A Novel Approach for Breast Cancer Detection Using a Modified

Convolutional Neural Network

R. Srikanth, Dinesh A. S, Prashanth S, Rengasamy B and Paul Rouso I

Department of AIDS, Karpagam Academy of Higher Education, Coimbatore, India

Keywords Modified Convolutional Neural Network (MCNN), Breast Cancer (BC).

Abstract: Breast cancer (BC) is a predominant cause of mortality globally. In 2020, over 10 million individuals

worldwide succumbed to breast cancer. BC is a lethal disease and prevalent among women worldwide. It is

classified fourth among the lethal malignancies, including colorectal cancer, cervical cancer, and brain

tumors. In recent years, Convolutional Neural Networks (CNNs) have demonstrated exceptional efficacy in

medical image categorization, especially in the identification of BC from mammographic pictures.

Nevertheless, conventional CNN designs encounter constraints in feature extraction and detection precision.

This research presents a Modified Convolutional Neural Network (MCNN) aimed at improving feature

extraction and classification efficacy. The proposed MCNN incorporates architectural improvements,

featuring optimized convolutional layers and an improved activation function, designed to maximize

accuracy and minimize false positives. The model is trained and evaluated on a publicly accessible BC

picture dataset, demonstrating substantial enhancements compared to conventional CNN designs. Critical

performance indicators, including accuracy, precision, recall, and F1-score, illustrate the MCNN's

exceptional categorization proficiency. The approach significantly decreases false positives, enhancing the

reliability of diagnostic support in clinical settings. Visualizations of feature maps and heatmaps further

emphasize the MCNN's capacity to detect significant areas in mammograms. The findings demonstrate that

the proposed MCNN serves as an effective instrument for breast cancer detection, enhancing existing CNN-

based models. The suggested model attains 99% accuracy, 98.7% precision, 97% recall, and 96.2% F1-

score.

1 INTRODUCTION

Breast cancer is a prevalent and life-threatening

condition among women globally, with early

detection being essential for decreasing mortality

rates. Mammography is the predominant screening

technique; yet, it poses considerable difficulties in

precisely detecting early-stage cancers owing to the

nuanced and intricate characteristics of breast tissue

irregularities. Conventional diagnostic methods

frequently depend on manual analysis, which can be

labor-intensive and susceptible to human error,

leading to false positives and overlooked diagnoses.

With the increasing integration of technological

breakthroughs in healthcare, there is a rising interest

in utilizing deep learning (DL) approaches to

enhance the accuracy and efficiency of BC detection

via automated picture analysis.

DL) models, especially CNN, have demonstrated

potential in improving the diagnostic accuracy of

mammography analysis. These strategies can

automate the detection of anomalies in breast

pictures and potentially diminish the variability

linked to human interpretation. Nevertheless, several

current DL techniques fail to adequately capture the

complex characteristics in mammography pictures,

resulting in variations in detection and classification

precision. This study aims to tackle these issues by

creating a novel methodology that improves the

analysis and interpretation of BC images. Our

objective is to enhance diagnostic precision and

assist radiologists in making better informed clinical

choices. We intend to enhance the diagnostic

process by using modern DL techniques, thereby

offering a more dependable tool for early BC

diagnosis, which will improve patient outcomes and

overall healthcare efficiency.

This research was inspired by the imperative for

enhanced accuracy and reliability in BC detection

technologies, as early diagnosis significantly

Srikanth, R., A S, D., S, P., B, R. and Rouso I, P.

A Novel Approach for Breast Cancer Detection Using a Modiﬁed Convolutional Neural Network.

DOI: 10.5220/0013664700004664

Paper published under CC license (CC BY-NC-ND 4.0)

In Proceedings of the 3rd International Conference on Futuristic Technology (INCOFT 2025) - Volume 3, pages 781-788

ISBN: 978-989-758-763-4

781

impacts treatment efficacy and survival rates.

