Kazim

Ali1, Faheem Abbas 2, Atif Iqbal3

1Kazim Ali, PhD Student, IT Department,

University of Central Punjab Lahore

2Faheem Hussain, MS Student,

IT Department, University

of Central Punjab Lahore

3Atif Iqbal, MS Student, IT Department,

University of Central Punjab Lahore

ABSTRACT

The

segmentation and detection of nuclei in Histopathology images is very important

in these days to indentify different diseases like cancer etc. A large variety

of approaches for segmentation of nuclei in breast cancer histopathology images

have been proposed. Digital pathology

represents one of the major evolutions in modern medicine. Pathological

examinations constitute the

gold standard in many medical protocols, and also

play a critical and legal role in the diagnosis process. In the conventional

cancer

diagnosis, pathologists analyze biopsies to make

diagnostic and prognostic assessments, mainly based on the cell morphology and

architecture distribution. They

vary not only in the core segmentation methods, but also in the pre- and post

processing steps that aim to improve segmentation performance. We will give an

overview of image analysis for the sake of nuclei segmentation and detection.

In this paper, we will use the concept of blob detection in digital image

processing by using three digital image processing algorithms which are

Laplacian of Gaussian (Log), Difference of Gaussian (Dog) and Determinant of

Hessian (Hog). This paper proposes a method of nuclei segmentation and

detection which purely based on image processing techniques. At the end of

paper we will compare results of these three algorithms on a cancer patient’s

dataset.

Keywords—

Segmentaq 2w`3tion, Log, Dog, Doh, Blod

detection.

INTRODUCTION

Breast

cancer is one of the most common cancer along with lung and bronchus cancer,

prostate cancer, colon cancer, and pancreatic cancer among others. Recently, computerized methods have been

rapidly evolving in the area of digital pathology, with growing applications

related to nuclei detection, segmentation, and classification. In cancer

research, these approaches have played, and will continue to play a key role in minimizing human intervention,

consolidating pertinent second opinions, and providing traceable clinical

information. Pathological studies have been conducted for numerous cancer

detection and grading applications, including brain, breast, cervix, lung, and

prostate cancer grading.

Breast

cancer is the most prevalent form of cancers among women and image analysis

methods that target this disease have a huge potential to reduce the workload

in a typical pathology lab and to improve the quality of the interpretation 1. This study investigated blob detection and

blob parameter estimation. Estimated blob parameters have many potential

applications, such as noise filtering and medical image analysis. Many

applications may benefit from these blob parameters, because they produce

simpler or more accurate algorithms 2. The utilization of data science and

machine learning approaches in medical fields proves to be prolific as such

approaches may be considered of great assistance in the decision making process

of medical practitioners. With an unfortunate increasing trend of breast cancer

cases comes also a big deal of data which is of significant use in furthering

clinical and medical research, and much more to the application of data science

and machine learning in the aforementioned domain 3. Pathological examination, in which a series of

H-stained histopathological slides are manually examined by pathologists

for disease diagnosis, is a time-consuming and labor-intensive task. More

importantly, this process is subjective, prone to error, and has large inter-

and intra observer variation. Due to the heterogeneity and morphological

complexity of tumors, it is a challenging task even for well-trained

pathologists to reach an agreement when diagnosing a tumor sample by visual

inspection of H-stained images 45.

Related

Work

1

The procedure can be divided into four main steps: 1) preprocessing with color

un mixing and morphological operators, 2) marker-controlled watershed

segmentation at multiple scales and with different markers, 3) post-processing

for rejection of false regions and 4) merging of the results from multiple

scales. The procedure was developed on a set of 21 breast cancer cases (subset

A) and tested on a separate validation set of 18 cases (subset B). The

evaluation was done in terms of both detection accuracy (sensitivity and

positive predictive value) and segmentation accuracy (Dice coefficient). 6 Starts with an overview of the tissue

preparation, staining and slide digitization processes followed by a discussion

of the different image processing techniques and applications, ranging from

analysis of tissue staining to computer-aided diagnosis, and prognosis of

breast cancer patients. 7 Proposes a method uses two observations,

above. First, real E-only images exhibit lower contrast when compared to

digitally separate E-only images using the wedge-finding method and second, the optical density of

some regions of the H image is significantly higher than the sum of the

optical densities of the corresponding Honly and E-only image. 2 Detects

image blobs and estimated parameters using the determinant of the Hessian

operator. To investigate differential detectors quantitatively, a mathematical

function was used to represent the blobs and to solve the parameters, including

the position, width, length, contrast, offset, and orientation, in a closed

form. 8 Compares six machine learning (ML) algorithms: GRU-SVM, Linear

Regression, Multilayer Perceptron (MLP), Nearest Neighbor (NN) search, Soft max

Regression, and Support Vector Machine (SVM) on the Wisconsin Diagnostic Breast

Cancer (WDBC) dataset by measuring their classification test accuracy and their

sensitivity and specificity values. 3 Presents, discusses, and extracts the major trends from an exhaustive

