Diagnosis of Idiopathic Pulmonary Fibrosis in High-Resolution Computed Tomography Scans Using a Combination of Handcrafted Radiomics and Deep Learning DOI Creative Commons
Turkey Refaee, Zohaib Salahuddin,

Anne-Noëlle Frix

et al.

Frontiers in Medicine, Journal Year: 2022, Volume and Issue: 9

Published: June 23, 2022

To develop handcrafted radiomics (HCR) and deep learning (DL) based automated diagnostic tools that can differentiate between idiopathic pulmonary fibrosis (IPF) non-IPF interstitial lung diseases (ILDs) in patients using high-resolution computed tomography (HRCT) scans. In this retrospective study, 474 HRCT scans were included (mean age, 64.10 years ± 9.57 [SD]). Five-fold cross-validation was performed on 365 Furthermore, an external dataset comprising 109 used as a test set. An HCR model, DL ensemble of model developed. A virtual in-silico trial conducted with two radiologists one pulmonologist the same set for performance comparison. The compared DeLong method McNemar test. Shapley Additive exPlanations (SHAP) plots Grad-CAM heatmaps post-hoc interpretability models, respectively. five-fold cross-validation, models achieved accuracies 76.2 6.8, 77.9 4.6, 85.2 2.7%, For diagnosis IPF ILDs set, HCR, DL, 76.1, 77.9, 85.3%, outperformed clinicians who mean accuracy 66.3 6.7% (p < 0.05) during trial. area under receiver operating characteristic curve (AUC) 0.917 which significantly higher than (0.817, p = 0.02) (0.823, 0.005). agreement 61.4%, specificity predictions when both agree 93 97%, SHAP analysis showed texture features most important focused clinically relevant part image. Deep complement each other serve useful clinical aids ILDs.

Language: Английский

Artificial Intelligence in Cervical Cancer Screening and Diagnosis DOI Creative Commons

Xin Hou,

Guangyang Shen,

Liqiang Zhou

et al.

Frontiers in Oncology, Journal Year: 2022, Volume and Issue: 12

Published: March 11, 2022

Cervical cancer remains a leading cause of death in women, seriously threatening their physical and mental health. It is an easily preventable with early screening diagnosis. Although technical advancements have significantly improved the diagnosis cervical cancer, accurate difficult owing to various factors. In recent years, artificial intelligence (AI)-based medical diagnostic applications been on rise excellent applicability cancer. Their benefits include reduced time consumption, need for professional personnel, no bias subjective We, thus, aimed discuss how AI can be used diagnosis, particularly improve accuracy The application challenges using treatment are also discussed.

Language: Английский

Citations

104
Systematic review of the radiomics quality score applications: an EuSoMII Radiomics Auditing Group Initiative DOI Creative Commons

Gaia Spadarella,

Arnaldo Stanzione, Tugba Akinci D’Antonoli

et al.

European Radiology, Journal Year: 2022, Volume and Issue: 33(3), P. 1884 - 1894

Published: Oct. 25, 2022

The main aim of the present systematic review was a comprehensive overview Radiomics Quality Score (RQS)-based reviews to highlight common issues and challenges radiomics research application evaluate relationship between RQS features.

Language: Английский

Citations

84
Radiomics and Its Feature Selection: A Review DOI Open Access
Wenchao Zhang, Yu Guo, Qiyu Jin

et al.

Symmetry, Journal Year: 2023, Volume and Issue: 15(10), P. 1834 - 1834

Published: Sept. 27, 2023

Medical imaging plays an indispensable role in evaluating, predicting, and monitoring a range of medical conditions. Radiomics, specialized branch imaging, utilizes quantitative features extracted from images to describe underlying pathologies, genetic information, prognostic indicators. The integration radiomics with artificial intelligence presents innovative avenues for cancer diagnosis, prognosis evaluation, therapeutic choices. In the context oncology, offers significant potential. Feature selection emerges as pivotal step, enhancing clinical utility precision radiomics. It achieves this by purging superfluous unrelated features, thereby augmenting model performance generalizability. goal review is assess fundamental process progress feature methods, explore their applications challenges research, provide theoretical methodological support future investigations. Through extensive literature survey, articles pertinent were garnered, synthesized, appraised. paper provides detailed descriptions how applied challenged different types various stages. also comparative insights into strategies, including filtering, packing, embedding methodologies. Conclusively, broaches limitations prospective trajectories

Language: Английский

Citations

45
The effect of feature normalization methods in radiomics DOI Creative Commons
Aydın Demircioğlu

Insights into Imaging, Journal Year: 2024, Volume and Issue: 15(1)

