Jose Gerardo Tamez Pena
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Research Subjects
Jose Gerardo Tamez Pena has a total of 261 co-authors. Here are the top co-authors by number of shared publications:
- Victor Trevino
Tecnologico de Monterrey, OriGen Project, Mexico
24 shared publications Saara Totterman
Qmetrics Technologies, Pittsford, NY, United States
21 shared publicationsS. Totterman
Qmetrics Technologies, Rochester, NY, USA
16 shared publications- José Manuel Celaya Padilla
Unidad Académica de Ingeniería Eléctrica, Universidad Autónoma de Zacatecas, Campus UAZ Siglo XXI, Carretera Zacatecas-Guadalajara Km. 6, Ejido La Escondida, Zacatecas 98160, Mexico
9 shared publications E. Schreyer
Qmetrics Technologies, Rochester, NY, USA
7 shared publicationsJ.M. Farber
7 shared publications
Latest Publications
- Open Access
Ensemble of radiomics and ConvNeXt for breast cancer diagnosis
Early diagnosis of breast cancer is crucial for improving survival rates. Radiomics and deep learning (DL) have shown significant potential in assisting radiologists with early cancer detection. This paper aims to critically assess the performance of radiomics, DL, and ensemble techniques in detecting cancer from screening mammograms. Two independent datasets were used: the RSNA 2023 Breast Cancer Detection Challenge (11,913 patients) and a Mexican cohort from the TecSalud dataset (19,400 patients). The ConvNeXtV1-small DL model was trained on the RSNA dataset and validated on the TecSalud dataset, while radiomics models were developed using the TecSalud dataset and validated with a leave-one-year-out approach. The ensemble method consistently combined and calibrated predictions using the same methodology. Results showed that the ensemble approach achieved the highest area under the curve (AUC) of 0.87, compared to 0.83 for ConvNeXtV1-small and 0.80 for radiomics. In conclusion, ensemble methods combining DL and radiomics predictions significantly enhance breast cancer diagnosis from mammograms.
Open AccessEnsemble of radiomics and ConvNeXt for breast cancer diagnosis
Early diagnosis of breast cancer is crucial for improving survival rates. Radiomics and deep learning (DL) have shown significant potential in assisting radiologists with early cancer detection. This paper aims to critically assess the performance of radiomics, DL, and ensemble techniques in detecting cancer from screening mammograms. Two independent datasets were used: the RSNA 2023 Breast Cancer Detection Challenge (11,913 patients) and a Mexican cohort from the TecSalud dataset (19,400 patients). The ConvNeXtV1-small DL model was trained on the RSNA dataset and validated on the TecSalud dataset, while radiomics models were developed using the TecSalud dataset and validated with a leave-one-year-out approach. The ensemble method consistently combined and calibrated predictions using the same methodology. Results showed that the ensemble approach achieved the highest area under the curve (AUC) of 0.87, compared to 0.83 for ConvNeXtV1-small and 0.80 for radiomics. In conclusion, ensemble methods combining DL and radiomics predictions significantly enhance breast cancer diagnosis from mammograms.
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 636-641
Comparison of ConvNeXt and Vision-Language Models for Breast Density Assessment in Screening Mammography
Mammographic breast density classification is essential for cancer risk assessment but remains challenging due to subjective interpretation and inter-observer variability. This study compares multimodal and CNN-based methods for automated classification using the BI-RADS system, evaluating BioMedCLIP and ConvNeXt across three learning scenarios: zero-shot classification, linear probing with textual descriptions, and fine-tuning with numerical labels. Results show that zero-shot classification achieved modest performance, while the fine-tuned ConvNeXt model outperformed the BioMedCLIP linear probe. Although linear probing demonstrated potential with pretrained embeddings, it was less effective than full fine-tuning. These findings suggest that despite the promise of multimodal learning, CNN-based models with end-to-end fine-tuning provide stronger performance for specialized medical imaging. The study underscores the need for more detailed textual representations and domain-specific adaptations in future radiology applications.
