Carlos A. Peña-Solorzano
Current Organizations / Institutes
The organizations/institutes from the scholar's latest article
Previous Organizations / Institutes
The organizations / institutes from the scholar's other articles in the past ten years
Research Subjects
Carlos A. Peña-Solorzano has a total of 67 co-authors. Here are the top co-authors by number of shared publications:
- Chun Fung Kwok
St. Vincent’s Institute of Medical Research, Melbourne 3065, Australia
6 shared publications - Michael S. Elliott
St Vincent's Institute of Medical Research, Melbourne, VIC, Australia
6 shared publications - Chong Wang
Stanford University
5 shared publications - Davis James McCarthy
Bioinformatics and Cellular Genomics, St. Vincent’s Institute of Medical Research, Fitzroy, Australia
4 shared publications - Gustavo Carneiro
University of Surrey
4 shared publications - Helen Frazer
St Vincent's BreastScreen St Vincent's Hospital Melbourne Melbourne Victoria Australia
4 shared publications
Latest Publications
- 7 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (8) , 5687-5699
Progressive Mining and Dynamic Distillation of Hierarchical Prototypes for Disease Classification and Localisation
Constructing effective representation of lesions is essential for disease classification and localization in medical image analysis. Prototype-based models address this by leveraging visual prototypes to capture representative lesion patterns, yet effectively handling the complexity of diverse lesion characteristics remains a critical challenge, as they typically rely on single-level, fixedsize prototypes and suffer from prototype redundancy. In this paper, we present HierProtoPNet, a new prototypebased framework designed to handle the complexity of lesions in medical images. HierProtoPNet leverages hierarchical visual prototypes across different semantic feature granularities to effectively capture diverse lesion patterns. To prevent redundancy and increase utility of the prototypes, we devise a novel prototype mining paradigm to progressively discover semantically distinct prototypes, offering multi-level complementary analysis of lesions. Also, we introduce a dynamic knowledge distillation strategy that allows transferring essential classification information across hierarchical levels, thereby improving generalisation performance. Comprehensive experiments show that HierProtoPNet achieves state-of-the-art classification performances in three benchmarks: binary breast cancer screening, multi-class retinal disease diagnosis, and multilabel chest X-ray classification. Quantitative assessments also illustrate HierProtoPNet's significant advantages in weakly-supervised disease localisation and segmentation.
- 7 April 2025
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Journal of Biomedical and Health Informatics
- Vol. 29 (8) , 5687-5699
Progressive Mining and Dynamic Distillation of Hierarchical Prototypes for Disease Classification and Localisation
Constructing effective representation of lesions is essential for disease classification and localization in medical image analysis. Prototype-based models address this by leveraging visual prototypes to capture representative lesion patterns, yet effectively handling the complexity of diverse lesion characteristics remains a critical challenge, as they typically rely on single-level, fixedsize prototypes and suffer from prototype redundancy. In this paper, we present HierProtoPNet, a new prototypebased framework designed to handle the complexity of lesions in medical images. HierProtoPNet leverages hierarchical visual prototypes across different semantic feature granularities to effectively capture diverse lesion patterns. To prevent redundancy and increase utility of the prototypes, we devise a novel prototype mining paradigm to progressively discover semantically distinct prototypes, offering multi-level complementary analysis of lesions. Also, we introduce a dynamic knowledge distillation strategy that allows transferring essential classification information across hierarchical levels, thereby improving generalisation performance. Comprehensive experiments show that HierProtoPNet achieves state-of-the-art classification performances in three benchmarks: binary breast cancer screening, multi-class retinal disease diagnosis, and multilabel chest X-ray classification. Quantitative assessments also illustrate HierProtoPNet's significant advantages in weakly-supervised disease localisation and segmentation.
- Open Access
- 30 August 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-12
Comparison of AI-integrated pathways with human-AI interaction in population mammographic screening for breast cancer
Artificial intelligence (AI) readers of mammograms compare favourably to individual radiologists in detecting breast cancer. However, AI readers cannot perform at the level of multi-reader systems used by screening programs in countries such as Australia, Sweden, and the UK. Therefore, implementation demands human-AI collaboration. Here, we use a large, high-quality retrospective mammography dataset from Victoria, Australia to conduct detailed simulations of five potential AI-integrated screening pathways, and examine human-AI interaction effects to explore automation bias. Operating an AI reader as a second reader or as a high confidence filter improves current screening outcomes by 1.9–2.5% in sensitivity and up to 0.6% in specificity, achieving 4.6–10.9% reduction in assessments and 48–80.7% reduction in human reads. Automation bias degrades performance in multi-reader settings but improves it for single-readers. This study provides insight into feasible approaches for AI-integrated screening pathways and prospective studies necessary prior to clinical adoption.
