BC2NetRF: Breast Cancer Classification from Mammogram Images Using Enhanced Deep Learning Features and Equilibrium-Jaya Controlled Regula Falsi-Based Features Selection

Diagnostics, Journal Year: 2023, Volume and Issue: 13(7), P. 1238 - 1238

Published: March 25, 2023

One of the most frequent cancers in women is breast cancer, and year 2022, approximately 287,850 new cases have been diagnosed. From them, 43,250 died from this cancer. An early diagnosis cancer can help to overcome mortality rate. However, manual using mammogram images not an easy process always requires expert person. Several AI-based techniques suggested literature. still, they are facing several challenges, such as similarities between non-cancer regions, irrelevant feature extraction, weak training models. In work, we proposed a automated computerized framework for classification. The improves contrast novel enhancement technique called haze-reduced local-global. enhanced later employed dataset augmentation. This step aimed at increasing diversity improving capability selected deep learning model. After that, pre-trained model named EfficientNet-b0 was fine-tuned add few layers. trained separately on original transfer concepts with static hyperparameters' initialization. Deep features were extracted average pooling layer next fused serial-based approach. optimized selection algorithm known Equilibrium-Jaya controlled Regula Falsi. Falsi termination function algorithm. finally classified machine classifiers. experimental conducted two publicly available datasets-CBIS-DDSM INbreast. For these datasets, achieved accuracy 95.4% 99.7%. A comparison state-of-the-art (SOTA) technology shows that obtained improved accuracy. Moreover, confidence interval-based analysis consistent results framework.

Language: Английский

---

Advancements and Prospects of Machine Learning in Medical Diagnostics: Unveiling the Future of Diagnostic Precision

Archives of Computational Methods in Engineering, Journal Year: 2024, Volume and Issue: unknown

Published: June 26, 2024

Language: Английский

---

Breast Tumor Classification with Enhanced Transfer Learning Features and Selection Using Chaotic Map-Based Optimization

International Journal of Computational Intelligence Systems, Journal Year: 2024, Volume and Issue: 17(1)

Published: Feb. 1, 2024

Abstract Among women, breast cancer remains one of the most dominant types. In year 2022, around 2,87,800 new cases were diagnosed, and 43,200 women faced mortality due to this disease. Analysis processing mammogram images is vital for its earlier identification thus helps in reducing rates facilitating effective treatment women. Accordingly, several deep-learning techniques have emerged classification. However, it still challenging requires promising solutions. This study proposed a newer automated computer-aided implementation The work starts with enhancing contrast using haze-reduced adaptive technique followed by augmentation. Afterward, EfficientNet-B4 pre-trained architecture trained both original enhanced sets mammograms individually static hyperparameters’ initialization. provides an output 1792 feature vectors each set then fused serial mid-value-based approach. final are optimized chaotic-crow-search optimization algorithm. Finally, obtained significant classified aid machine learning algorithms. evaluation made INbreast CBIS-DDSM databases. framework attained balanced computation time maximum classification performance 98.459 96.175% accuracies on databases, respectively.

Language: Английский

---

A Framework for Lung and Colon Cancer Diagnosis via Lightweight Deep Learning Models and Transformation Methods

Diagnostics, Journal Year: 2022, Volume and Issue: 12(12), P. 2926 - 2926

Published: Nov. 23, 2022

Among the leading causes of mortality and morbidity in people are lung colon cancers. They may develop concurrently organs negatively impact human life. If cancer is not diagnosed its early stages, there a great likelihood that it will spread to two organs. The histopathological detection such malignancies one most crucial components effective treatment. Although process lengthy complex, deep learning (DL) techniques have made feasible complete more quickly accurately, enabling researchers study lot patients short time period for less cost. Earlier studies relied on DL models require computational ability resources. Most them depended individual extract features high dimension or perform diagnoses. However, this study, framework based multiple lightweight proposed utilizes several transformation methods feature reduction provide better representation data. In context, histopathology scans fed into ShuffleNet, MobileNet, SqueezeNet models. number acquired from these subsequently reduced using principal component analysis (PCA) fast Walsh-Hadamard transform (FHWT) techniques. Following that, discrete wavelet (DWT) used fuse FWHT's obtained three Additionally, models' PCA concatenated. Finally, diminished as result FHWT-DWT fusion processes four distinct machine algorithms, reaching highest accuracy 99.6%. results show can distinguish variants with lower complexity compared existing methods. also prove utilizing reduce offer superior interpretation data, thus improving diagnosis procedure.

