Deep Radiogenomics Sequencing for Breast Tumor Gene-Phenotype Decoding Using Dynamic Contrast Magnetic Resonance Imaging.

IF 3 4区医学 Q2 RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING

Molecular Imaging and Biology Pub Date : 2025-01-15 DOI:10.1007/s11307-025-01981-x

Isaac Shiri, Yazdan Salimi, Pooya Mohammadi Kazaj, Sara Bagherieh, Mehdi Amini, Abdollah Saberi Manesh, Habib Zaidi

{"title":"Deep Radiogenomics Sequencing for Breast Tumor Gene-Phenotype Decoding Using Dynamic Contrast Magnetic Resonance Imaging.","authors":"Isaac Shiri, Yazdan Salimi, Pooya Mohammadi Kazaj, Sara Bagherieh, Mehdi Amini, Abdollah Saberi Manesh, Habib Zaidi","doi":"10.1007/s11307-025-01981-x","DOIUrl":null,"url":null,"abstract":"Purpose: We aim to perform radiogenomic profiling of breast cancer tumors using dynamic contrast magnetic resonance imaging (MRI) for the estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) genes.Methods: The dataset used in the current study consists of imaging data of 922 biopsy-confirmed invasive breast cancer patients with ER, PR, and HER2 gene mutation status. Breast MR images, including a T1-weighted pre-contrast sequence and three post-contrast sequences, were enrolled for analysis. All images were corrected using N4 bias correction algorithms. Based on all images and tumor masks, a bounding box of 128 × 128 × 68 was chosen to include all tumor regions. All networks were implemented in 3D fashion with input sizes of 128 × 128 × 68, and four images were input to each network for multi-channel analysis. Data were randomly split into train/validation (80%) and test set (20%) with stratification in class (patient-wise), and all metrics were reported in 20% of the untouched test dataset.Results: For ER prediction, SEResNet50 achieved an AUC mean of 0.695 (CI95%: 0.610-0.775), a sensitivity of 0.564, and a specificity of 0.787. For PR prediction, ResNet34 achieved an AUC mean of 0.658 (95% CI: 0.573-0.741), a sensitivity of 0.593, and a specificity of 0.734. For HER2 prediction, SEResNext101 achieved an AUC mean of 0.698 (95% CI: 0.560-0.822), a sensitivity of 0.750, and a specificity of 0.625.Conclusion: The current study demonstrated the feasibility of imaging gene-phenotype decoding in breast tumors using MR images and deep learning algorithms with moderate performance.","PeriodicalId":18760,"journal":{"name":"Molecular Imaging and Biology","volume":" ","pages":""},"PeriodicalIF":3.0000,"publicationDate":"2025-01-15","publicationTypes":"Journal Article","fieldsOfStudy":null,"isOpenAccess":false,"openAccessPdf":"","citationCount":"0","resultStr":null,"platform":"Semanticscholar","paperid":null,"PeriodicalName":"Molecular Imaging and Biology","FirstCategoryId":"3","ListUrlMain":"https://doi.org/10.1007/s11307-025-01981-x","RegionNum":4,"RegionCategory":"医学","ArticlePicture":[],"TitleCN":null,"AbstractTextCN":null,"PMCID":null,"EPubDate":"","PubModel":"","JCR":"Q2","JCRName":"RADIOLOGY, NUCLEAR MEDICINE & MEDICAL IMAGING","Score":null,"Total":0}

引用次数: 0

Abstract

Purpose: We aim to perform radiogenomic profiling of breast cancer tumors using dynamic contrast magnetic resonance imaging (MRI) for the estrogen receptor (ER), progesterone receptor (PR), and human epidermal growth factor receptor 2 (HER2) genes.

Methods: The dataset used in the current study consists of imaging data of 922 biopsy-confirmed invasive breast cancer patients with ER, PR, and HER2 gene mutation status. Breast MR images, including a T1-weighted pre-contrast sequence and three post-contrast sequences, were enrolled for analysis. All images were corrected using N4 bias correction algorithms. Based on all images and tumor masks, a bounding box of 128 × 128 × 68 was chosen to include all tumor regions. All networks were implemented in 3D fashion with input sizes of 128 × 128 × 68, and four images were input to each network for multi-channel analysis. Data were randomly split into train/validation (80%) and test set (20%) with stratification in class (patient-wise), and all metrics were reported in 20% of the untouched test dataset.

