A novel loss function to reproduce texture features for deep learning‐based MRI‐to‐CT synthesis DOI Open Access

Siqi Yuan,

Yuxiang Liu, Ran Wei

et al.

Medical Physics, Journal Year: 2023, Volume and Issue: 51(4), P. 2695 - 2706

Published: Dec. 3, 2023

Abstract Background Studies on computed tomography (CT) synthesis based magnetic resonance imaging (MRI) have mainly focused pixel‐wise consistency, but the texture features of regions interest (ROIs) not received appropriate attention. Purpose This study aimed to propose a novel loss function reproduce ROIs and consistency for deep learning‐based MRI‐to‐CT synthesis. The method was expected assist multi‐modality studies radiomics. Methods retrospectively enrolled 127 patients with nasopharyngeal carcinoma. CT MRI images were collected each patient, then rigidly registered as pre‐procession. We proposed gray‐level co‐occurrence matrix (GLCM)‐based improve reproducibility features. could be embedded into present framework image In this study, typical model selected baseline, which contained Unet trained mean square error (MSE) function. designed experiments supervise different prove its effectiveness. concordance correlation coefficient (CCC) GLCM feature employed evaluate features, are Besides, we used publicly available dataset brain tumors verify our Results Compared improved quality metrics (MAE: 107.5 106.8 HU; SSIM: 0.9728 0.9730). CCC values in GTVnx significantly from 0.78 ± 0.12 0.82 0.11 ( p < 0.05 paired t ‐test). Generally, > 90% (22/24) GLCM‐based compared where Informational Measure Correlation most (CCC: 0.74 0.83). For public dataset, also shows With added, ability ROIs. Conclusions reproduced synthesis, would benefit radiomics

Language: Английский

Advancements in Automatic Kidney Segmentation Using Deep Learning Frameworks and Volumetric Segmentation Techniques for CT Imaging: A Review DOI
Vishal Kumar Kanaujia, Awadhesh Kumar, Satya Prakash Yadav

et al.

Archives of Computational Methods in Engineering, Journal Year: 2024, Volume and Issue: 31(5), P. 3151 - 3169

Published: Feb. 19, 2024

Language: Английский

Citations

4
Radiomics in head and neck squamous cell carcinoma – a leap towards precision oncology DOI Creative Commons
Pranjal Rai, Abhishek Mahajan

Journal for ImmunoTherapy of Cancer, Journal Year: 2025, Volume and Issue: 13(4), P. e011692 - e011692

Published: April 1, 2025

Immunotherapy has revolutionized head and neck squamous cell carcinoma (HNSCC) treatment, with neoadjuvant chemoimmunotherapy showing promising pathological complete response rates (36–42%). Lin et al introduce a radiomics-clinical nomogram using MRI-derived intratumoral peritumoral features to predict pCR, addressing critical clinical gap. Their model, emphasizing the region (within 3 mm), achieved high predictive accuracy area under curve (AUC) >0.8. While multicenter design enhances generalizability, standardizing imaging protocols remains challenge. Integrating radiomics Neck Imaging Reporting Data System could refine post-treatment assessment. This study advances precision oncology in HNSCC, offering non-invasive tool for personalized treatment strategies. Future directions include artificial intelligence-driven radiogenomics enhance prediction patient selection.

Language: Английский

Citations

0
ComBat models for harmonization of resting-state EEG features in multisite studies DOI Creative Commons
Alberto Jaramillo‐Jimenez, Diego Tovar, Yorguin-José Mantilla-Ramos

et al.

Clinical Neurophysiology, Journal Year: 2024, Volume and Issue: 167, P. 241 - 253

Published: Sept. 24, 2024

Language: Английский

Citations

3
Robustness of textural analysis features in quantitative 99 mTc and 177Lu SPECT-CT phantom acquisitions DOI Creative Commons
Alastair J. Gemmell,

Colin M Brown,

Surajit Ray

et al.