- 18 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE)
- p. 636-641
Comparison of ConvNeXt and Vision-Language Models for Breast Density Assessment in Screening Mammography
Mammographic breast density classification is essential for cancer risk assessment but remains challenging due to subjective interpretation and inter-observer variability. This study compares multimodal and CNN-based methods for automated classification using the BI-RADS system, evaluating BioMedCLIP and ConvNeXt across three learning scenarios: zero-shot classification, linear probing with textual descriptions, and fine-tuning with numerical labels. Results show that zero-shot classification achieved modest performance, while the fine-tuned ConvNeXt model outperformed the BioMedCLIP linear probe. Although linear probing demonstrated potential with pretrained embeddings, it was less effective than full fine-tuning. These findings suggest that despite the promise of multimodal learning, CNN-based models with end-to-end fine-tuning provide stronger performance for specialized medical imaging. The study underscores the need for more detailed textual representations and domain-specific adaptations in future radiology applications.
- Open Access
- 11 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
A ConvNeXt-Transformer Approach for Automated Conclusion Generation from Mammography
Breast cancer remains a critical global health concern, particularly in low-resource settings where limited access to radiologists delays diagnosis. While digital mammography enables early detection, its effectiveness is often constrained by radiological expertise shortages. This study proposes a multimodal deep learning system to automate the generation of radiology report conclusions from mammogram images, supporting diagnostic decision-making and reducing radiologist workload. An encoder-decoder architecture is implemented, combining a ConvNeXt model for image feature extraction with a Transformer-based decoder for text generation. The model is trained and evaluated on the RSNA Screening Mammography dataset and a curated dataset from Tec Salud in Mexico. Focusing on BI-RADS classification and biopsy recommendations, the system achieves high-quality text generation (BLEU, ROUGE-L, METEOR up to 0.98) and substantial agreement with expert assessments (Cohen's kappa of 0.93). Performance remains fair to moderate for more complex categories. The system shows strong potential as a prioritization tool in screening campaigns, particularly in underserved regions, and offers a scalable solution adaptable to other diagnostic imaging tasks.
- 11 June 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in TechRxiv
Open AccessA ConvNeXt-Transformer Approach for Automated Conclusion Generation from Mammography
Breast cancer remains a critical global health concern, particularly in low-resource settings where limited access to radiologists delays diagnosis. While digital mammography enables early detection, its effectiveness is often constrained by radiological expertise shortages. This study proposes a multimodal deep learning system to automate the generation of radiology report conclusions from mammogram images, supporting diagnostic decision-making and reducing radiologist workload. An encoder-decoder architecture is implemented, combining a ConvNeXt model for image feature extraction with a Transformer-based decoder for text generation. The model is trained and evaluated on the RSNA Screening Mammography dataset and a curated dataset from Tec Salud in Mexico. Focusing on BI-RADS classification and biopsy recommendations, the system achieves high-quality text generation (BLEU, ROUGE-L, METEOR up to 0.98) and substantial agreement with expert assessments (Cohen's kappa of 0.93). Performance remains fair to moderate for more complex categories. The system shows strong potential as a prioritization tool in screening campaigns, particularly in underserved regions, and offers a scalable solution adaptable to other diagnostic imaging tasks.
- Open Access
- 30 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 9265-9275
Multiview Multimodal Feature Fusion for Breast Cancer Classification Using Deep Learning
The increasing incidence and mortality of breast cancer pose significant global challenges for women. Deep learning (DL) has shown superior diagnostic performance in breast cancer classification compared to human experts. However, most DL methods have relied on unimodal features, which may limit the performance of diagnostic models. Recent studies focus on multimodal data along with multiple views of mammograms, typically two: Cranio-Caudal (CC) and Medio-Lateral-Oblique (MLO). Combining multimodal data has shown improvements in classification effectiveness over single-modal systems. In this study, we compiled a multimodal dataset comprising imaging and textual data (combination of clinical and radiological features). We propose a DL-based multiview multimodal feature fusion (MMFF) strategy for breast cancer classification that utilizes images (four views of mammograms) and tabular data (extracted from radiological reports) from our newly developed in-house dataset. Various augmentation techniques are applied to both imaging and textual data to expand the training dataset size. Imaging features were extracted using a Squeeze-and-Excitation (SE) network-based ResNet50 model, while textual features were extracted using an artificial neural network (ANN). Afterwards, extracted features from both modalities were fused using a late feature fusion strategy. Finally, fused features were fed into the ANN for the final classification of breast cancer. In our study, we compared the performance of our proposed MMFF model with single-modal models (image only) and models built on textual data. The performance was evaluated using accuracy, precision, sensitivity, F1 score and area under the receiver operating characteristic curve (AUC) metrics. Our model MMFF achieved an AUC of 0.965 for benign vs malignant classification as compared to image-only (ResNet50 = 0.545), text-only (ANN = 0.688, SVM = 0.842) and other multimodal approaches (ResNet50+ANN = 0.748, EfficientNetb7+ANN = 0.874).
