View of Detection and Tracking Breast Cancer Using Image Processing: Advance Studies Dr.

(1)

http://annalsofrscb.ro 14205

Detection and Tracking Breast Cancer Using Image Processing: Advance Studies

Dr. Mukta Jagdish ^a*, Eldalaine Torres Vargas ^b,Dina Arrieta Ramos ^c, Natalia Parraci Nuñez^d

a Lecturer, Mahatma Gandhi Government College, Mayabunder, North and Middle Andaman, Andaman and Nicobar Islands, India

b Universidad Nacional de Ucayali, Perú.

[email protected] https://orcid.org/0000-0002-6786-0975

c CESFAM Los Quillayes, Santiago, Chile.

[email protected]

d CESFAM Los Quillayes, Santiago, Chile.

[email protected]

* Corresponding Author: Dr. Mukta Jagdish

Lecturer, Mahatma Gandhi Government College, Mayabunder, North and Middle Andaman, Andaman and Nicobar Islands, India

Abstract- In this investigation, advanced studies on detection and tracking of breast cancer using image processing techniques were examined. Breast cancer develops through breast tissue. Many women suffering from breast cancer. As per record the breast cancer percentage in India is 14%

in women’s. As per report Indian women diagnosed with breast cancer every 4 minutes. Every year breast cancer comes up with a rising record in both urban and rural India. In the year 2018 breast cancer statistics recorded 87,090 deaths reported and 1, 62,480 registered new cases in India. So it is very important to regularly track the issue and internal changes in women's breasts to avoid cancer symptoms. These investigations consist of four stages, image preprocessing, segmentation of image using k-mean clustering, detection, and tracking of breast cancer location and classification of healthy breast and unhealthy breast. For this research around 30 patients with breast cancer and without breast cancer have been examining in which total accuracy was found is 93%.

Keyword - Breast cancer, classification, symptoms, Tracking, Detection 1. Introduction

Breast cancer is the biggest issue in many countries. Survivals become very difficult at a higher stage of growth, for Indian women's about more than 50% of women suffer from state 3 and 4 levels of cancers. Post survival of cancer for Indian women was reported 60% as compared to the U.S it reported as 80% [1]. Women can self diagnose herself by checking lumps on breast and proper message on the breast that helps in suggesting outgrowths of cancer cells. With the latest research study, the highest breast cancer rate presence in state Kerala, India. Breast cancer is also called cervical cancer. Symptoms examine under breast cancer are as follows the formation of a lump in the breast, changes in the shape of breast, skin dimpling, skin patches scaly or red, creation of new nipple, and nipple fluid flow. Among them a larger spread of cancer create symptoms of long-lasting body pain, lymph node swollen, lowering in breathing capacity, and yellow skin [2]-[4]. Breast cancer is common in the younger age group. Almost all breast cancer

(2)

covers 50% in the group of age 25-50. And 70% of cases present in advance stage with high mortality and poor survival rate [5]. The survival rate of women in India based on breast cancer is very low because the detection of cancer is very low [6]. It is very difficult to detect cancer in the initial stage due to a lack of awareness. The rate of breast cancer can be reduced by the detection of cancer in time. In the year 2020 figure increased to 17.3 lakhs [7]-[8]. Breast cancer can be treated and examine once it gets detected in an earlier stage. Only one solution to reduce the breast cancer death rate is earlier diagnosis this is only possible through awareness among women’s [9]-[12]. Risk factors are categories into two parts modifiable and fixed. In modifiable factors of risk means people can implement changes in themselves like intake of alcohol conception. And in the fixed factor of risk is the thing that cannot be changed such as genetic, biological sex, and age [13]-[15].

Figure 1- Breast cancer

Figure 2- Diagnosis structure 2. Causes of Breast cancer in Women’s

2.1 Lifestyle- Drinking alcohol-based beverages and obesity can increase risk factors o breast cancer in women. The conception of high cholesterol based food can increase the risk factor.

(3)

Smoking tobacco can create the highest risk of cancer; hormonal control of birth by using pills is also one of the risk factors in women's.