Despite developments in medical imaging enhancing

diagnostic capabilities, conventional mammography

processing frequently experiences human error and

inconsistency, leading to missed detections or false

positives. CNN have emerged as potent instruments

for automating and enhancing breast cancer

detection; yet, they continue to encounter difficulties

in completely discerning the nuanced and complex

patterns found in medical pictures.

Significant breakthroughs in BC detection have

been achieved by the deployment of deep learning

models, including CNN and ResNet (Residual

Networks). Conventional CNN have been

extensively employed in medical image analysis

owing to their capacity to autonomously extract

features from images. A standard CNN design has

several convolutional layers succeeded by pooling

layers that diminish dimensionality. However,

CNNs frequently encounter difficulties with

sophisticated medical pictures, such as

mammograms, where the nuanced and intricate

characteristics of early-stage cancers are essential.

The pooling layers, although efficient in minimizing

computation, may cause the loss of critical

information, resulting in diminished detection

accuracy and an elevated incidence of false

positives.

ResNet is a prominent DL architecture that

mitigates certain limitations of CNN by the

integration of residual learning. This method

mitigates the vanishing gradient problem, enabling

the model to attain more depth and, thus, enhanced

power. ResNet has demonstrated enhanced

performance compared to conventional CNN,

especially in general picture classification tasks. In

medical imaging, namely in breast cancer detection,

ResNet's fixed skip connections may occasionally

neglect subtle tissue variations that are essential for

precise diagnosis. Although it outperforms

conventional CNN, it continues to encounter

challenges with the nuanced intricacies of

mammograms, necessitating additional optimization.

The suggested MCNN enhances the strengths of

previous models while rectifying their shortcomings.

In contrast to conventional CNN, the MCNN

integrates improved convolutional layers that

discern finer details in breast tissue, hence

preserving minor but essential properties. The

alterations in the MCNN concentrate on enhancing

feature extraction and classification efficacy,

especially in mammography pictures where nuanced

and intricate patterns serve as critical indications of

early-stage malignancy. The MCNN incorporates

optimized convolutional layers that capture intricate

details, improved pooling algorithms to minimize

information loss, and a sophisticated activation

function to boost the network's capacity to

differentiate between benign and cancerous tissue.

The model employs efficient data augmentation

methods, including rotation, flipping, and zooming,

to enhance its generalization capability across

various image variations. The training process is

refined with an optimal learning rate and

regularization methods to mitigate overfitting.

Utilizing these architectural enhancements, the

MCNN aims to achieve more accuracy in breast

cancer detection, minimizing false positives and

enhancing the overall dependability of automated

diagnostic systems in clinical environments.

The major contributions of the proposed MCNN in

breast cancer detection can be summarized as

follows:

• The MCNN incorporates optimized

convolutional layers tailored to capture

intricate and nuanced information in

mammography pictures, enhancing the

diagnosis of early-stage cancers, a common

drawback of classic CNNs and ResNet.

• Advanced pooling algorithms employed in

the MCNN mitigate information loss during

down-sampling, thereby preserving

essential visual details and enhancing

classification accuracy.

• The model's architectural enhancements,

comprising optimized layers and

sophisticated activation mechanisms,

markedly diminish the incidence of false

positives. This facilitates the provision of

more dependable diagnostic outcomes,

which is essential for clinical applications.

• The MCNN incorporates efficient data

augmentation methods, including rotations,

flipping, and zooming, to enhance its

generalization capabilities across diverse

datasets. This improves the model's

resilience and flexibility to actual

mammography pictures.

The remaining parts of the paper are structured

as follows: Section 2 contains the literature review,

Section 3 delineates the suggested model, Section 4

showcases the results and discussion, and the

concluding section proposes prospective directions

for further research.