- 30 December 2024
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Access
- Vol. 13, 9265-9275
Open AccessMultiview Multimodal Feature Fusion for Breast Cancer Classification Using Deep Learning
The increasing incidence and mortality of breast cancer pose significant global challenges for women. Deep learning (DL) has shown superior diagnostic performance in breast cancer classification compared to human experts. However, most DL methods have relied on unimodal features, which may limit the performance of diagnostic models. Recent studies focus on multimodal data along with multiple views of mammograms, typically two: Cranio-Caudal (CC) and Medio-Lateral-Oblique (MLO). Combining multimodal data has shown improvements in classification effectiveness over single-modal systems. In this study, we compiled a multimodal dataset comprising imaging and textual data (combination of clinical and radiological features). We propose a DL-based multiview multimodal feature fusion (MMFF) strategy for breast cancer classification that utilizes images (four views of mammograms) and tabular data (extracted from radiological reports) from our newly developed in-house dataset. Various augmentation techniques are applied to both imaging and textual data to expand the training dataset size. Imaging features were extracted using a Squeeze-and-Excitation (SE) network-based ResNet50 model, while textual features were extracted using an artificial neural network (ANN). Afterwards, extracted features from both modalities were fused using a late feature fusion strategy. Finally, fused features were fed into the ANN for the final classification of breast cancer. In our study, we compared the performance of our proposed MMFF model with single-modal models (image only) and models built on textual data. The performance was evaluated using accuracy, precision, sensitivity, F1 score and area under the receiver operating characteristic curve (AUC) metrics. Our model MMFF achieved an AUC of 0.965 for benign vs malignant classification as compared to image-only (ResNet50 = 0.545), text-only (ANN = 0.688, SVM = 0.842) and other multimodal approaches (ResNet50+ANN = 0.748, EfficientNetb7+ANN = 0.874).
- Open Access
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 31-39
Comparative Analysis of Deep Learning Models for Breast Cancer Classification on Multimodal Data
Rising breast cancer incidence and mortality represent significant global challenges for women. Deep learning has demonstrated superior diagnostic performance in breast cancer classification compared to human experts. However, most deep learning methods have relied on unimodal features, potentially limiting the performance of diagnostic models. Additionally, most studies conducted so far have used a single view of digital mammograms, which significantly reduces model performance due to limited overall perspective and generalizability. To address these limitations, we collected a multiview multimodal dataset, including digital mammograms four views two craniocaudal (CC), two mediolateral oblique (MLO) one for each breast, and textual data extracted from radiological reports. We propose a multimodal deep learning architecture for breast cancer classification, utilizing images (digital mammograms) and textual data (radiological reports) from our new in-house dataset. In addition, various augmentation techniques are applied to both imaging and textual data to enhance the training data size. In our investigation, we explored the performance of six state-of-the-art (SOTA) deep learning architectures: VGG16, VGG19, ResNet34, MobileNetV3, EfficientNetB7, and a vision transformer (ViT) as an imaging feature extractors. For textual feature extraction, we employed an artificial neural network (ANN). Afterwards, features were fused using an early fusion and late fusion strategy. The fused imaging and textual features were then inputted into an ANN classifier for breast cancer classification. We evaluated various feature extractors and an ANN classifier combinations, finding that VGG19 in association with ANN achieved the highest accuracy at 0.951. In terms of precision, again VGG19 and ANN combination surpassed other SOTA CNN and attention-based architectures, achieving a score of 0.95. The best sensitivity score of 0.893 was recorded by VGG16+ANN, followed by VGG19+ANN with 0.884. The highest F1 score of 0.922 was achieved by VGG19+ANN. VGG16+ANN achieved the best area under the curve (AUC) score of 0.929, closely followed by VGG19+ANN with a score of 0.915.