Results: For ER prediction, SEResNet50 achieved an AUC mean of 0.695 (CI95%: 0.610-0.775), a sensitivity of 0.564, and a specificity of 0.787. For PR prediction, ResNet34 achieved an AUC mean of 0.658 (95% CI: 0.573-0.741), a sensitivity of 0.593, and a specificity of 0.734. For HER2 prediction, SEResNext101 achieved an AUC mean of 0.698 (95% CI: 0.560-0.822), a sensitivity of 0.750, and a specificity of 0.625.

Conclusion: The current study demonstrated the feasibility of imaging gene-phenotype decoding in breast tumors using MR images and deep learning algorithms with moderate performance.

查看原文本刊更多论文

使用动态对比磁共振成像进行乳腺肿瘤基因表型解码的深度放射基因组测序。

目的：我们的目标是利用动态对比磁共振成像（MRI）对乳腺癌肿瘤进行雌激素受体（ER）、孕激素受体（PR）和人表皮生长因子受体2 （HER2）基因的放射基因组谱分析。方法：本研究使用的数据集包括922例活检证实的ER、PR和HER2基因突变状态的浸润性乳腺癌患者的影像学数据。乳腺MR图像，包括t1加权前对比序列和三个后对比序列，被纳入分析。所有图像均采用N4偏置校正算法进行校正。基于所有图像和肿瘤蒙版，选择一个128 × 128 × 68的边界框来包含所有肿瘤区域。所有网络均以3D方式实现，输入尺寸为128 × 128 × 68，每个网络输入4张图像进行多通道分析。数据被随机分成训练/验证（80%）和测试集（20%），并按类别（患者方向）分层，所有指标在20%的未动测试数据集中报告。结果：SEResNet50预测ER的AUC平均值为0.695 (CI95%: 0.610-0.775)，敏感性为0.564，特异性为0.787。对于PR预测，ResNet34的AUC平均值为0.658 (95% CI: 0.573-0.741)，敏感性为0.593，特异性为0.734。对于HER2预测，SEResNext101的AUC平均值为0.698 (95% CI: 0.60 -0.822)，敏感性为0.750，特异性为0.625。结论：本研究证明了利用磁共振图像和深度学习算法进行乳腺肿瘤基因表型解码成像的可行性。

本文章由计算机程序翻译，如有差异，请以英文原文为准。

求助全文

约1分钟内获得全文求助全文

来源期刊

Molecular Imaging and Biology 医学-核医学

CiteScore

6.90

自引率

3.20%

发文量

审稿时长

3 months

期刊介绍： Molecular Imaging and Biology (MIB) invites original contributions (research articles, review articles, commentaries, etc.) on the utilization of molecular imaging (i.e., nuclear imaging, optical imaging, autoradiography and pathology, MRI, MPI, ultrasound imaging, radiomics/genomics etc.) to investigate questions related to biology and health. The objective of MIB is to provide a forum to the discovery of molecular mechanisms of disease through the use of imaging techniques. We aim to investigate the biological nature of disease in patients and establish new molecular imaging diagnostic and therapy procedures. Some areas that are covered are: Preclinical and clinical imaging of macromolecular targets (e.g., genes, receptors, enzymes) involved in significant biological processes. The design, characterization, and study of new molecular imaging probes and contrast agents for the functional interrogation of macromolecular targets. Development and evaluation of imaging systems including instrumentation, image reconstruction algorithms, image analysis, and display. Development of molecular assay approaches leading to quantification of the biological information obtained in molecular imaging. Study of in vivo animal models of disease for the development of new molecular diagnostics and therapeutics. Extension of in vitro and in vivo discoveries using disease models, into well designed clinical research investigations. Clinical molecular imaging involving clinical investigations, clinical trials and medical management or cost-effectiveness studies.