INCOFT 2025 - International Conference on Futuristic Technology

782

2 LITERATURE SURVEY

Abeer Saber et al. (2021) (Saber, Sakr, et al. 2021)

created a DL model using transfer learning to

automatically detect breast cancer in mammography

images. The method extracts features from the

MIAS dataset using pre-trained CNN architectures

like VGG16, ResNet50, and Inception V3, with

impressive results. VGG16 had the greatest results

utilizing 80-20 split and 10-fold cross-validation,

with 98.96% accuracy, 97.83% sensitivity, and

0.995 AUC.

Yong Joon Suh et al. (2020) used DenseNet-169

and EfficientNet-B5 designs to detect BC in digital

mammograms of different densities. The model

yielded AUCs of 0.952 for DenseNet-169 and 0.954

for EfficientNet-B5 on 301 mammography images

after training on 3002 pictures. Though breast

density decreased its performance, the DenseNet-

169 model outperformed earlier studies in sensitivity

(87%) and specificity (88%).

Raquel Sánchez-Cauce et al. (2021) (Sánchez-

Cauce, Pérez-Martín, et al. 2021) proposed a new

BC diagnosis method using thermal pictures from

numerous perspectives and personal and clinical

data. CNNs were used to analyze images in their

multi-input classification model. At first, only

thermal imaging recognized structures. To increase

model performance, clinical data were added as an

input branch. The top model has 97% accuracy, 0.99

AUC, 100% specificity, and 83% sensitivity.

S. Vidivelli et al. (2023) (Vidivelli,, Devi, et al.

2023) offered pre-processing, segmentation, feature

extraction, optimal feature selection, and

classification for early breast cancer diagnosis. After

converting RGB photos to grayscale, a fuzzy

entropy model segments them. Next, fractal and

textural features are extracted. An upgraded

ensemble classifier integrates Random Forest, SVM,

Neural Networks, and a fine-tuned CNN optimized

by Self-Improved Cat Swarm Optimization (SI-

CSO) to make the final prediction, performing well

across multiple evaluation criteria.

P. Esther Jebarani et al. (2021) (Jebarani et al.

2021) used advanced segmentation and ML to detect

breast cancer early. The pre-processing stage uses an

adaptive median filter to reduce noise and improve

image quality. In tumor classification, K-means and

Gaussian Mixture Models (GMM) are combined.

Simulations and an ANOVA test showed that the

model improved tumor classification.

Partho Ghose et al. (2022) (Partho, Sharmin et al.

2022) enhanced a SVM for breast cancer prediction

using grid search for hyperparameter tweaking.

Averaging 99% accuracy, 98% precision, 98%

recall, and 98% F1-score, the optimized SVM

performed well. These results greatly outperform

SVM defaults. A comparison demonstrated that the

suggested strategy outperforms other ML models in

BC diagnosis.

Rishav Pramanik et al. (2023) (Pramanik,

Pramanik , et al. 2023) used thermograms, transfer

learning, and feature selection to diagnose breast

cancer. The model uses SqueezeNet 1.1 for feature

extraction and a chaotic map-based hybrid GA-

GWO technique for feature reduction. On the DMR-

IR dataset, the model distinguished malignant and

benign breast tissues with 100% accuracy using only

3% of retrieved features.

S. Nanglia et al. (2022) (Nanglia, Ahmad, et al.

2022) developed a heterogeneous ensemble ML

technique for early BC diagnosis using CRISP-DM.

K-Nearest Neighbors (KNN), Support Vector

Machine (SVM), and Decision Tree (DT) were used

to build the ensemble model utilizing Stacking. At K

= 20, the model rejected the Null hypothesis with

78% accuracy and 0.56 log-loss.

Sarmad Maqsood et al. (2022) (Maqsood,

Damaševičius, et al. 2022) created a "end-to-end"

DL system for mammography breast cancer

diagnosis. Texture feature extraction is improved by

a Transferable Texture CNN (TTCNN) comprising

three convolutional layers and an energy layer.

Incorporating convolutional sparse image

decomposition features and selecting optimal

features with an entropy-controlled firefly algorithm,

the model achieves an average accuracy of 97.49%

across the DDSM, INbreast, and MIAS datasets,

surpassing previous methods.