- 28 October 2024
- Published by Association for Computing Machinery (ACM)
- p. 31-39
Open AccessComparative Analysis of Deep Learning Models for Breast Cancer Classification on Multimodal Data
Rising breast cancer incidence and mortality represent significant global challenges for women. Deep learning has demonstrated superior diagnostic performance in breast cancer classification compared to human experts. However, most deep learning methods have relied on unimodal features, potentially limiting the performance of diagnostic models. Additionally, most studies conducted so far have used a single view of digital mammograms, which significantly reduces model performance due to limited overall perspective and generalizability. To address these limitations, we collected a multiview multimodal dataset, including digital mammograms four views two craniocaudal (CC), two mediolateral oblique (MLO) one for each breast, and textual data extracted from radiological reports. We propose a multimodal deep learning architecture for breast cancer classification, utilizing images (digital mammograms) and textual data (radiological reports) from our new in-house dataset. In addition, various augmentation techniques are applied to both imaging and textual data to enhance the training data size. In our investigation, we explored the performance of six state-of-the-art (SOTA) deep learning architectures: VGG16, VGG19, ResNet34, MobileNetV3, EfficientNetB7, and a vision transformer (ViT) as an imaging feature extractors. For textual feature extraction, we employed an artificial neural network (ANN). Afterwards, features were fused using an early fusion and late fusion strategy. The fused imaging and textual features were then inputted into an ANN classifier for breast cancer classification. We evaluated various feature extractors and an ANN classifier combinations, finding that VGG19 in association with ANN achieved the highest accuracy at 0.951. In terms of precision, again VGG19 and ANN combination surpassed other SOTA CNN and attention-based architectures, achieving a score of 0.95. The best sensitivity score of 0.893 was recorded by VGG16+ANN, followed by VGG19+ANN with 0.884. The highest F1 score of 0.922 was achieved by VGG19+ANN. VGG16+ANN achieved the best area under the curve (AUC) score of 0.929, closely followed by VGG19+ANN with a score of 0.915.
- 9 October 2024
- Published by Springer Nature
- p. 59-69
Performance Evaluation of Deep Learning and Transformer Models Using Multimodal Data for Breast Cancer Classification
- 9 October 2024
- Published by Springer Nature
- p. 59-69
Performance Evaluation of Deep Learning and Transformer Models Using Multimodal Data for Breast Cancer Classification
- Open Access
- 18 December 2023
- Published by Frontiers Media SA in Frontiers in Neurology
- Vol. 14, 1282833
Clinical performance of a multiparametric MRI-based post concussive syndrome index
Introduction: Diffusion Tensor Imaging (DTI) has revealed measurable changes in the brains of patients with persistent post-concussive syndrome (PCS). Because of inconsistent results in univariate DTI metrics among patients with mild traumatic brain injury (mTBI), there is currently no single objective and reliable MRI index for clinical decision-making in patients with PCS.Purpose: This study aimed to evaluate the performance of a newly developed PCS Index (PCSI) derived from machine learning of multiparametric magnetic resonance imaging (MRI) data to classify and differentiate subjects with mTBI and PCS history from those without a history of mTBI.Materials and methods: Data were retrospectively extracted from 139 patients aged between 18 and 60 years with PCS who underwent MRI examinations at 2 weeks to 1-year post-mTBI, as well as from 336 subjects without a history of head trauma. The performance of the PCS Index was assessed by comparing 69 patients with a clinical diagnosis of PCS with 264 control subjects. The PCSI values for patients with PCS were compared based on the mechanism of injury, time interval from injury to MRI examination, sex, history of prior concussion, loss of consciousness, and reported symptoms.Results: Injured patients had a mean PCSI value of 0.57, compared to the control group, which had a mean PCSI value of 0.12 (p = 8.42e-23) with accuracy of 88%, sensitivity of 64%, and specificity of 95%, respectively. No statistically significant differences were found in the PCSI values when comparing the mechanism of injury, sex, or loss of consciousness.Conclusion: The PCSI for individuals aged between 18 and 60 years was able to accurately identify patients with post-concussive injuries from 2 weeks to 1-year post-mTBI and differentiate them from the controls. The results of this study suggest that multiparametric MRI-based PCSI has great potential as an objective clinical tool to support the diagnosis, treatment, and follow-up care of patients with post-concussive syndrome. Further research is required to investigate the replicability of this method using other types of clinical MRI scanners.