EJNMMI Physics, Journal Year: 2025, Volume and Issue: 12(1)

Published: April 17, 2025

Abstract Background Textural Analysis features in molecular imaging require to be robust under repeat measurement and independent of volume for optimum use clinical studies. Recent EANM SNMMI guidelines radiomics provide advice on the potential phantoms identify (Hatt EJNMMI, 2022). This study applies suggested SPECT quantification two radionuclides, 99 m Tc 177 Lu. Methods Acquisitions were made with a uniform phantom test dependency customised ‘Revolver’ phantom, based PET described Hatt (EJNMMI, 2022) but local adaptations SPECT. Each was filled separately Sixty-seven extracted tested robustness dependency. Results Features showing high or Coefficient Variation (indicating poor repeatability) removed from list that may suitable After feature reduction, there 39 33 Lu remaining. Conclusion The Revolver repeatable is possible quantitative using Selection such likely centre-dependent due differences camera performance as well acquisition reconstruction protocols.

Language: Английский

Citations

0
Multimodal Imaging Approach for Tumor Treatment Response Evaluation in the Era of Immunotherapy DOI
Geewon Lee, Seung Hwan Moon, Jong‐Hoon Kim

et al.

Investigative Radiology, Journal Year: 2024, Volume and Issue: unknown

Published: July 17, 2024

Immunotherapy is likely the most remarkable advancement in lung cancer treatment during past decade. Although immunotherapy provides substantial benefits, their therapeutic responses differ from those of conventional chemotherapy and targeted therapy, some patients present unique response patterns that cannot be judged under current measurement standards. Therefore, monitoring can challenging, such as differentiation between real pseudo-response. This review outlines various tumor to discusses methods for quantifying computed tomography (CT) 18F-fluorodeoxyglucose positron emission (PET) field cancer. Emerging technologies magnetic resonance imaging (MRI) non-FDG PET tracers are also explored. With responses, role essential both anatomical radiological (CT/MRI) molecular changes (PET imaging). Multiple aspects must considered when assessing using CT PET. Finally, we introduce multimodal approaches integrate nonimaging data, discuss future directions assessment prediction immunotherapy.

Language: Английский

Citations

2
Performance of baseline FDG-PET/CT radiomics for prediction of bone marrow minimal residual disease status in the LyMa-101 trial DOI Creative Commons
Caroline Bodet‐Milin,

Cyrille Morvant,

Thomas Carlier

et al.

Scientific Reports, Journal Year: 2023, Volume and Issue: 13(1)

Published: Oct. 24, 2023

The prognostic value of 18F-Fluorodeoxyglucose positron emission tomography/computed tomography (FDG-PET/CT) at baseline or the predictive minimal residual disease (MRD) detection appear as potential tools to improve mantle cell lymphoma (MCL) patients' management. LyMa-101, a phase 2 trial LYSA group (ClinicalTrials.gov:NCT02896582) reported induction therapy with obinutuzumab, CD20 monoclonal antibody. Herein, we investigated added radiomic features (RF) derived from FDG-PET/CT diagnosis for MRD prediction. 59 MCL patients included in LyMa-101 have been independently, blindly and centrally reviewed. RF were extracted area highest uptake total metabolic tumor volume (TMTV). Two models machine learning used compare several combinations prediction before autologous stem transplant consolidation (ASCT). Each algorithm was generated without constrained feature selections clinical laboratory parameters. Both algorithms showed better discrimination performances negative vs positive lesion than TMTV. use biological clear loss sensitivity status ASCT, regardless model. These data plead importance compared parameters also reinforced previously made hypothesis that prognosis is linked most aggressive contingent, within uptake.

Language: Английский

Citations

4
Development and Validation of Prognostic Models Using Radiomic Features from Pre-Treatment Positron Emission Tomography (PET) Images in Head and Neck Squamous Cell Carcinoma (HNSCC) Patients DOI Open Access
Mahima Merin Philip,

Jessica Watts,

Fergus McKiddie

et al.