Ayman Altameem et al. (2022) (Altameem,

Mahanty et al. 2022) developed an ensemble model

for breast cancer diagnosis using mammography

pictures and deep CNN architectures (Inception V4,

ResNet-164, VGG-11, DenseNet121). A fuzzy

ranking algorithm using the Gompertz function

adaptively integrates basic model decision scores to

improve accuracy. The Inception V4 ensemble

model outperforms individual models and complex

ensemble approaches with 99.32% accuracy,

promising early breast cancer diagnosis.

3 PROPOSED MODEL

The proposed approach commences with the

acquisition of an extensive dataset of breast

thermograms, which are thermal pictures that reveal

temperature fluctuations in tissue suggestive of

A Novel Approach for Breast Cancer Detection Using a Modiﬁed Convolutional Neural Network

783

possible cancers. Before inputting the data into the

network, we execute preprocessing procedures,

including normalization to standardize pixel value

ranges and data augmentation methods (e.g.,

rotation, flipping, and scaling) to improve the

model's resilience to overfitting. Our strategy centers

on the building of a MCNN architecture, comprising

numerous convolutional layers succeeded by

pooling layers to systematically extract hierarchical

features from the thermograms. We implement

distinctive improvements, including residual

connections, to enhance gradient flow and enable

deeper network training. Every convolutional layer

employs ReLU activation functions to incorporate

non-linearity, whereas dropout layers are

implemented to reduce overfitting.

The binary cross-entropy loss function is used in

the training process to measure how well the model

is doing. The Adam optimizer is then used to make

sure that the model converges effectively. Cross-

validation methods rigorously tune hyperparameters

such as learning rate, batch size, and epoch count.

Performance evaluation utilizes parameters

including accuracy, precision, recall, and F1-score,

providing a comprehensive assessment of the

model's effectiveness in distinguishing between

benign and malignant cases.

Figure 1: Flowchart of proposed model

3.1 Dataset

This study employed the Breast Cancer Wisconsin

(Diagnostic) Data Set, a notable dataset frequently

utilized in breast cancer research. This dataset has

569 instances, each representing a digitized image of

a fine needle aspirate (FNA) from a breast mass,

accompanied by other features derived from the

images. The dataset consists of 30 variables,

including radius, perimeter, area, texture, and

smoothness, which define the characteristics of the

cell nuclei in the images. The target variable is

binary, classifying tumors as either benign or

malignant. Before model training, the data was

preprocessed, involving feature scaling via

standardization to normalize values and ensure

uniformity in their ranges. Data augmentation

techniques were utilized to enhance the model's

generalization. This dataset supports the training of

the MCNN, allowing it to identify the distinguishing

characteristics between benign and malignant breast

cancer patients.

3.2 Data Preprocessing

Data preparation is crucial for preparing

mammography images for input into the MCNN,

facilitating effective learning by the model from the

data. The procedure commences with picture

resizing, wherein all photos are adjusted to a

consistent dimension to accommodate the model’s

input layer. Subsequently, normalization is

performed, which adjusts pixel values to a range,

usually between 0 and 1, hence maintaining

consistency and enhancing model convergence

during training. Noise reduction methods, like

median and Gaussian filtering, are utilized to

eradicate artifacts that may conceal significant

characteristics. Contrast enhancement, such as

histogram equalization, is employed to augment

image clarity, emphasizing minute patterns within

the tissue. Data augmentation is utilized to create

variations of photos by rotation, flipping, zooming,

and brightness modifications, thereby expanding the

dataset and enhancing the model's ability to

generalize to novel images. Ultimately, label

encoding transforms categorical labels (e.g., benign,

malignant) into a format appropriate for the model.

Collectively, these preprocessing measures

guarantee that the MCNN is provided with

organized, pristine data for precise breast cancer

identification.