- 18 December 2023
- Published by Frontiers Media SA in Frontiers in Neurology
- Vol. 14, 1282833
Open AccessClinical performance of a multiparametric MRI-based post concussive syndrome index
Introduction: Diffusion Tensor Imaging (DTI) has revealed measurable changes in the brains of patients with persistent post-concussive syndrome (PCS). Because of inconsistent results in univariate DTI metrics among patients with mild traumatic brain injury (mTBI), there is currently no single objective and reliable MRI index for clinical decision-making in patients with PCS.Purpose: This study aimed to evaluate the performance of a newly developed PCS Index (PCSI) derived from machine learning of multiparametric magnetic resonance imaging (MRI) data to classify and differentiate subjects with mTBI and PCS history from those without a history of mTBI.Materials and methods: Data were retrospectively extracted from 139 patients aged between 18 and 60 years with PCS who underwent MRI examinations at 2 weeks to 1-year post-mTBI, as well as from 336 subjects without a history of head trauma. The performance of the PCS Index was assessed by comparing 69 patients with a clinical diagnosis of PCS with 264 control subjects. The PCSI values for patients with PCS were compared based on the mechanism of injury, time interval from injury to MRI examination, sex, history of prior concussion, loss of consciousness, and reported symptoms.Results: Injured patients had a mean PCSI value of 0.57, compared to the control group, which had a mean PCSI value of 0.12 (p = 8.42e-23) with accuracy of 88%, sensitivity of 64%, and specificity of 95%, respectively. No statistically significant differences were found in the PCSI values when comparing the mechanism of injury, sex, or loss of consciousness.Conclusion: The PCSI for individuals aged between 18 and 60 years was able to accurately identify patients with post-concussive injuries from 2 weeks to 1-year post-mTBI and differentiate them from the controls. The results of this study suggest that multiparametric MRI-based PCSI has great potential as an objective clinical tool to support the diagnosis, treatment, and follow-up care of patients with post-concussive syndrome. Further research is required to investigate the replicability of this method using other types of clinical MRI scanners.
- 7 April 2023
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12465, 86-591
Glaucoma classification using a morphological-convolutional neural network trained with extreme learning machine
Mathematical morphology is a technique frequently used in image processing, it has several applications such as segmentation, filtering, compression, edge detection, and feature extraction. Considering this last application here is presented a morphological and convolutional neural network (MCNN) that takes advantage of the different types of morphological operations – erosion, dilation, opening, and closing – by including them in a single layer. Three independent neural networks are used to learn information per channel, Random Forest is used at the end, fed with the three outputs of the NNs. The classification performance of the method was compared against three common CNN architectures: ResNet-18, ShuffleNet-V2, and MobileNet-V2. Two training approaches were used: training from scratch and using transfer learning. A glaucoma classification was conducted using the ORIGA dataset. The MCNN method obtained an AUC of 0.704 (0.672, 0.743 95% CI) with a performance similar to the other CNN methods when trained using transfer learning.