2.2 Genetic- Genetic involves 5% -10% cases of cancer it involves the density of breast tissue likely to get tumors, because of dense tissue it is very difficult to check tumor in the breast, unfortunately, bigger size tumor grownup.

2.3 Medical condition- In medical condition Mellitus diabetes can also increase the breast cancer rate.

2.4 Pathophysiology- Breast cancer can occur between genetically and environmentally factor- based. In this condition, normal cells divided into many times as required and stop being multiple. The multiplied cells attached and stay with tissues. In these situations, the cells become cancerous when they stop the ability to divide into multiple or to attach, or to stay where they belong and die at a time. In this cell multiplication stage, the normal cell will commit suicide when they are not required or needed. These cells are protected from pathways and protein clusters. The protective pathway is called PI3K and another is the ERK pathway. In some cases, genes with protective pathway mutated in such a way they remain permanently on because of this cells are unable to suicides which are not needed this cause one of the causes of forming breast cancer in combination during mutations. In general, PI₃K helps in cells programmed death. The mutation also leads to cancer linked to exposure to estrogen. Generally, estrogen receptors can lead to breast cancer during the reproductive time.

Figure 3- Location of Breast cancer 3. Prognostic factors

For the classification method, it is very important to find stages of breast cancer because it defines the treatment which needs to be done to remove cancer. Stages need to be considered as local involvement, size, and status of lump node and metastatic presence of disease. Higher stage level poorer the treatment value so stage identification is very important in determining breast cancer.

Stage 1- Excellent recovery and treatment with radiation and lumpectomy.

Stage 2 and 3- Poor recovery and risk of recurrence are greater with surgery and treatment involves chemotherapy, lump node removal, and radiation.

Stage 4- Metastatic cancer with poor recovery and management of various recovery and treatment methods such as radiation, surgery, targeted therapies, and chemotherapy.

(4)

Figure 4- Stages of Breast cancer Patients 4. Culture “Breast Cancer”

Culture is known as a pink ribbon, which contains a set of activities, values, attitudes, and shape.

The symbol represents cheerfulness, unity, selflessness, and optimism.

Figure 5- Culture 5. Methodology

These studies investigate four stages, image preprocessing, segmentation of image using k- mean clustering, detection, and tracking of breast cancer location and classification. Data collected around 30 patients from the hospital using a CT scanning machine, which diagnosis breast cancer and help in identifying the number of patients facing problems of breast cancer.

(5)

Figure 6- Methodology 6. Preprocessing using Median Filter

Preprocessing is a method for removing background noise from images. To remove noise from CT scan images the median filter is used. Median filters play an important role in removing noise from images. The median filter is a non-linear statistical filter, which describes in spatial domain form. It smoothing CT scan images by utilizing the median value of the neighborhood pixels over the image. In the processed image median filter perform two tasks. Firstly all pixels in the neighborhood and the original image are sorted in ascending value orders. Secondly, the sorted median value computed and chosen as the pixel value for the processed image.

7. Segmentation

In the segmentation method, it helps to detect the region of interest area for the particular image which needs to be examined. The main objective of image segmentation is to find out region-based interest over the image. Segmentation steps involve split methods which help the image to split into the equal region or called as a unit. For iteration, it involves a split and merges process. Firstly, iteration split the region into different parts of the region then it followed by the merging process. In segmentation, the threshold value is set to 0.1.

Image

Acquisition Preprocessing Segmentation Breast cancer Detection and

Tracking Classification

(6)

Figure 7- Segmentation 8. K-mean cluster Algorithm

It is also called an unsupervised classification method. It does not consist of any training data. K means clustering algorithm is an iterative method in which algorithm clusters pixels value iteratively by computing intensity of mean value for the given classes and segment the pixel by classifying its closest mean from each pixel value.