Language: Английский

---

A Hybrid Workflow of Residual Convolutional Transformer Encoder for Breast Cancer Classification Using Digital X-ray Mammograms

Biomedicines, Journal Year: 2022, Volume and Issue: 10(11), P. 2971 - 2971

Published: Nov. 18, 2022

Breast cancer, which attacks the glandular epithelium of breast, is second most common kind cancer in women after lung and it affects a significant number people worldwide. Based on advantages Residual Convolutional Network Transformer Encoder with Multiple Layer Perceptron (MLP), this study proposes novel hybrid deep learning Computer-Aided Diagnosis (CAD) system for breast lesions. While backbone residual network employed to create features, transformer utilized classify according self-attention mechanism. The proposed CAD has capability recognize two scenarios: Scenario A (Binary classification) B (Multi-classification). Data collection preprocessing, patch image creation splitting, artificial intelligence-based lesion identification are all components execution framework that applied consistently across both cases. effectiveness AI model compared against three separate models: custom CNN, VGG16, ResNet50. Two datasets, CBIS-DDSM DDSM, construct test system. Five-fold cross validation data used evaluate accuracy performance results. suggested achieves encouraging evaluation results, overall accuracies 100% 95.80% binary multiclass prediction challenges, respectively. experimental results reveal could identify benign malignant tissues significantly, important radiologists recommend further investigation abnormal mammograms provide optimal treatment plan.

Language: Английский

---

BO-ALLCNN: Bayesian-Based Optimized CNN for Acute Lymphoblastic Leukemia Detection in Microscopic Blood Smear Images

Sensors, Journal Year: 2022, Volume and Issue: 22(15), P. 5520 - 5520

Published: July 24, 2022

Acute lymphoblastic leukemia (ALL) is a deadly cancer characterized by aberrant accumulation of immature lymphocytes in the blood or bone marrow. Effective treatment ALL strongly associated with early diagnosis disease. Current practice for initial performed through manual evaluation stained smear microscopy images, which time-consuming and error-prone process. Deep learning-based human-centric biomedical has recently emerged as powerful tool assisting physicians making medical decisions. Therefore, numerous computer-aided diagnostic systems have been developed to autonomously identify images. In this study, new Bayesian-based optimized convolutional neural network (CNN) introduced detection microscopic To promote classification performance, architecture proposed CNN its hyperparameters are customized input data Bayesian optimization approach. The technique adopts an informed iterative procedure search hyperparameter space optimal set that minimizes objective error function. trained validated using hybrid dataset formed integrating two public datasets. Data augmentation adopted further supplement image boost performance. search-derived model recorded improved performance image-based on test set. findings study reveal superiority Bayesian-optimized over other deep learning models.

Language: Английский

---

A survey of multimodal hybrid deep learning for computer vision: Architectures, applications, trends, and challenges

Information Fusion, Journal Year: 2023, Volume and Issue: 105, P. 102217 - 102217

Published: Dec. 30, 2023

Language: Английский

---

Histopathological Analysis for Detecting Lung and Colon Cancer Malignancies Using Hybrid Systems with Fused Features