- 7 April 2023
- Published by SPIE-Intl Soc Optical Eng
- Vol. 12465, 86-591
Glaucoma classification using a morphological-convolutional neural network trained with extreme learning machine
Mathematical morphology is a technique frequently used in image processing, it has several applications such as segmentation, filtering, compression, edge detection, and feature extraction. Considering this last application here is presented a morphological and convolutional neural network (MCNN) that takes advantage of the different types of morphological operations – erosion, dilation, opening, and closing – by including them in a single layer. Three independent neural networks are used to learn information per channel, Random Forest is used at the end, fed with the three outputs of the NNs. The classification performance of the method was compared against three common CNN architectures: ResNet-18, ShuffleNet-V2, and MobileNet-V2. Two training approaches were used: training from scratch and using transfer learning. A glaucoma classification was conducted using the ORIGA dataset. The MCNN method obtained an AUC of 0.704 (0.672, 0.743 95% CI) with a performance similar to the other CNN methods when trained using transfer learning.
- Open Access
- 28 February 2023
- Published by Springer Nature in Research Square
Iterative Decorrelation Analysis, Unit of Measure Preserving Transformations and Latent Biomarker Discovery
Background Numerous biomarker discovery studies and exploratory clinical studies extract a large set of measurable variables, which often have varying degrees of correlation among them. This data collinearity can impact statistical model interpretation and hinder the discovery of potential associations between measured variables and the observed outcome. Exploratory Factor Analysis (EFA), Principal Component Analysis (PCA), and Machine-Learning (ML) can be used to discover latent variables associated with disease progression or outcome by computing transformation matrices, but the interpretation of unsupervised/supervised latent variables in high-dimensional datasets can be challenging. Results This study describe and reports the performance of the iterative decorrelation analysis algorithm (IDeA). The algorithm iteratively analyzes the correlation matrix of the data, updating the transformation coefficients until it reaches the desired correlation goal. The output of IDeA is a basis-transformation matrix that preserves the data dimensionality and unit of measure of the original observed variables. The main advantages of the IDeA basis transformations are sparsity and interpretability. The transformation does not alter uncorrelated features, thus statistical modeling and biomarker discovery in the new transformed basis can be a combination of novel latent variables and a sizable subset of unaltered variables. The algorithm was tested on five multidimensional/hyperdimensional and multimodal sets, demonstrating the effect of decorrelation parameters, interpretability, and latent biomarker discovery. Conclusions The iterative decorrelation approach provides a simple to use tool for researchers to explore the association between correlated features in hyperdimensional/multimodal settings and to decorrelate significant associations via latent variables that preserve the unit of measurement. An efficient computer implementation of IDeA is available in the FRESA.CAD R package (https://cran.r-project.org/web/packages/FRESA.CAD/index.html).
- 28 February 2023
- Published by Springer Nature in Research Square
Open AccessIterative Decorrelation Analysis, Unit of Measure Preserving Transformations and Latent Biomarker Discovery
Background Numerous biomarker discovery studies and exploratory clinical studies extract a large set of measurable variables, which often have varying degrees of correlation among them. This data collinearity can impact statistical model interpretation and hinder the discovery of potential associations between measured variables and the observed outcome. Exploratory Factor Analysis (EFA), Principal Component Analysis (PCA), and Machine-Learning (ML) can be used to discover latent variables associated with disease progression or outcome by computing transformation matrices, but the interpretation of unsupervised/supervised latent variables in high-dimensional datasets can be challenging. Results This study describe and reports the performance of the iterative decorrelation analysis algorithm (IDeA). The algorithm iteratively analyzes the correlation matrix of the data, updating the transformation coefficients until it reaches the desired correlation goal. The output of IDeA is a basis-transformation matrix that preserves the data dimensionality and unit of measure of the original observed variables. The main advantages of the IDeA basis transformations are sparsity and interpretability. The transformation does not alter uncorrelated features, thus statistical modeling and biomarker discovery in the new transformed basis can be a combination of novel latent variables and a sizable subset of unaltered variables. The algorithm was tested on five multidimensional/hyperdimensional and multimodal sets, demonstrating the effect of decorrelation parameters, interpretability, and latent biomarker discovery. Conclusions The iterative decorrelation approach provides a simple to use tool for researchers to explore the association between correlated features in hyperdimensional/multimodal settings and to decorrelate significant associations via latent variables that preserve the unit of measurement. An efficient computer implementation of IDeA is available in the FRESA.CAD R package (https://cran.r-project.org/web/packages/FRESA.CAD/index.html).