overview of various nuclei detection, segmentation, feature computation, and

classification techniques used in histopathology imagery, specifically in

hematoxylin–eosin and immunohistochemical staining protocols. 5 Presents

an automated cell nuclei segmentation approach that works with H-stained

images. A color de-convolution algorithm was first applied to the image to get

the hematoxylin channel. Using a morphological operation and thresholding

technique on the hematoxylin channel image, candidate target nuclei and

background regions were detected, which were then used as markers for a

marker-controlled watershed transform segmentation algorithm. Moreover, post processing

was conducted to split the touching nuclei. For each segmented region from the

previous steps, the regional maximum value positions were identified as

potential nuclei centers. These maximum values were further grouped into

k-clusters, and the locations within each cluster were connected with the

minimum spanning tree technique. Then, these connected positions were utilized

as new markers for a watershed segmentation approach. The final number of

nuclei at each region was determined by minimizing an objective function that

iterated all of the possible k-values. The proposed method was applied to the

pathological images of the tumor tissues from The Cancer Genome Atlas study.

Experimental results show that the proposed method can lead to promising

results in terms of segmentation accuracy and separation of touching nuclei.

9 Detection of nuclei can be accomplished by using a CNN as a classifier and

applying sliding window through the whole image. The result will be a

probability map which indicates the probability of each pixel to be the

centroid of a nucleus. A CNN classifier consists of two parts: (i) the feature

extraction part that includes a few convolution layers followed by pooling

layers and an activation function such as a sigmoid, tanh, and ReLU; and (ii)

the classification part which is a few fully connected layers complemented by a

loss function. 10 presents a robust

and accurate novel method for segmenting cell nuclei using a combination of

ideas. The image foreground is extracted automatically using a graph cuts based

binarization. Next, nuclear seed points are detected by a novel method combining

multi-scale Laplacian of Gaussian filtering constrained by distance map based

adaptive scale selection. These points are used to perform an initial

segmentation that is refined using a second graph cuts based algorithm

incorporating the method of alpha expansions and graph coloring to reduce computational

complexity. 11 Proposes a novel segmentation algorithm for detecting

individual nuclei from Hematoxylin and Eosin (H) stained breast

histopathology images. This detection framework estimates a nuclei saliency map

using tensor voting followed by boundary extraction of the nuclei on the saliency

map using a Loopy Back Propagation (LBP) algorithm on a Markov Random Field (MRF).

The method was tested on both whole-slide images and frames of breast cancer

histopathology images.

Proposed

Method

The

proposed method is based on blob detection technique which is used in digital

image processing to detect stars and galaxies in satellite images. But we are

trying to use this technique to detect nuclei in breast cancer images. The

proposed method based on this technique is given by

Fig.

1

The

programming language and libraries which are used in the proposed method to

detect nuclei are given under.

Python Lanuage

Scikit-image: It

is an open source image processing library for the Python programming

language. It includes algorithms for segmentation, geometric

transformations, color space manipulation, analysis, filtering,

morphology, feature detection, and more.

Numpy: It is used to array processing for numbers,

strings, records, and objects).

Matplotlib: It is

a plotting library for the Python programming language and its numerical

mathematics extension NumPy.

The image processing algorithms which are used in this work

are given below.

3.1

Laplacian of Gassian:

Laplacian filters are derivative

filters used to find areas of rapid change (edges) in images. Since derivative

filters are very sensitive to noise, it is common to smooth the image (e.g.,

using a Gaussian filter) before applying the Laplacian. This two-step process

is called the Laplacian of Gaussian (LoG) operation. The mathmatical equation

is:

3.2

Difference of Gaussian:

The DOG performs edge detection by performing a Gaussian blur on

an image at a specified theta (also known as sigma or standard deviation). The

resulting image is a blurred version of the source image. The module then

performs another blur with a sharper theta that blurs the image less than

previously. The final image is then calculated by replacing each pixel with the

difference between the two blurred images. The equation is given.

3.3 Determinant of Hessian: The

Hessian blob detector 42, 43 is based on a 2 × 2 matrix of second-order derivatives of

image intensity I(x, y), called the Hessian matrix. This matrix can be used

to analyze local image structures and it is expressed in the form

H(x, y, ?) =

where Ixx, Ixy, and Iyy

are second-order image derivatives computed using Gaussian function of

standard deviation ?. In order to detect interest features, it searches

for a subset of points where the derivatives responses are high in two

orthogonal directions. That is, the detector searches for points where the

determinant of the Hessian matrix has local maxima.

det(H) = IxxIyy – I2

xy

3.4 Blob Detection Algorithm

Analysis:

LoG is the most accurate and slowest

approach. It computes the Laplacian of Gaussian images with successively

increasing standard deviation and stacks them up in a cube. Blobs are local maxima

in this cube. Detecting larger blobs is especially slower because of larger

kernel sizes during convolution. Only bright blobs on dark backgrounds are

detected.

DoG

is a faster approximation of LoG approach. In this case the image is blurred

with increasing standard deviations and the differences between two

successively blurred images are stacked up in a cube. This method suffers from

the same disadvantage as LoG approach for detecting larger blobs. Blobs are

again assumed to be bright on dark.