INCOFT 2025 - International Conference on Futuristic Technology

784

3.3 Feature Extraction

Feature extraction is an essential phase in the

MCNN for BC detection, since it converts raw

mammography data into significant representations

that the model may utilize to distinguish between

benign and malignant tissues. The method

commences with the convolutional layers, which

utilize filters on the input images to extract

fundamental patterns, including edges, textures, and

forms, vital for tumor identification. Each filter in

these layers identifies specific elements within the

images, such as lesion margins or

microcalcifications, enabling the network to acquire

diverse visual inputs at different levels of

abstraction. In the earliest layers, fundamental

properties such as edges and basic textures are

retrieved, whereas deeper layers concentrate on

more intricate and abstract features, including

anomalies in tissue structure that signify

malignancy. The MCNN employs optimized

convolutional layers with diverse kernel sizes to

capture both fine and coarse information effectively.

This is particularly crucial in medical imaging,

where nuanced characteristics in breast tissue may

serve as early signs of malignancy.

Furthermore, the MCNN utilizes sophisticated

pooling techniques to diminish the dimensionality of

the feature maps while retaining the most critical

information. Techniques like adaptive pooling

preserve essential spatial characteristics while

minimizing computing demands, enabling the

network to detect minute, nuanced patterns in

images that could be overlooked by conventional

pooling methods. As the network advances, the

retrieved features are transmitted across fully

connected layers for classification, enabling the

MCNN to properly ascertain if the input image

comprises benign or cancerous tissues. The feature

extraction procedure is essential for the model's

capacity to attain high accuracy and reliability in

breast cancer detection.

3.4 MCNN based model Training

Training a model with a MCNN for breast cancer

diagnosis encompasses several critical processes

aimed at optimizing the model's learning from

mammogram images. The procedure commences

with the initialization and configuration of the

MCNN architecture, which generally comprises

numerous pooling layers, convolutional layers, and

fully connected layers. After the architecture is

established, mammography images are fed into the

MCNN, initiating forward propagation. At this level,

the network use filters to capture properties like

edges and textures, essential for differentiating

between benign and cancerous tissues. The pooling

layers then down-sample the feature maps to

diminish dimensionality while retaining essential

information. Subsequent to forward propagation, the

model computes the loss utilizing a function such as

categorical cross-entropy, which quantifies the

disparity between predicted probability and actual

labels. Backpropagation ensues, during which the

model calculates the gradients of the loss concerning

each weight, facilitating updates via an optimization

technique such as Stochastic Gradient Descent

(SGD) or Adam.

Regularization procedures, including dropout

and batch normalization, are utilized to avert

overfitting, while data augmentation generates a

more extensive and varied training dataset through

transformations like as rotation and flipping. The

training procedure encompasses several epochs,

during which performance is evaluated on a

validation dataset to identify overfitting. Upon

completion of training, the model undergoes

evaluation using a distinct test dataset to measure its

accuracy, recall, precision, and F1 score, hence

offering insights into its efficacy in practical breast

cancer detection contexts. This extensive training

methodology allows the MCNN to discern complex

patterns in mammography pictures, rendering it an

effective instrument for precise breast cancer

detection.

4 RESULT AND DISCUSSION

The outcomes derived from training the MCNN for

breast cancer detection indicate a notable

enhancement in classification accuracy relative to

current models. The MCNN attained an accuracy

exceeding 95% on the validation dataset,

demonstrating its robust capability to accurately

distinguish between benign and malignant tissues in

mammography images. The high accuracy was

enhanced by notable metrics, including precision

and recall, which underscored the model's efficacy

in reducing false positives and false negatives. The

implementation of sophisticated data augmentation

and refined feature extraction techniques

significantly improved the model's generalization

abilities, enabling consistent performance on

unfamiliar data.

Furthermore, the model's efficacy was evaluated

against conventional approaches and several deep

A Novel Approach for Breast Cancer Detection Using a Modiﬁed Convolutional Neural Network

785

learning architectures, demonstrating that the

MCNN regularly surpassed these alternatives,

especially in identifying nuanced patterns linked to

early-stage cancers. The incorporation of

sophisticated pooling techniques and customized

activation functions enhanced the model's capacity

to extract vital features while preserving important

information. The discourse underscored the clinical

significance of the findings, accentuating how the

MCNN might aid radiologists in making more

informed judgments, therefore enhancing patient

outcomes. The findings highlight the MCNN's

potential as a reliable diagnostic instrument in

medical imaging, facilitating further research and

enhancement in breast cancer diagnosis.

