Colon and lung cancer classification from multi-modal images using resilient and efficient neural network architectures

Heliyon, Год журнала: 2024, Номер 10(9), С. e30625 - e30625

Опубликована: Май 1, 2024

Automatic classification of colon and lung cancer images is crucial for early detection accurate diagnostics. However, there room improvement to enhance accuracy, ensuring better diagnostic precision. This study introduces two novel dense architectures (D1 D2) emphasizes their effectiveness in classifying from diverse images. It also highlights resilience, efficiency, superior performance across multiple datasets. These were tested on various types datasets, including NCT-CRC-HE-100K (set 100,000 non-overlapping image patches hematoxylin eosin (H&E) stained histological human colorectal (CRC) normal tissue), CRC-VAL-HE-7K 7180 N=50 patients with adenocarcinoma, no overlap NCT-CRC-HE-100K), LC25000 (Lung Colon Cancer Histopathological Image), IQ-OTHNCCD (Iraq-Oncology Teaching Hospital/National Center Diseases), showcasing cancers histopathological Computed Tomography (CT) scan underscores the multi-modal capability proposed models. Moreover, addresses imbalanced particularly IQ-OTHNCCD, a specific focus model resilience robustness. To assess overall performance, conducted experiments different scenarios. The D1 achieved an impressive 99.80% accuracy dataset, Jaccard Index (J) 0.8371, Matthew's Correlation Coefficient (MCC) 0.9073, Cohen's Kappa (Kp) 0.9057, Critical Success (CSI) 0.8213. When subjected 10-fold cross-validation LC25000, averaged (avg) 99.96% (avg J, MCC, Kp, CSI 0.9993, 0.9987, 0.9853, 0.9990), surpassing recent reported performances. Furthermore, ensemble D2 reached 93% (J, 0.7556, 0.8839, 0.8796, 0.7140) exceeding benchmarks aligning other results. Efficiency evaluations For instance, training only 10% resulted high rates 99.19% 0.9840, 0.9898, 0.9837) (D1) 99.30% 0.9863, 0.9913, 0.9861) (D2). In NCT-CRC-HE-100K, 99.53% 0.9906, 0.9946, 0.9906) 30% dataset testing remaining 70%. CRC-VAL-HE-7K, 95% 0.8845, 0.9455, 0.9452, 0.8745) 96% 0.8926, 0.9504, 0.9503, 0.8798), respectively, outperforming previously results closely others. Lastly, just significant outperformance InceptionV3, Xception, DenseNet201 benchmarks, achieving rate 82.98% 0.7227, 0.8095, 0.8081, 0.6671). Finally, using explainable AI algorithms such as Grad-CAM, Grad-CAM++, Score-CAM, Faster along emphasized versions, we visualized features last layer well CT-scan samples. models, multi-modality, robustness, efficiency classification, hold promise advancements medical They have potential revolutionize improve healthcare accessibility worldwide.

Язык: Английский

---

An effective multiclass skin cancer classification approach based on deep convolutional neural network

Cluster Computing, Год журнала: 2024, Номер unknown

Опубликована: Июнь 17, 2024

Abstract Skin cancer is one of the most dangerous types due to its immediate appearance and possibility rapid spread. It arises from uncontrollably growing cells, rapidly dividing cells in area body, invading other bodily tissues, spreading throughout body. Early detection helps prevent progress reaching critical levels, reducing risk complications need for more aggressive treatment options. Convolutional neural networks (CNNs) revolutionize skin diagnosis by extracting intricate features images, enabling an accurate classification lesions. Their role extends early detection, providing a powerful tool dermatologists identify abnormalities their nascent stages, ultimately improving patient outcomes. This study proposes novel deep convolutional network (DCNN) approach classifying The proposed DCNN model evaluated using two unbalanced datasets, namely HAM10000 ISIC-2019. compared with transfer learning models, including VGG16, VGG19, DenseNet121, DenseNet201, MobileNetV2. Its performance assessed four widely used evaluation metrics: accuracy, recall, precision, F1-score, specificity, AUC. experimental results demonstrate that outperforms (DL) models utilized these datasets. achieved highest accuracy ISIC-2019 $$98.5\%$$ 98.5 % $$97.1\%$$ 97.1 , respectively. These show how competitive successful overcoming problems caused class imbalance raising accuracy. Furthermore, demonstrates superior performance, particularly excelling terms recent studies utilize same which highlights robustness effectiveness DCNN.

