Fast and Efficient Lung Abnormality Identification With Explainable AI: A Comprehensive Framework for Chest CT Scan and X-Ray Images

IEEE Access, Journal Year: 2024, Volume and Issue: 12, P. 31117 - 31135

Published: Jan. 1, 2024

A novel automated multi-classification approach is proposed for the anticipation of lung abnormalities using chest X-ray and CT images. The study leverages a publicly accessible dataset with an insufficient unbalanced number images, addressing this issue by employing data augmentation DCGAN to balance dataset. Various preprocessing procedures are applied improve features reduce noise in pictures. As base model, vision trans-former convolution-based compact convolutional transformer (CCT) model utilized. To determine best configuration, ablation performed on original CCT scan image dimensions 32 x 32. Following that, trained evaluate performance entirely other modality. performances compared six pre-trained models 32x32 While traditional achieved modest performance, test accuracies ranging from 43% 77% 49% 73% requiring lengthy training times, suggested exceptionally well, obtaining 99.77% 95.37% X-ray, respectively short duration 10-12 40-42 seconds/epoch. Robustness demonstrated through progressive reduction findings indicating that maintains good even reduced An explainable AI technique Grad-CAM used explain model's judgment. Grad-CAM-based color visualization shown assessments help health specialists make quick, confident decisions. This deep learning techniques detect anomalies, it addressed challenges time computational complexity.

Language: Английский

---

Malignancy pattern analysis of breast ultrasound images using clinical features and a graph convolutional network

Digital Health, Journal Year: 2024, Volume and Issue: 10

Published: Jan. 1, 2024

Early diagnosis of breast cancer can lead to effective treatment, possibly increase long-term survival rates, and improve quality life. The objective this study is present an automated analysis classification system for using clinical markers such as tumor shape, orientation, margin, surrounding tissue. novelty uniqueness the lie in approach considering medical features based on radiologists.

Language: Английский

---

Evaluating the effect of tissue stimulation at different frequencies on breast lesion classification based on nonlinear features using a novel radio frequency time series approach

Heliyon, Journal Year: 2024, Volume and Issue: 10(13), P. e33133 - e33133

Published: June 20, 2024

ObjectiveRadio Frequency Time Series (RF TS) is a cutting-edge ultrasound approach in tissue typing. The RF TS does not provide dynamic insights into the propagation medium; when and probe are fixed. We previously proposed innovative RFTSDP method which data recorded while stimulating tissue. Applying stimulation can unveil mechanical characteristics of echo.Materials MethodsIn this study, an apparatus was developed to induce vibrations at different frequencies medium. Data were collected from four PVA phantoms simulating nonlinear behaviors healthy, fibroadenoma, cyst, cancerous breast tissues. Raw focused, raw, beamformed ultrafast under conditions no stimulation, constant force, various vibrational stimulations using Supersonic Imagine Aixplorer clinical/research imaging system. domain (TD), spectral, features extracted each TS. Support Vector Machine (SVM), Random Forest, Decision Tree algorithms employed for classification.ResultsThe optimal outcome achieved SVM classifier considering 19 applying vibration frequency 65 Hz. classification accuracy, specificity, precision 98.44 0.20%, 99.49 0.01%, 98.53 0.04%, respectively. RFTSDP, notable 24.45% improvement accuracy observed compared case fixed assessing raw focused data.ConclusionsExternal appropriate frequency, as applied incorporates beneficial information about medium its TS, improve characterization.

Language: Английский

---

Innovative breast cancer detection using a segmentation-guided ensemble classification framework

Biomedical Engineering Letters, Journal Year: 2024, Volume and Issue: 15(1), P. 179 - 191

Published: Oct. 18, 2024

Language: Английский

---

Detection of explosives in dustbins using deep transfer learning based multiclass classifiers

Applied Intelligence, Journal Year: 2024, Volume and Issue: 54(2), P. 2314 - 2347

Published: Jan. 1, 2024

Language: Английский

---

PalScDiff: A diffusion-based framework with progressive augmentation learning and semantic consistency for breast ultrasound tumor segmentation

Journal of Intelligent & Fuzzy Systems, Journal Year: 2024, Volume and Issue: unknown, P. 1 - 15

Published: March 23, 2024

Background: Breast cancer diagnosis relies on accurate lesion segmentation in medical images. Automated computer-aided reduces clinician workload and improves efficiency, but existing image methods face challenges model performance generalization. Objective: This study aims to develop a generative framework using denoising diffusion for efficient breast Methods: We design novel framework, PalScDiff, that leverages probabilistic reconstruct the label distribution images, thereby enabling sampling of diverse, plausible outcomes. Specifically, with condition corresponding image, PalScDiff learns estimate masses region probability through step by step. Furthermore, we Progressive Augmentation Learning strategy incrementally handle irregular blurred tumors. Moreover, multi-round is employed achieve robust mass segmentation. Results: Our experimental results show outperforms established models such as U-Net transformer-based alternatives, achieving an accuracy 95.15%, precision 79.74%, Dice coefficient 77.61%, Intersection over Union (IOU) 81.51% . Conclusion: The proposed demonstrates promising capabilities cancer.

Language: Английский

---

Graph regularized least squares regression for automated breast ultrasound imaging

Neurocomputing, Journal Year: 2024, Volume and Issue: unknown, P. 129065 - 129065

Published: Dec. 1, 2024

Language: Английский

---

Can Vision Transformers Be the Next State-of-the-Art Model for Oncology Medical Image Analysis?

AI in Precision Oncology, Journal Year: 2024, Volume and Issue: 1(6), P. 286 - 305

Published: Dec. 1, 2024

Language: Английский

---

Spectrum is a Picture: Feasibility Study of Two-Dimensional Convolutional Neural Networks in Spectral Processing

Published: Jan. 1, 2024

In spectral data processing, a spectrum is usually represented in form of numeric vector for further processing. The same approach has been used also treatment by convolutional neural networks (CNN), initially purposed, however, image Analyzing as two-dimensional picture rather than one-dimensional potentially can improve the accuracy regression and classification models. purpose this work was to test assumption. We explored potential 2D-CNN with two case studies: Mössbauer - predict parameters numerically using (.bmp) files spectra near-infrared biological tissues. compared performance CNN types input data: pictures vectors. results indicate that proposed be helpful certain cases.

Language: Английский

---

EAH-Net: A Novel Ensemble Attention-Based Hybrid Architecture for Breast Cancer Diagnosis Utilizing Ultrasound Images

Published: Oct. 28, 2024

Breast cancer is a complex and often fatal malignancy in women worldwide, requiring thorough medical examinations. Accurately detecting breast challenging due to its diverse forms, stages, symptoms, diagnostic techniques. With advancements artificial intelligence, an automated computerized method can potentially aid radiologists the early detection of cancer. This study presents novel robust deep neural network, EAH-Net, for diagnosis using ultrasound images. The EAH-Net architecture comprises ensemble attention module, modified UNet model that performs segmentation by isolating regions interest, hybrid approach classify cancers accurately. Besides, we employed explainable AI techniques highlight most significant regions, assisting making more informed decisions. proposed framework yields promising outcomes across Jaccard, Precision, Recall, Specificity, Dice metrics, averaging 89.26 ± 0.36, 91.79 1.13, 92.98 1.08, 99.38 0.35, 95.26 0.45 percents, respectively. classification demonstrates outstanding performance with accuracy 98.48 0.18%. Overall, offers reliable computer-aided solution diagnosis.

Language: Английский

---