Published: Jan. 7, 2024

Abstract Objectives In radiomics, different feature normalization methods, such as z-Score or Min–Max, are currently utilized, but their specific impact on the model is unclear. We aimed to measure effect predictive performance and selection. Methods employed fifteen publicly available radiomics datasets compare seven methods. Using four selection classifier we used cross-validation area under curve (AUC) of resulting models, agreement selected features, calibration. addition, assessed whether before introduces bias. Results On average, difference between methods was relatively small, with a gain at most + 0.012 in AUC when comparing (mean AUC: 0.707 ± 0.102) no 0.719 0.107). However, some datasets, reached 0.051. The performed best, while tanh transformation showed worst even decreased overall performance. While quantile performed, slightly worse than z-Score, it outperformed all other one out three datasets. features by only mild, reaching 62%. Applying did not introduce significant Conclusion choice method influenced depended strongly dataset. It impacted set features. Critical relevance statement Feature plays crucial role preprocessing influences complicating interpretation. Key points • radiomic models measured. Normalization similarly differed more Different led sets impeding Model calibration largely affected method. Graphical

Language: Английский

Citations

20
The application of radiomics in cancer imaging with a focus on lung cancer, renal cell carcinoma, gastrointestinal cancer, and head and neck cancer: A systematic review DOI
Roberta Fusco, Vincenza Granata, Sergio Venanzio Setola

et al.

Physica Medica, Journal Year: 2025, Volume and Issue: 130, P. 104891 - 104891

Published: Jan. 8, 2025

Language: Английский

Citations

2
Towards a safe and efficient clinical implementation of machine learning in radiation oncology by exploring model interpretability, explainability and data-model dependency DOI Creative Commons
Ana María Barragán Montero, Adrien Bibal, Margerie Huet Dastarac

et al.

Physics in Medicine and Biology, Journal Year: 2022, Volume and Issue: 67(11), P. 11TR01 - 11TR01

Published: April 14, 2022

Abstract The interest in machine learning (ML) has grown tremendously recent years, partly due to the performance leap that occurred with new techniques of deep learning, convolutional neural networks for images, increased computational power, and wider availability large datasets. Most fields medicine follow popular trend and, notably, radiation oncology is one those are at forefront, already a long tradition using digital images fully computerized workflows. ML models driven by data, contrast many statistical or physical models, they can be very complex, countless generic parameters. This inevitably raises two questions, namely, tight dependence between datasets feed them, interpretability which scales its complexity. Any problems data used train model will later reflected their performance. This, together low makes implementation into clinical workflow particularly difficult. Building tools risk assessment quality assurance must involve then main points: data-model dependency. After joint introduction both ML, this paper reviews risks current solutions when applying latter workflows former. Risks associated as well interaction, detailed. Next, core concepts interpretability, explainability, dependency formally defined illustrated examples. Afterwards, broad discussion goes through key applications vendors’ perspectives ML.

Language: Английский

Citations

53
Trends and statistics of artificial intelligence and radiomics research in Radiology, Nuclear Medicine, and Medical Imaging: bibliometric analysis DOI
Burak Koçak, Bettina Baeßler, Renato Cuocolo

et al.

European Radiology, Journal Year: 2023, Volume and Issue: 33(11), P. 7542 - 7555

Published: June 14, 2023

Language: Английский

Citations

31
Emerging technologies for cancer therapy using accelerated particles DOI
Christian Graeff, Lennart Volz, Marco Durante

et al.

Progress in Particle and Nuclear Physics, Journal Year: 2023, Volume and Issue: 131, P. 104046 - 104046

Published: April 8, 2023

Language: Английский

Citations

28
Rethinking the Role of AI with Physicians in Oncology: Revealing Perspectives from Clinical and Research Workflows DOI Open Access
Himanshu Verma, Jakub Mlynář, Roger Schaer

et al.

Published: April 19, 2023

Significant and rapid advancements in cancer research have been attributed to Artificial Intelligence (AI). However, AI's role impact on the clinical side has limited. This discrepancy manifests due overlooked, yet profound, differences practices oncology. Our contribution seeks scrutinize physicians' engagement with AI by interviewing 7 medical-imaging experts disentangle its future alignment across workflows, diverging from existing "one-size-fits-all" paradigm within Human-Centered discourses. analysis revealed that trust is less dependent their general acceptance of AI, but more contestable experiences AI. Contestability, underpins need for personal supervision outcomes processes, i.e., clinician-in-the-loop. Finally, we discuss tensions desired attributes such as explainability control, contextualizing them divergent intentionality scope workflows.

Language: Английский

Citations

28
Are deep models in radiomics performing better than generic models? A systematic review DOI Creative Commons
Aydın Demircioğlu

European Radiology Experimental, Journal Year: 2023, Volume and Issue: 7(1)

Published: March 15, 2023

Application of radiomics proceeds by extracting and analysing imaging features based on generic morphological, textural, statistical defined formulas. Recently, deep learning methods were applied. It is unclear whether models (DMs) can outperform (GMs).

Language: Английский

Citations

24