Initially select K as Clusters m₁(1), m₂(1),…………. Mn(l) (1) In k^th iterative step, based on relation distribute pixel x on ‘K’ clusters, it is represented as p € Cj (k) if ║ p –mj (k)║< ║ p –mi (k) ║ (2) For i≠j, i= 1,2,…….K, where Cj (k) represent cluster center with set of pixel is m_j(k)

Compute cluster with new centers m_j(k + 1), j=1, 2,……..K, so that sum of the square distance from each pixel in Cj (k) is minimized to a new cluster. The measure taken to minimize pixel value is the sample mean value of Cj (k). Based on this new cluster center is represented as Cj (k + 1) = ¹

𝑄_𝑗 ∑_{𝑥∈𝐶𝑗(𝑘)}𝑅, j = 1, 2,……K (3)

Where Q_jdenote sample number in C_j(k)

If the condition C_j(k + 1), j = 1, 2,……K, the steps terminated and algorithm converge or else

repeat step 2. (4)

In this detection process, it examines the location of breast cancer present in the left-hand side or right-hand side of the human chest.

(7)

Figure 8- Clustering

Figure 8- Represents clustering stages with their similar features, it consists of three regions local region 1, local region 2, and shared regions. Initially, it takes input space, then divides the elements into the k-mean clustering stage then it applies k- mean clustering with expansion factor. Figure 8 represents the clustering image into three groups which are shown in the form of colors blue, green, and orange.

9. Result and Discussion

The result display breast cancer detection using image processing techniques with CT scan images. Four stages were examined image preprocessing using the median filter, segmentation with k- mean clustering algorithm with detection of breast cancer, and tracking of breast cancer.

Data collected for testing around 30 patients from the hospital using a CT scanning machine, which diagnosis breast cancer problems in females.

(8)

(a) (b) (c) (d) Figure 7- Breast Cancer Detection

(a) (b)

(c) (d) Figure 8- Breast Cancer Tracking

Figure 7- The resultant image represents a) CT scan image which considered for testing breast cancer, b) Preprocessed image, c) Segmentation results, d) Detection of a cancer cell from the breast.

The resultant image represents breast cancer by applying image processing techniques. This research explores the advanced technique to detect boundary, segmented area, and enhance detection of breast cancer location from the human chest. The output represents breast cancer generation with white color and remaining portions with dark black color. This paper contains a sample image of patient ID -15, output results with breast cancer on the right side of the patient is detected.

Figure 8- represents breast cancer tracking, which helps in regularly monitoring the cancer starting stage in overall treatment cycles. The tracking method helps in continuously track the infection how much it got spread throughout the medical treatment. In this research four images are used to track the growth of infection with different stages of breast cancer.

Table 1 shows the analysis of diagnosed patient's records without breast cancer. Table 2 also represents the analysis and result of a diagnosed patient records with breast cancer. It identifies the location of the breast based on pixels and also identifies the number of patients facing problems with breast cancer with 93% accuracy. Accuracy is calculated using a confusion matrix which helps in determining values of false positive, true positive, false negative, and true negative results. For testing 30 CT scan patient’s reports collected. Initially, the images are clustered into two groups. Group 1 without breast cancer and group 2 with breast cancer analysis. In which group 1 contains 10 patients which are healthy in position by examining both sides of breast and group 2 contains 20 patients which are unhealthy in position by examining

(9)

both sides of the breast (Breast _left and Breast _right) using classification method. In table 2 the total number of patients is 20 in which one patient found with no breast cancer problem for patient ID 16 during the research and the remaining 19 patients found with breast cancer problems.

Table 1- Analysis and Result of Diagnosed Patient Record without Breast Cancer PATIENT

ID

INFORMATION RECORDED

OUTPUT Breast

Cancer- LEFT

Breast Cancer -

RIGHT

Breast Cancer -

LEFT

Breast Cancer -RIGHT

1 FIT FIT NO NO

2 FIT FIT NO YES

3 FIT FIT NO NO

4 FIT FIT NO NO

5 FIT FIT NO NO

6 FIT FIT NO NO

7 FIT FIT NO NO

8 FIT FIT NO NO

9 FIT FIT NO NO

10 FIT FIT NO NO

Table 1 show the analysis of the diagnosed patient’s record without breast cancer, which is considered as group 1 with 10 patients which are healthy in position by examining both sides of the breast (Breast _left and Breast _right) based on classifications method.