Cancers, Journal Year: 2024, Volume and Issue: 16(12), P. 2195 - 2195

Published: June 11, 2024

High-dimensional radiomics features derived from pre-treatment positron emission tomography (PET) images offer prognostic insights for patients with head and neck squamous cell carcinoma (HNSCC). Using 124 PET clinical variables (age, sex, stage of cancer, site cancer) a cohort 232 patients, we evaluated four survival models-penalized Cox model, random forest, gradient boosted model support vector machine-to predict all-cause mortality (ACM), locoregional recurrence/residual disease (LR) distant metastasis (DM) probability during 36, 24 months follow-up, respectively. We developed models five-fold cross-validation, selected the best-performing each outcome based on concordance index (C-statistic) integrated Brier score (IBS) validated them in an independent 102 patients. The penalized demonstrated better performance ACM (C-statistic = 0.70, IBS 0.12) DM 0.08) while forest displayed LR 0.76, 0.07). conclude that ML-based can aid clinicians quantifying prognosis determining effective treatment strategies, thereby improving favorable outcomes HNSCC

Language: Английский

Citations

1
Computer Vision—Radiomics & Pathognomics DOI
Alexandra T. Bourdillon

Otolaryngologic Clinics of North America, Journal Year: 2024, Volume and Issue: 57(5), P. 719 - 751

Published: June 22, 2024

Language: Английский

Citations

1
An Innovative and Efficient Diagnostic Prediction Flow for Head and Neck Cancer: A Deep Learning Approach for Multi-Modal Survival Analysis Prediction Based on Text and Multi-Center PET/CT Images DOI Creative Commons
Zhaonian Wang,

Chundan Zheng,

Han Xu

et al.

Diagnostics, Journal Year: 2024, Volume and Issue: 14(4), P. 448 - 448

Published: Feb. 17, 2024

Objective: To comprehensively capture intra-tumor heterogeneity in head and neck cancer (HNC) maximize the use of valid information collected clinical field, we propose a novel multi-modal image–text fusion strategy aimed at improving prognosis. Method: We have developed tailored diagnostic algorithm for HNC, leveraging deep learning-based model that integrates both image text information. For part, used cross-attention mechanism to fuse between PET CT, image, Q-former architecture also improved traditional prognostic by introducing time as variable construction model, finally obtained corresponding results. Result: assessed efficacy our methodology through compilation multicenter dataset, achieving commendable outcomes validations. Notably, results metastasis-free survival (MFS), recurrence-free (RFS), overall (OS), progression-free (PFS) were follows: 0.796, 0.626, 0.641, 0.691. Our demonstrate notable superiority over utilization CT independently, exceed result derived without textual Conclusions: not only validates effectiveness aiding diagnosis, but provides insights optimizing analysis. The study underscores potential approach enhancing prognosis contributing advancement personalized medicine HNC.

Language: Английский

Citations

1
A novel loss function to reproduce texture features for deep learning‐based MRI‐to‐CT synthesis DOI Open Access

Siqi Yuan,

Yuxiang Liu, Ran Wei

et al.

Medical Physics, Journal Year: 2023, Volume and Issue: 51(4), P. 2695 - 2706

Published: Dec. 3, 2023

Abstract Background Studies on computed tomography (CT) synthesis based magnetic resonance imaging (MRI) have mainly focused pixel‐wise consistency, but the texture features of regions interest (ROIs) not received appropriate attention. Purpose This study aimed to propose a novel loss function reproduce ROIs and consistency for deep learning‐based MRI‐to‐CT synthesis. The method was expected assist multi‐modality studies radiomics. Methods retrospectively enrolled 127 patients with nasopharyngeal carcinoma. CT MRI images were collected each patient, then rigidly registered as pre‐procession. We proposed gray‐level co‐occurrence matrix (GLCM)‐based improve reproducibility features. could be embedded into present framework image In this study, typical model selected baseline, which contained Unet trained mean square error (MSE) function. designed experiments supervise different prove its effectiveness. concordance correlation coefficient (CCC) GLCM feature employed evaluate features, are Besides, we used publicly available dataset brain tumors verify our Results Compared improved quality metrics (MAE: 107.5 106.8 HU; SSIM: 0.9728 0.9730). CCC values in GTVnx significantly from 0.78 ± 0.12 0.82 0.11 ( p < 0.05 paired t ‐test). Generally, > 90% (22/24) GLCM‐based compared where Informational Measure Correlation most (CCC: 0.74 0.83). For public dataset, also shows With added, ability ROIs. Conclusions reproduced synthesis, would benefit radiomics

Language: Английский

Citations

3