A novel Deep Learning architecture for lung cancer detection and diagnosis from Computed Tomography image analysis

Healthcare Analytics, Journal Year: 2024, Volume and Issue: 5, P. 100316 - 100316

Published: March 21, 2024

Timely identification of lung nodules, which are precursors to cancer, and their evaluation can significantly reduce the incidence rate. Computed Tomography (CT) is primary technique used for cancer screening due its high resolution. Identifying white, spherical shadows as nodules in CT images essential accurately detecting cancer. Convolutional Neural Network (CNN)-based methods have performed better than traditional techniques various medical image applications. However, challenges still need be addressed insufficient annotated datasets, significant intra-class variations, substantial inter-class similarities, hinder practical use. Manually labeling position on slices critical distinguishing between benign malignant cases, but it an unreliable time-consuming process. Insufficient data class imbalance factors that may result overfitting below-par performance. The paper presents a novel Deep Learning (DL) framework detect classify input images. It introduces 3D-VNet architecture accurate segmentation pulmonary 3D-ResNet designed classification. model achieves Dice Similarity Coefficient (DSC) 99.34% LUNA16 dataset while reducing false positives 0.4%. classification shows performance metrics with accuracy, sensitivity, specificity 99.2%, 98.8%, 99.6%, respectively. network outperforms previous by calibrating sizes shapes excellent robustness. model's show suggested method current approaches regarding specificity, sensitivity F1-Score.

Language: Английский

---

Diabetic retinopathy detection using ensembled transfer learning based thrice CNN with SVM classifier

Multimedia Tools and Applications, Journal Year: 2024, Volume and Issue: 83(27), P. 70089 - 70115

Published: Jan. 30, 2024

Language: Английский

---

EDTNet: A spatial aware attention-based transformer for the pulmonary nodule segmentation

PLoS ONE, Journal Year: 2024, Volume and Issue: 19(11), P. e0311080 - e0311080

Published: Nov. 15, 2024

Accurate segmentation of lung lesions in CT-scan images is essential to diagnose cancer. The challenges nodule diagnosis arise due their small size and diverse nature. We designed a transformer-based model EDTNet (Encoder Decoder Transformer Network) for PNS (Pulmonary Nodule Segmentation). Traditional CNN-based encoders decoders are hindered by inability capture long-range spatial dependencies, leading suboptimal performance complex object tasks. To address the limitation, we leverage an enhanced attention-based Vision (ViT) as encoder decoder EDTNet. integrates two successive transformer blocks, patch-expanding layer, down-sampling layers, up-sampling layers improve capabilities. In addition, ESLA (Enhanced aware local attention) EGLA global blocks added provide attention features. Furthermore, skip connections introduced facilitate symmetrical interaction between corresponding enabling retrieval intricate details output. compared with several models on DS1 DS2, including Unet, ResUNet++, U-NET 3+, DeepLabV3+, SegNet, Trans-Unet, Swin-UNet, demonstrates superior quantitative visual results. On DS1, achieved 96.27%, 95.81%, 96.15% precision, IoU (Intersection over Union), DSC (Sorensen–Dice coefficient). Moreover, has demonstrated sensitivity, SDC 98.84%, 96.06% 97.85% DS2.

Language: Английский

---

Introducing Radex: Adaptive Parameterized Feature Extraction from Medical Images

Lecture notes in computer science, Journal Year: 2025, Volume and Issue: unknown, P. 278 - 294

Published: Jan. 1, 2025

Language: Английский

---

Multi-objective deep learning for lung cancer detection in CT images: enhancements in tumor classification, localization, and diagnostic efficiency

Discover Oncology, Journal Year: 2025, Volume and Issue: 16(1)

Published: April 15, 2025

This study aims to develop and evaluate an advanced deep learning framework for the detection, classification, localization of lung tumors in computed tomography (CT) scan images. The research utilized a dataset 1608 CT images, including 623 cancerous 985 non-cancerous cases, all carefully labeled accurate tumor classification (benign or malignant), localization. preprocessing involved optimizing window settings, adjusting slice thickness, applying data augmentation techniques enhance model's robustness generalizability. proposed model incorporated innovative components such as transformer-based attention layers, adaptive anchor-free mechanisms, improved feature pyramid network. These features enabled efficiently handle tasks. was split into 70% training, 15% validation, testing. A multi-task loss function used balance three objectives optimize performance. Evaluation metrics included mean average precision (mAP), intersection over union (IoU), accuracy, precision, recall. demonstrated outstanding performance, achieving mAP 96.26%, IoU 95.76%, 98.11%, recall 98.83% on test dataset. It outperformed existing models, You Only Look Once (YOLO)v9 YOLOv10, with YOLOv10 95.23% YOLOv9 95.70%. showed faster convergence, better stability, superior detection capabilities, particularly localizing smaller tumors. Its significantly diagnostic accuracy operational efficiency. offers robust scalable solution cancer providing real-time inference, learning, high accuracy. holds significant potential clinical integration improve outcomes patient care.

Language: Английский

---

Quantum-enhanced hybrid feature engineering in thoracic CT image analysis for state-of-the-art nodule classification: an advanced lung cancer assessment

Biomedical Physics & Engineering Express, Journal Year: 2024, Volume and Issue: 10(4), P. 045005 - 045005

Published: April 25, 2024

Abstract The intricate nature of lung cancer treatment poses considerable challenges upon diagnosis. Early detection plays a pivotal role in mitigating its escalating global mortality rates. Consequently, there are pressing demands for robust and dependable early diagnostic systems. However, the technological limitations complexity disease make it challenging to implement an efficient screening system. AI-based CT image analysis techniques showing significant contributions development computer-assisted (CAD) systems screening. Various existing research groups working on implementing assessing classifying cancer. different structures inside is high comprehension information inherited by them more complex even after applying advanced feature extraction selection techniques. Traditional classical may struggle capture interdependencies between features. They get stuck local optima sometimes require additional exploration strategies. also with combinatorial optimization problems when applied prominent space. This paper proposed methodology overcome using Vision Transformer (FexViT) Feature Quantum Computing based Quadratic unconstrained binary (QC-FSelQUBO) technique. algorithm shows better performance compared other showed as evaluated necessary output measures, such accuracy, Area under roc (receiver operating characteristics) curve, precision, sensitivity, specificity, obtained 94.28%, 99.10%, 96.17%, 90.16% 97.46%. further advancement CAD essential meet demand reliable diagnosis cancer, which can be addressed leading quantum computation growing technology ahead.

Language: Английский

---

Diabetic retinopathy detection using EADBSC and improved dilated ensemble CNN-based classification

Multimedia Tools and Applications, Journal Year: 2023, Volume and Issue: 83(11), P. 33573 - 33595

Published: Sept. 21, 2023

Language: Английский

---

An ensemble reinforcement learning-assisted deep learning framework for enhanced lung cancer diagnosis

Swarm and Evolutionary Computation, Journal Year: 2024, Volume and Issue: 91, P. 101767 - 101767

Published: Nov. 15, 2024

Language: Английский

---

Eisoc with ifodpso and dcnn classifier for diabetic retinopathy recognition system

Multimedia Tools and Applications, Journal Year: 2023, Volume and Issue: 83(14), P. 42561 - 42583

Published: Oct. 9, 2023

Language: Английский

---

Identification of glaucoma in fundus images utilizing gray wolf optimization with deep convolutional neural network-based resnet50 model

Multimedia Tools and Applications, Journal Year: 2023, Volume and Issue: 83(16), P. 49301 - 49319

Published: Nov. 2, 2023

Language: Английский

---