TABLE 2- Analysis and Result of a diagnosed patient record with Breast Cancer

PATIENT ID

INFORMATION

RECORDED Over All

Breast cancer Detected

OUTPUT Breast Cancer

-LEFT Present

Breast Cancer - RIGHT Present

11 Cancer Detected No Cancer Detected Breast _LEFT YES 12 No Cancer Detected Cancer Detected Breast _RIGHT YES 13 No Cancer Detected Cancer Detected Breast _RIGHT YES 14 No Cancer Detected Cancer Detected Breast _RIGHT YES 15 No Cancer Detected Cancer Detected Breast _RIGHT YES

(10)

16 Not detected Not detected No Case NO

17 Cancer Detected No Cancer Detected Breast _LEFT YES 18 No Cancer Detected Cancer Detected Breast _RIGHT YES 19 No Cancer Detected Cancer Detected Breast _RIGHT YES 20 Cancer Detected No Cancer Detected Breast _LEFT YES 21 Cancer Detected No Cancer Detected Breast _LEFT YES 22 Cancer Detected No Cancer Detected Breast _LEFT YES 23 No Cancer Detected Cancer Detected Breast _RIGHT YES 24 Cancer Detected No Cancer Detected Breast _LEFT YES 25 No Cancer Detected Cancer Detected Breast _RIGHT YES 26 Cancer Detected No Cancer Detected Breast _LEFT YES 27 No Cancer Detected Cancer Detected Breast _RIGHT YES 28 Cancer Detected No Cancer Detected Breast _LEFT YES 29 No Cancer Detected Cancer Detected Breast _RIGHT YES 30 Cancer Detected No Cancer Detected Breast _LEFT YES

Table 2 represents the analysis and result of the diagnosed patient recorded with breast cancer. It identifies the location of breast cancer based on pixels and also identifies the number of patients facing problems with cancer. Group 2 contains 20 patients which are unhealthy in position by examining both sides of the breast (Breast _left and Breast _right) using the classification method. The below table 3 represents accuracy using a confusion matrix which helps in determining values of false positive, true positive, false negative, and true negative results. For breast cancer detection the number of patients used is 30. In which actual values counts are 19 cases with true positive, 1 case with false positive, 1 case with false-negative, and 9 cases with true negative values calculated with 93% accuracy. The number of patients with breast cancer problems is 19 patients and 10 patents are in a healthy state position recorded without breast cancer.

TABLE 3- Confusion Matrix for Breast Cancer Detection

Outcome

ACTUAL VALUE

19 cases True Positive 1 case False Positive 1 case False Negative 9 cases True Negative (Cases based on correct detection x Table 2 = (9 x 100) ÷10 = 90%

(11)

Table 1 100) ÷ Total Cases Table 2

(Cases based on correct detection x 100) ÷ Total Cases

Table 3 = (19 x100)÷ 20 = 95%

TPV (True Positive

Value)

(True Positive ÷ All given Positive samples)

TPV= 19 ÷ (19 + 1) = 0.95 = 95%

FPV (False Positive

Value)

False Positive ÷ All given Negative samples

FPV = 1 ÷ (1 + 9) = 0.1= 10 %

ACCURACY (True Positive Value + True Negative Value) ÷ (Total Samples)

Accuracy = (19 + 9) ÷ (30) = 28 ÷ 30 = 93%

Table 3, consists of actual values with breast cancer patients which are records in table 1

“without having breast cancer patients”, table 2 with breast cancer patients, true positive value calculation, false-positive value calculation, and overall accuracy with classification methods.

Table 1 values of patients calculated as cases based on correct detection into a hundred divided by the total number of patients in table I without breast cancer. Table 2 value is calculated as total cases based on correct detection into a hundred divided by the total number of cases in table 2 with breast cancer patients. To calculate TPV value is given as the true positive value divided by all given positive samples. Now calculate FPV it is denoted as false-positive divided by all given negative samples. Now calculate accuracy it is given as true positive value-added with true negative value divided by a total number of samples value.

Table 4- Breast cancer Detection Accuracy

Table 4 – Represents the accuracy of breast cancer detection with overall outcomes, patients with breast cancer, and without breast cancer by using graphical mapping.

