Lung Cancer Detection Systems Applied to Medical Images: A State-of-the-Art Survey

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

Published: May 22, 2024

Abstract Lung cancer represents a significant global health challenge, transcending demographic boundaries of age, gender, and ethnicity. Timely detection stands as pivotal factor for enhancing both survival rates post-diagnosis quality life. Artificial intelligence (AI) emerges transformative force with the potential to substantially enhance accuracy efficiency Computer-Aided Diagnosis (CAD) systems lung cancer. Despite burgeoning interest, notable gap persists in literature concerning comprehensive reviews that delve into intricate design architectural facets these systems. While existing furnish valuable insights result summaries model attributes, glaring absence prevails offering reliable roadmap guide researchers towards optimal research directions. Addressing this automated within medical imaging, survey adopts focused approach, specifically targeting innovative models tailored solely image analysis. The endeavors meticulously scrutinize merge knowledge pertaining components intended functionalities models. In adherence PRISMA guidelines, systematically incorporates analyzes 119 original articles spanning years 2019–2023 sourced from Scopus WoS-indexed repositories. is underpinned by three primary areas inquiry: application AI CAD systems, intricacies designs, comparative analyses latest advancements To ensure coherence depth analysis, surveyed methodologies are categorically classified seven distinct groups based on their foundational Furthermore, conducts rigorous review references discerns trend observations designs associated tasks. Beyond synthesizing knowledge, serves highlights avenues further critical domain. By providing facilitating informed decision-making, aims contribute body study propel field.

Language: Английский

---

Diagnosis Aid System for Colorectal Cancer Using Low Computational Cost Deep Learning Architectures

Electronics, Journal Year: 2024, Volume and Issue: 13(12), P. 2248 - 2248

Published: June 7, 2024

Colorectal cancer is the second leading cause of cancer-related deaths worldwide. To prevent deaths, regular screenings with histopathological analysis colorectal tissue should be performed. A diagnostic aid system could reduce time required by medical professionals, and provide an initial approach to final diagnosis. In this study, we analyze low computational custom architectures, based on Convolutional Neural Networks, which can serve as high-accuracy binary classifiers for screening using images. For purpose, carry out optimization process obtain best performance model in terms effectiveness a classifier cost reducing number parameters. Subsequently, compare results obtained previous work same field. Cross-validation reveals high robustness models classifiers, yielding superior accuracy outcomes 99.4 ± 0.58% 93.2 1.46% lighter model. The achieved exceeding 99% test subset low-resolution images significantly reduced layer count, sized at 11% those used studies. Consequently, estimate projected reduction up 50% costs compared most lightweight proposed existing literature.

Language: Английский

---

Predictive analytics of complex healthcare systems using deep learning based disease diagnosis model

Scientific Reports, Journal Year: 2024, Volume and Issue: 14(1)

Published: Nov. 11, 2024

Cancer is a life-threatening disease resulting from genetic disorder and range of metabolic anomalies. In particular, lung colon cancer (LCC) are among the major causes death in humans. The histopathological diagnoses critical detecting this kind cancer. This diagnostic testing substantial part patient's treatment. Thus, recognition classification LCC cutting-edge research regions, particularly biological healthcare medical fields. Earlier diagnosis can significantly reduce risk fatality. Machine learning (ML) deep (DL) models used to hasten these analyses, allowing researcher workers analyze considerable proportion patients limited time at low price. manuscript proposes Predictive Analytics Complex Healthcare Systems Using DL-based Disease Diagnosis Model (PACHS-DLBDDM) method. proposed PACHS-DLBDDM method majorly concentrates on detection LCC. At primary stage, methodology utilizes Gabor Filtering (GF) preprocess input imageries. Next, employs Faster SqueezeNet generate feature vectors. addition, convolutional neural network with long short-term memory (CNN-LSTM) approach classify To optimize hyperparameter values CNN-LSTM approach, Chaotic Tunicate Swarm Algorithm (CTSA) was implemented improve accuracy classifier results. simulation examined image dataset. performance validation model portrays superior value 99.54% over other DL models.

Language: Английский

---

Vision Transformers for Breast Cancer Human Epidermal Growth Factor Receptor 2 Expression Staging without Immunohistochemical Staining

American Journal Of Pathology, Journal Year: 2023, Volume and Issue: 194(3), P. 402 - 414

Published: Dec. 12, 2023

Accurate staging of human epidermal growth factor receptor 2 (HER2) expression is vital for evaluating breast cancer treatment efficacy. However, it typically involves costly and complex immunohistochemical staining, along with hematoxylin eosin staining. This work presents customized vision transformers HER2 in using only eosin–stained images. The proposed algorithm comprised three modules: a localization module weakly localizing critical image features spatial transformers, an attention global learning via loss to determine proximity level based on input images by calculating ordinal loss. Results, reported 95% CIs, reveal the approach's success staging: area under receiver operating characteristic curve, 0.9202 ± 0.01; precision, 0.922 sensitivity, 0.876 specificity, 0.959 0.02 over fivefold cross-validation. Comparatively, this approach significantly outperformed conventional transformer models state-of-the-art convolutional neural network (P < 0.001). Furthermore, surpassed existing methods when evaluated independent test data set. holds great importance, aiding while circumventing time-consuming staining procedure, thereby addressing diagnostic disparities low-resource settings low-income countries.

Language: Английский

---

Automated histopathological detection and classification of lung cancer with an image pre-processing pipeline and spatial attention with deep neural networks

Cogent Engineering, Journal Year: 2024, Volume and Issue: 11(1)

Published: May 22, 2024

Lung Cancer is a major cancer in the world and specifically India. Histopathological examination of tumorous tissue biopsy gold standard method used to clinically identify type, sub-type, stage cancer. Two most prevalent forms lung cancer: Adenocarcinoma & Squamous Cell Carcinoma account for nearly 80% all cases, which makes classifying two subtypes high importance. Proposed this study data pre-processing pipeline H&E-stained images along with customized EfficientNetB3-based Convolutional Neural Network employing spatial attention, trained on public three-class histopathological image dataset. The employed before training, validation testing helps enhance features removes biases due stain variations increased model robustness. usage pre-trained CNN deep learning generalize better weights, while attention mechanism On three-fold validation, classifier bagged accuracies 0.9943 ± 0.0012 0.9947 0.0018 combined F1-Scores 0.9942 0.0042 0.9833 0.0216 over respectively. performance its computational efficiency could enable easy deployment our without necessitating infrastructure overhaul.

Language: Английский

---