Bioengineering, Journal Year: 2023, Volume and Issue: 10(3), P. 383 - 383

Published: March 21, 2023

Lung and colon cancer are among humanity's most common deadly cancers. In 2020, there were 4.19 million people diagnosed with lung cancer, more than 2.7 died worldwide. Some develop simultaneously due to smoking which causes leading an abnormal diet, also cancer. There many techniques for diagnosing notably the biopsy technique its analysis in laboratories. Due scarcity of health centers medical staff, especially developing countries. Moreover, manual diagnosis takes a long time is subject differing opinions doctors. Thus, artificial intelligence solve these challenges. this study, three strategies developed, each two systems early histological images LC25000 dataset. Histological have been improved, contrast affected areas has increased. The GoogLeNet VGG-19 models all produced high dimensional features, so redundant unnecessary features removed reduce dimensionality retain essential by PCA method. first strategy dataset ANN uses crucial separately. second combined VGG-19. One system reduced dimensions combined, while other then dimensions. third fusion CNN (GoogLeNet VGG-19) handcrafted features. With reached sensitivity 99.85%, precision 100%, accuracy 99.64%, specificity AUC 99.86%.

Language: Английский

---

Benign and Malignant Breast Tumor Classification in Ultrasound and Mammography Images via Fusion of Deep Learning and Handcraft Features

Entropy, Journal Year: 2023, Volume and Issue: 25(7), P. 991 - 991

Published: June 28, 2023

Breast cancer is a disease that affects women in different countries around the world. The real cause of breast particularly challenging to determine, and early detection necessary for reducing death rate, due high risks associated with cancer. Treatment period can increase life expectancy quality women. CAD (Computer Aided Diagnostic) systems perform diagnosis benign malignant lesions using technologies tools based on image processing, helping specialist doctors obtain more precise point view fewer processes when making their by giving second opinion. This study presents novel system automated diagnosis. proposed method consists stages. In preprocessing stage, an segmented, mask lesion obtained; during next extraction deep learning features performed CNN—specifically, DenseNet 201. Additionally, handcrafted (Histogram Oriented Gradients (HOG)-based, ULBP-based, perimeter area, eccentricity, circularity) are obtained from image. designed hybrid uses CNN architecture extracting features, along traditional methods which several handcraft following medical properties purpose later fusion via statistical criteria. During where analyzed, genetic algorithms as well mutual information selection algorithm, followed classifiers (XGBoost, AdaBoost, Multilayer perceptron (MLP)) stochastic measures, applied choose most sensible group among features. experimental validation two modalities design, types studies—mammography (MG) ultrasound (US)—the databases mini-DDSM (Digital Database Screening Mammography) BUSI (Breast Ultrasound Images Dataset) were used. Novel evaluated compared recent state-of-the-art systems, demonstrating better performance commonly used criteria, obtaining ACC 97.6%, PRE 98%, Recall F1-Score IBA 95% abovementioned datasets.

Language: Английский

---

Deep Learning Cascaded Feature Selection Framework for Breast Cancer Classification: Hybrid CNN with Univariate-Based Approach

Mathematics, Journal Year: 2022, Volume and Issue: 10(19), P. 3631 - 3631

Published: Oct. 4, 2022

With the help of machine learning, many problems that have plagued mammography in past been solved. Effective prediction models need normal and tumor samples. For medical applications such as breast cancer diagnosis framework, it is difficult to gather labeled training data construct effective learning frameworks. Transfer an emerging strategy has recently used tackle scarcity by transferring pre-trained convolutional network knowledge into domain. Despite well reputation transfer based on Convolutional Neural Networks (CNN) for imaging, several hurdles still exist achieve a prominent classification performance. In this paper, we attempt solve Feature Dimensionality Curse (FDC) problem deep features are derived from CNNs. Such raised due high space dimensionality extracted with respect small size available Therefore, novel cascaded feature selection framework proposed networks univariate-based paradigm. Deep AlexNet, VGG, GoogleNet randomly selected extract shallow INbreast mammograms, whereas univariate helps overcome curse multicollinearity issues features. The optimized key via approach statistically significant (p-value ≤ 0.05) good capability efficiently train models. Using optimal features, could promising evaluation performance terms 98.50% accuracy, 98.06% sensitivity, 98.99% specificity, 98.98% precision. seems be beneficial develop practical reliable computer-aided (CAD) classification.

Language: Английский

---