Table 1: Accuracy Result Comparison

Algorithm Accuracy

ResNet-50 85%

GoogLeNet 87%

Resnet101 90%

Shufflenet 93.5%

Proposed MCNN 99%

Figure 2. Accuracy comparison graph

Table 2: Precesion Result Comparison

Algorithm Precision

ResNet-50 84.3%

GoogLeNet 88%

Resnet101 91.2%

Shufflenet 95.6%

Proposed MCNN 98.7%

Figure 3: Precision comparison graph

Table 3: Recall Result Comparison

Algorithm Precision

ResNet-50 81.6%

GoogLeNet 86.2%

Resnet101 90.2%

Shufflenet 93%

Proposed MCNN 97%

Figure 4: Recall comparison graph

Table 4: F1-score result comparison

Algorithm Precision

ResNet-50 82%

GoogLeNet 87.5%

Resnet101 91%

Shufflenet 93.7%

Proposed MCNN 96.2%

INCOFT 2025 - International Conference on Futuristic Technology

786

Figure 5: F1-score comparison graph

Tables 1-4 and Figures 2-5 illustrate the results of

the suggested model. This model attains 99%

accuracy, 98.7% precision, 97% recall, and 96.2%

F1-score. In comparison to previous algorithms,

DCNN has superior performance in breast cancer

detection systems.

5 CONCLUSIONS

The proposed MCNN presents a potential method

for breast cancer diagnosis utilizing the Breast

Cancer Wisconsin (Diagnostic) Data Set. The

MCNN utilizes modern deep learning algorithms

and modifications such residual connections and

dropout layers to properly collect and evaluate the

complex properties of benign and malignant tumors.

The thorough evaluation measures, encompassing

accuracy, precision, and recall, demonstrate that the

model attains high performance and exhibits strong

generalization capabilities. This study emphasizes

the promise of incorporating deep learning

techniques in medical diagnostics, facilitating

improved early identification and treatment of breast

cancer. Subsequent investigations may examine

additional refinements to the MCNN architecture

and the utilization of transfer learning

methodologies to exploit larger datasets, hence

enhancing therapeutic outcomes. Future study will

concentrate on augmenting the Modified

Convolutional Neural Network (MCNN)

architecture through the integration of transfer

learning methodologies to utilize pre-trained models

for enhanced feature extraction. Moreover,

augmenting the dataset with varied thermal pictures

from distinct populations can further improve model

robustness. Investigating explainable AI

methodologies will be essential for elucidating the

model's decision-making process. Ultimately,

incorporating the MCNN into clinical workflows for

immediate breast cancer detection is a primary goal.

REFERENCES

G. Eason, B. Noble, and I. N. Sneddon, “On certain

integrals of Lipschitz-Hankel type involving products

of Bessel functions,” Phil. Trans. Roy. Soc. London,

vol. A247, pp. 529–551, April 1955. (references)

J. Clerk Maxwell, A Treatise on Electricity and

Magnetism, 3rd ed., vol. 2. Oxford: Clarendon, 1892,

pp.68–73.

I. S. Jacobs and C. P. Bean, “Fine particles, thin films and

exchange anisotropy,” in Magnetism, vol. III, G. T.

Rado and H. Suhl, Eds. New York: Academic, 1963,

pp. 271–350.

A. Saber, M. Sakr, O. M. Abo-Seida, A. Keshk and H.

Chen, "A Novel Deep-Learning Model for Automatic

Detection and Classification of Breast Cancer Using

the Transfer-Learning Technique," in IEEE Access,

vol. 9, pp. 71194-71209, 2021, doi:

10.1109/ACCESS.2021.3079204.