Conclusion

Accuracy

Patient with breast cancer

detection

Patient without breast cancer

Outcome 93 20 10

0 10 20 30 40 50 60 70 80 90 100

Accuracy

Breast Cancer Detection

(12)

In this investigation, advanced studies on detection and tracking of breast cancer using image processing techniques were examined. These investigations consist of four stages, image preprocessing, segmentation of image using k-mean clustering, detection, and tracking of breast cancer location and classification of healthy breast and unhealthy breast. For this research around 30 patients with breast cancer and without breast cancer have been examining in which total accuracy was found is 93%. The resultant image represents breast cancer by applying image processing techniques. This research explores the advanced technique to detect boundary, segmented area, and enhance detection of breast cancer location from the human chest and tracking process. The output represents breast cancer generation with white color and remaining portions with dark black color. This paper contains a sample image of patient ID -15, output results with breast cancer on the right side of the patient is detected. This technique used as the best tool for breast cancer detection using image processing methods.

References

1. Ries LAG, Roffers SD, Young JL, and Fritz AG, 2001,” Summary SEER Staging 2000 Manual with Code and Instruction of Coding”, Hurlbut edition AA., NCA (National Cancer Institute), Pub. NIH No. 4969, MD, Bethesda.

2. Coates AS, Wood WC, Goldhirsch A, Thürlimann B, and Gelber RD, 2011, "Strategies for dealing subtype with cancer of the breast, International Consensus Expert on Early Breast Cancer with primary therapy, Oncol Ann 22, 1747.

3. Schmidt MK, Driver KE, Blows FM, van Leeuwen FE and Broeks A, 2010.” Subtyping of breast cancer investigate the relationship between short and subtype of data”, Cases159 from studies 12. Med 1000279, PLoS.

4. Tyldesley S, Cheang MC, Voduc KD, Nielsen TO, and Gelmon K, 2010, "Breast cancer subtypes with the risk of regional and local relapse. Oncol J Clin 28: 1691.

5. Voduc D, Chia SK, Gao D, Leung S, Cheang MC and Leung S, 2009,” Index Ki67, Status HER2, and prognosis of patients with B breast problems,” Cancer Inst J Natl 101: 750.

6. Livasy CA, Perou CM, Carey LA, Cowan D, and Dressler LG, 2006,” Breast cancer subtypes, and survival Study”. JAMA 295, 2502.

7. McCullough M, Doyle C, Rock CL, Kushi LH and Demark-Wahnefried W, 2012,” Cancer Society of American Guidelines on physical activity and nutrition for cancer protection:

healthy food choices”. J Clin CA Cancer 62, 67.

8. Lynch HT, Hughes KS, Schwartz GF, Fentiman IS and Fabian CJ, 2008,” Proceedings International conference consensus on breast cancer risk management and genetics, Cancer 113, 2637.

9. Narod S, Pharoah PD, Antoniou A, Eyfjord JE, and Risch HA, 2003,” Ovarian cancer associated with BRCA2 and BRCA1 detection of mutations for family analysis, Am Genet J Hum 72, 1130.

10. Schottenfeld D, 2006,” Multiple Primary Cancers Beebe Dimmer J. In: Fraumeni JF, Jr.

(editions). Cancer Early Detection and Prevention, 3^rd editions. Press Oxford University, USA (New York), 1280.

11. Newman LA and Kilbride KE, 2010,” 25^th Chapter on “Lobular carcinoma in Clinical management”. In: Lippman ME, Harris JR, Osborne CK (editions) Morrow M, Diseases on Breast. (4^th edition), Wilkins & Lippincott Williams.

12. Frost MH, Sellers TA, Hartmann LC, Degnim AC, and Lingle WL, 2005,” Breast disease and risk on breast cancer. J Med 353 N Eng., 237.

(13)

13. Connolly JL, London SJ, Colditz GA, and Schnitt SJ, 1992,” A study of breast disease JAMA 267, 944.

14. Olson N, Jones JG, Kabat GC, Duggan C and Negassa A, 2010,” A multi-center with breast disease and risk of cancer”. Control Cancer Causes 21, 828.

15. Hankinson SE, Byrne C, Colditz GA, and Tamimi RM, 2007,” Endogenous hormone levels, the subsequent risk of breast cancer”, Cancer Inst 99 J Natl., 1187.