DoH

is the fastest approach. It detects blobs by finding maxima in the matrix of

the Determinant of Hessian of the image. The detection speed is independent of

the size of blobs as internally the implementation uses box filters instead of

convolutions. Bright on dark as well as dark on bright blobs is detected. The

downside is that small blobs (<3px) are not detected accurately.
4.
Dataset
The
proposed methodology was tested and evaluated on de-identified and de-linked
images of histopathology specimens from the Department of Pathology, Christian Medical
College Hospital (CMC), the proposed method was validated on eight
representative images of H stained breast cancer histopathology sections.
5. Result
The resultant
output of LoG, DoG and Doh algorithms are given below.
Fig
(a). Original image
Fig(b). after applying LoG algorithm
Fig(c). After applying DoG
Fig(d). After applying
DoH
5.1
Performance Evaluation Metrics: The following metrics are used for performance evaluation.
Recall=R = Tp / ( Tp+ Tn)
Precision=P = Tp / (Tp + Fp )
Where
True
Positive (TP):
correctly identified nuclei.
False
Positive (FP):
incorrectly identified non nuclei.
True
Negative (TN):
correctly rejected.
False
Negative (FN):
incorrectly rejected nuclei.
Table
no. 1
Algorithms
Manual
Count of Nuclei
Count of Nuclei by Proposed Method
Recall
Precision
LoG
12115
11886
98%
100%
DoG
12115
3850
31%
100%
DoH
12115
11213
92%
100%
Conclusion
In this work we have used three
algorithms, LoG, DoG and HoG for nuclei detection in cancer image dataset. The
LoG achieve 98% recall and 100% precision, the DoG achieves 31% recall and 100%
precision and, the DoH achieves 92% recall and 100% precision. According to
above results, it is proved that the performance of LoG and HoG is much better
than DoG. So DoG algorithm is not efficient, we cannot use it in this specific
scenario. We only talk about the perfomance LoG and DoH. According table No.1
results LoG is slightly better than DoH. After a literature review we find the
DoG is costly than DoH. Therefor in our point of view about these two algorithms
is (i) LoG will be used when we have more computational power and required more
and more accuracy (ii) the DoH will be better in the environment where less
computational power is available but when we can compromise on slightly low
accuracy.
REFERENCES
1
Breast Cancer Histopathology Image
analysis by Mitko Veta.
2 Blob Detectczzion with the Determinant of the Hessian
by Xiaopeng Xu. Computer and
Information Engineering College, Inner Mongolia Normal University, Hohhot,
010022, China.
3 Methods
for Nuclei Detection, Segmentation, and Classification in Digital
Histopathology: A Review—Current Status and Future Potential. Humayun Irshad, Student Member, IEEE, Antoine
Veillard, Ludovic Roux, and Daniel Racoceanu, Member, IEEE. IEEE REVIEWS IN BIOMEDICAL ENGINEERING, VOL. 7,
2014.
4
Convolutional Neural Networks for Biomedical Image Analysis, Alex Kalinin, PhD
Candidate DCM, University of Michigan June 1, 2017.
5
Faliu Yi, Junzhou
Huang, Lin Yang, Yang Xie, uanghua Xiao, "Automatic extraction of cell nuclei
from H&E-stained histopathological images," J. Med. Imag. 4(2), 027502
(2017), doi: 10.1117/1.JMI.4.2.027502.
6
Breast Cancer Histopathology Image Analysis: A Review
Mitko
Veta, Josien P. W. Pluim, Paul J. van Diest, and Max A. Viergever. IEEE
TRANSACTIONS ON BIOMEDICAL ENGINEERING, VOL. 61, NO. 5, MAY 2014.
7
ALGORITHM AND BENCHMARK DATASET FOR STAIN SEPARATION IN HISTOLOGY IMAGES.
Michael T. McCann, Joshita Majumdar, Cheng Peng, Carlos A. Castro, Jelena Kova.
Magee-Womens Research Institute and Foundation of the University of Pittsburgh,
Pittsburgh, PA, USA.
8
On Breast Cancer Detection: An Application of Machine
Learning
Algorithms on the Wisconsin Diagnostic Dataset
Abien
Fred M. Agarap. Department of Computer Science
Adamson
University Manila, Philippines. arXiv:1711.07831v1 cs.LG 20 Nov 2017.
9
Detection of Nuclei in H Stained
Sections Using Convolutional Neural Networks Mina Khoshdeli, Richard Congy, and
Bahram Parvin, Biomedical and Electrical Engineering Department, University of
Nevada, Reno, NV, U.S.A, Amador Valley
High School, Pleasanton, Ca, U.S.A.
10
Improved Automatic Detection &
Segmentation of Cell Nuclei in Histopathology Images Yousef Al-Kofahi,
Wiem Lassoued, William Lee, and Badrinath Roysam, Senior Member, IEEE.
11 Automated Segmentation
of Nuclei in Breast Cancer Histopathology Images, Maqlin Paramanandam, Michael O'Byrne, Bidisha Ghosh, Joy John Mammen,
Marie Therese Manipadam, RobinsonThamburaj, Vikram Pakrashi.
12 http://bioimage.ucsb.edu research/bio-segmentation
/.