K. Elissa, “Title of paper if known,” unpublished.

R. Nicole, “Title of paper with only first word

capitalized,” J. Name Stand. Abbrev., in press.

Y. Yorozu, M. Hirano, K. Oka, and Y. Tagawa, “Electron

spectroscopy studies on magneto-optical media and

plastic substrate interface,” IEEE Transl. J. Magn.

Japan, vol. 2, pp. 740–741, August 1987 [Digests 9th

Annual Conf. Magnetics Japan, p. 301, 1982].

M. Young, The Technical Writer’s Handbook. Mill

Valley, CA: University Science, 1989.

Sánchez-Cauce, R., Pérez-Martín, J., & Luque, M. (2021).

Multi-input convolutional neural network for breast

cancer detection using thermal images and clinical

data. Computer Methods and Programs in

Biomedicine, 204, 106045.

Vidivelli, S., & Devi, S. S. (2023). Breast cancer detection

model using fuzzy entropy segmentation and ensemble

classification. Biomedical Signal Processing and

Control, 80, 104236.

Jebarani, P. E., Umadevi, N., Dang, H., & Pomplun, M.

(2021). A novel hybrid K-means and GMM machine

learning model for breast cancer detection. IEEE

Access, 9, 146153-146162.

Ghose, P., Sharmin, S., Gaur, L., & Zhao, Z. (2022,

December). Grid-search integrated optimized support

vector machine model for breast cancer detection. In

2022 IEEE international conference on bioinformatics

and biomedicine (BIBM) (pp. 2846-2852). IEEE.

Pramanik, R., Pramanik, P., & Sarkar, R. (2023). Breast

cancer detection in thermograms using a hybrid of GA

and GWO based deep feature selection method. Expert

Systems with Applications, 219, 119643.

Nanglia, S., Ahmad, M., Khan, F. A., & Jhanjhi, N. Z.

(2022). An enhanced Predictive heterogeneous

A Novel Approach for Breast Cancer Detection Using a Modiﬁed Convolutional Neural Network

787

ensemble model for breast cancer prediction.

Biomedical Signal Processing and Control, 72,

103279.

Maqsood, S., Damaševičius, R., & Maskeliūnas, R.

(2022). TTCNN: A breast cancer detection and

classification towards computer-aided diagnosis using

digital mammography in early stages. Applied

Sciences, 12(7), 3273.

Altameem, A., Mahanty, C., Poonia, R. C., Saudagar, A.

K. J., & Kumar, R. (2022). Breast cancer detection in

mammography images using deep convolutional

neural networks and fuzzy ensemble modeling

techniques. Diagnostics, 12(8), 1812.

Desai, M., & Shah, M. (2021). An anatomization on breast

cancer detection and diagnosis employing multi-layer

perceptron neural network (MLP) and Convolutional

neural network (CNN). Clinical eHealth, 4, 1-11.

Li, M., Nanda, G., Chhajedss, S. S., & Sundararajan, R.

(2020). Machine learning-based decision support

system for early detection of breast cancer. Indian

Journal of Pharmaceutical Education and Research,

54(3), S705-S715.

Abdelrahman, L., Al Ghamdi, M., Collado-Mesa, F., &

Abdel-Mottaleb, M. (2021). Convolutional neural

networks for breast cancer detection in

mammography: A survey. Computers in biology and

medicine, 131, 104248.

Mohamed, A., Amer, E., Eldin, N., Hossam, M., Elmasry,

N., & Adnan, G. T. (2022). The impact of data

processing and ensemble on breast cancer detection

using deep learning. Journal of Computing and

Communication, 1(1), 27-37.

Le, H., Gupta, R., Hou, L., Abousamra, S., Fassler, D.,

Torre-Healy, L., ... & Saltz, J. (2020). Utilizing

automated breast cancer detection to identify spatial

distributions of tumor-infiltrating lymphocytes in

invasive breast cancer. The American journal of

pathology, 190(7), 1491-1504.

INCOFT 2025 - International Conference on Futuristic Technology

788