Язык: Английский

---

Deep feature selection using adaptive β-Hill Climbing aided whale optimization algorithm for lung and colon cancer detection

Biomedical Signal Processing and Control, Год журнала: 2023, Номер 83, С. 104692 - 104692

Опубликована: Фев. 16, 2023

Язык: Английский

---

ApneaNet: A hybrid 1DCNN-LSTM architecture for detection of Obstructive Sleep Apnea using digitized ECG signals

Biomedical Signal Processing and Control, Год журнала: 2023, Номер 84, С. 104754 - 104754

Опубликована: Март 3, 2023

Язык: Английский

---

Comparative analysis of machine learning and deep learning models for improved cancer detection: A comprehensive review of recent advancements in diagnostic techniques

Expert Systems with Applications, Год журнала: 2024, Номер 255, С. 124838 - 124838

Опубликована: Июль 23, 2024

Язык: Английский

---

Denoising of poisson-corrupted microscopic biopsy images using fourth-order partial differential equation with ant colony optimization

Biomedical Signal Processing and Control, Год журнала: 2024, Номер 93, С. 106207 - 106207

Опубликована: Март 14, 2024

Язык: Английский

---

The development of an efficient artificial intelligence-based classification approach for colorectal cancer response to radiochemotherapy: deep learning vs. machine learning

Scientific Reports, Год журнала: 2025, Номер 15(1)

Опубликована: Янв. 2, 2025

Colorectal cancer (CRC) is a form of that impacts both the rectum and colon. Typically, it begins with small abnormal growth known as polyp, which can either be non-cancerous or cancerous. Therefore, early detection colorectal second deadliest after lung cancer, highly beneficial. Moreover, standard treatment for locally advanced widely accepted around world, chemoradiotherapy. Then, in this study, seven artificial intelligence models including decision tree, K-nearest neighbors, Adaboost, random forest, Gradient Boosting, multi-layer perceptron, convolutional neural network were implemented to detect patients responder non-responder radiochemotherapy. For finding potential predictors (genes), three feature selection strategies employed mutual information, F-classif, Chi-Square. Based on models, four different scenarios developed five, ten, twenty thirty features selected designing more accurate classification paradigm. The results study confirm neighbors provided terms accuracy, by 93.8%. Among methods, information F-classif showed best results, while Chi-Square produced worst results. suggested successfully applied robust approach response radiochemotherapy medical studies.

Язык: Английский

---

ResNetFed: Federated Deep Learning Architecture for Privacy-Preserving Pneumonia Detection from COVID-19 Chest Radiographs

Journal of Healthcare Informatics Research, Год журнала: 2023, Номер 7(2), С. 203 - 224

Опубликована: Июнь 1, 2023

Personal health data is subject to privacy regulations, making it challenging apply centralized data-driven methods in healthcare, where personalized training frequently used. Federated Learning (FL) promises provide a decentralized solution this problem. In FL, siloed used for the model ensure privacy. paper, we investigate viability of federated approach using detection COVID-19 pneumonia as use case. 1411 individual chest radiographs, sourced from public repository COVIDx8 are The dataset contains radiographs 753 normal lung findings and 658 related pneumonias. We partition unevenly across five separate silos order reflect typical FL scenario. For binary image classification analysis these propose ResNetFed, pre-trained ResNet50 modified federation so that supports Differential Privacy. addition, customized strategy with radiographs. experimental results show ResNetFed clearly outperforms locally trained models. Due uneven distribution silos, observe models perform significantly worse than (mean accuracies 63% 82.82%, respectively). particular, shows excellent performance underpopulated achieving up +34.9 percentage points higher accuracy compared local Thus, can assist initial screening medical centers privacy-preserving manner.

Язык: Английский

---

Handling imbalanced class in melanoma: Kemeny–Young rule based optimal rank aggregation and Self-Adaptive Differential Evolution Optimization

Engineering Applications of Artificial Intelligence, Год журнала: 2023, Номер 125, С. 106738 - 106738

Опубликована: Июль 11, 2023

Язык: Английский

---

Improved Water Strider Algorithm With Convolutional Autoencoder for Lung and Colon Cancer Detection on Histopathological Images

IEEE Access, Год журнала: 2023, Номер 12, С. 949 - 956

Опубликована: Дек. 25, 2023

Lung and colon cancers are deadly diseases that can develop concurrently in organs undesirably affect human life some special cases. The detection of these from histopathological images poses a complex challenge medical diagnostics. Advanced image processing techniques, including deep learning algorithms, offer solution by analyzing intricate patterns structures slides. integration artificial intelligence analysis not only improves the proficiency cancer but also holds potential to increase prognostic assessments, eventually contributing effective treatment strategies for patients with lung cancers. This manuscript presents an Improved Water Strider Algorithm Convolutional Autoencoder Colon Cancer Detection (IWSACAE-LCCD) on HIs. major aim IWSACAE-LCCD technique aims detect cancer. For noise removal process, median filtering (MF) approach be used. Besides, convolutional neural network based MobileNetv2 model applied as feature extractor IWSA hyperparameter optimizer. Finally, autoencoder (CAE) presence To enhance results technique, series simulations were performed. obtained highlighted outperforms other approaches terms different measures.

Язык: Английский

---