior to H&E in the examination of tissue sections, and concluded that MT would be a valuable tool for analysing autopsy tissues The % area covered by stains and intensity measurement aid in the visualization of cells and tissues However, poor coverage and intensity may obscure accurate and reliable diagnosis
5 Conclusion The quantitative and comparative analysis of MT in both % area and intensity measurement with other techniques in this study clearly demonstrates that MT is better than other staining methods Therefore, MT was recommended for routine use alongside with
H&E in diagnostic histopathology
Fig 6 Matched pairs analysis of intensity of stains.
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 7
Trang 8Table 1
Pair-wise correlation of % area measurements by stain methods.
Variable By variable Correlation Count Lower 95% Upper 95% Significant probability Graphical representation
Table 2
Pair-wise correlation of intensity measurements by stain methods.
Variable by Variable Correlation Count Lower 95% Upper 95% Significant Probability Graphical Representation
F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 8
Trang 91 Dio K Computer-aided Diagnosis in Medical Imaging: Historical Review, Current
Status and Future Potential vol 31 2007:198e211 (4e5)
2 Mendez AJ, Tahoces PG, Lado MJ, Souto M, Vidal JJ Computer-aided diagnosis:
automatic detection of malignant masses in digitized mammograms Med Phys.
1998 Jun;25:957e964
3 Doyle S, Feldman M, Tomaszewski J, Madabhushi A A boosted bayesian
multi-resolution classifier for prostate cancer detection from digitized needle
bi-opsies IEEE Trans Biomed Eng 2012;59(5):1205e1218 http://dx.doi.org/
10.1109/TBME.2010.2053540 (Epub 2010 Jun 21).
4 Gurcan MN, Boucheron L, Can A, Madabhushi A, Raipoot, Yener B
Histopath-ological image analysis: a review IEEE Rev Biomed Eng 2009;2:147e171
5 Basavanhally AN, Ganesan S, Agner S, et al Computerized image-based
detection and grading of lymphocytic infiltration in HER2þ breast cancer
his-topathology IEEE Trans Biomed Eng 2010 Mar;57(3):642e653
6 Murphy DB Fundamentals of Light Microscopy and Electronic Imaging
Wiley-Liss; 2001
7 Schneider CA, Rasband WS, Eliceiri KW NIH Image to ImageJ: 25 years of image
analysis Natl Methods 2012;9(7):671e675
8 Collins TJ ImageJ for microscopy BioTech 2007;43(1):25e30
9 Ellen CJ Quantitative analysis of histological staining and fluorescence using
imageJ Anat Rec 2013;296(3):378e381
10 Andrea CE, Bleggi-Torres LF, Alves MTS Analise da morfometria nuclear:
descriç~ao da metodologia e o papel dos softwares de ediç~ao de imagem Braz J
Pathol Lab Med 2008;44(1):51e57
11 Rangan GK, Tesch GH Quantification of renal pathology by image analysis
(Methods in Renal Research) Nephrology 2007;12(6):553e558
12 Avwioro OG Staining In: Histochemistry and Tissue Pathology Principles and
Techniques third ed Claverianum Press Nigeria Limited; 2014:133e168
13 Suvarna SK, Layton C, Bancroft JO Bancroft's Theory and Practice of
Histopath-ological Techniques seventh ed Churchill Livingstone; 2013
14 Harris HF On the rapid conversion of haematoxlin into haematein in staining
reactions J Appl Microsc Lab 1900;3:777
15 Masson P Some histological methods Trichrome staining and their preliminary
technique Bull Int Assoc Med 1929;2:75
16 McManus JFA, Mowry RW Staining Methods, Histologic and Histochemical London: Harper & Row; 1964:268
17 Shum MW, Hon JKY A modified phosphotugstic acid haematoxylin stain for formalin fixed tissue J Med Lab Technol 1969;26:38
18 Southgate HW Note on preparing mucicarmine J Pathology Bacteriol 1927;30:
729
19 Mowry RW Alcian blue technique for the histochemical study of acid carbo-hydrates J Histochem Cytochem 1956;4:407
20 Verhoef FH Some staining methods of wide applicability, including a rapid differential stain for elastic tissue J Am Med Assoc 1908;50:876
21 Horobin RW Structural staining relationships in histochemistry and Biological Staining I J Microsc 1980;119:345e355
22 Prento P Staining of Macromolecules Biotechnic and Histochemistry vol 84 2009:139e158
23 Horobin RW, Bennion PJ The interrelation of the size and substantivity of dyes: therole of van der waal's attractions and hydrophobic bonding in biological staining Histochromemie 1973;33:191e204
24 Stephen MF, Jagdish B, Michael JS, Douglas W, William CW, Marcos R Patho-logic fibrosis and matrix connective tissue in the subaortic myocardium of patients with hypertrophic cardiomyopathy J Am Cell Cardiol 1991;17(6): 1343e1351
25 Street JM, Souza ACP, Alvarez-Prats A, et al Automated quantification of renal fibrosis with Sirius Red and polarization contrast microscopy Physiol Rep 2014;2(7) http://dx.doi.org/10.14814/phy2.12088
26 Kumar V, Abul K, Jon CA Robbins Basic Pathology ninth ed Elsevier Saunders; 2013:689
27 Vanni R Uterus Leiomyoma Atlas Genet Cytogenet Oncol Haematol http:// atlasgeneticsoncology.org/Tumors/leiomyomID5031.html Retrieved 12 of March 2016.
28 NewScientist https://books.google.com.ng/books/about/New_Scientist.html? id¼dyM2kv5r7-gC Acccessed 13, March, 2016.
29 Ouyang J, Guzman M, Desoto-Lapaix F, Pincus MR, Wieczorek R Utility of desmin and a Masson's trichrome method to detect early acute myocardial infarction in autopsy tissues Int J Clin Exp Pathol 2010;3(1):98e105 F.M Onyije et al / Journal of Cancer Research and Practice xxx (2016) 1e9 9