- 30 August 2024
- Published by Springer Nature in Nature Communications
- Vol. 15 (1) , 1-12
Open AccessComparison of AI-integrated pathways with human-AI interaction in population mammographic screening for breast cancer
Artificial intelligence (AI) readers of mammograms compare favourably to individual radiologists in detecting breast cancer. However, AI readers cannot perform at the level of multi-reader systems used by screening programs in countries such as Australia, Sweden, and the UK. Therefore, implementation demands human-AI collaboration. Here, we use a large, high-quality retrospective mammography dataset from Victoria, Australia to conduct detailed simulations of five potential AI-integrated screening pathways, and examine human-AI interaction effects to explore automation bias. Operating an AI reader as a second reader or as a high confidence filter improves current screening outcomes by 1.9–2.5% in sensitivity and up to 0.6% in specificity, achieving 4.6–10.9% reduction in assessments and 48–80.7% reduction in human reads. Automation bias degrades performance in multi-reader settings but improves it for single-readers. This study provides insight into feasible approaches for AI-integrated screening pathways and prospective studies necessary prior to clinical adoption.
- 10 May 2024
- Published by Elsevier in Medical Image Analysis
- Vol. 96, 103192
BRAIxDet: Learning to detect malignant breast lesion with incomplete annotations
- 10 May 2024
- Published by Elsevier in Medical Image Analysis
- Vol. 96, 103192
BRAIxDet: Learning to detect malignant breast lesion with incomplete annotations
- 21 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 43 (1) , 392-404
An Interpretable and Accurate Deep-Learning Diagnosis Framework Modeled With Fully and Semi-Supervised Reciprocal Learning
The deployment of automated deep-learning classifiers in clinical practice has the potential to streamline the diagnosis process and improve the diagnosis accuracy, but the acceptance of those classifiers relies on both their accuracy and interpretability. In general, accurate deep-learning classifiers provide little model interpretability, while interpretable models do not have competitive classification accuracy. In this paper, we introduce a new deep-learning diagnosis framework, called InterNRL, that is designed to be highly accurate and interpretable. InterNRL consists of a student-teacher framework, where the student model is an interpretable prototype-based classifier (ProtoPNet) and the teacher is an accurate global image classifier (GlobalNet). The two classifiers are mutually optimised with a novel reciprocal learning paradigm in which the student ProtoPNet learns from optimal pseudo labels produced by the teacher GlobalNet, while GlobalNet learns from ProtoPNet’s classification performance and pseudo labels. This reciprocal learning paradigm enables InterNRL to be flexibly optimised under both fully- and semi-supervised learning scenarios, reaching state-of-the-art classification performance in both scenarios for the tasks of breast cancer and retinal disease diagnosis. Moreover, relying on weakly-labelled training images, InterNRL also achieves superior breast cancer localisation and brain tumour segmentation results than other competing methods.
- 21 August 2023
- Published by Institute of Electrical and Electronics Engineers (IEEE) in IEEE Transactions on Medical Imaging
- Vol. 43 (1) , 392-404
An Interpretable and Accurate Deep-Learning Diagnosis Framework Modeled With Fully and Semi-Supervised Reciprocal Learning
The deployment of automated deep-learning classifiers in clinical practice has the potential to streamline the diagnosis process and improve the diagnosis accuracy, but the acceptance of those classifiers relies on both their accuracy and interpretability. In general, accurate deep-learning classifiers provide little model interpretability, while interpretable models do not have competitive classification accuracy. In this paper, we introduce a new deep-learning diagnosis framework, called InterNRL, that is designed to be highly accurate and interpretable. InterNRL consists of a student-teacher framework, where the student model is an interpretable prototype-based classifier (ProtoPNet) and the teacher is an accurate global image classifier (GlobalNet). The two classifiers are mutually optimised with a novel reciprocal learning paradigm in which the student ProtoPNet learns from optimal pseudo labels produced by the teacher GlobalNet, while GlobalNet learns from ProtoPNet’s classification performance and pseudo labels. This reciprocal learning paradigm enables InterNRL to be flexibly optimised under both fully- and semi-supervised learning scenarios, reaching state-of-the-art classification performance in both scenarios for the tasks of breast cancer and retinal disease diagnosis. Moreover, relying on weakly-labelled training images, InterNRL also achieves superior breast cancer localisation and brain tumour segmentation results than other competing methods.
- Open Access
- 1 March 2023
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 5 (2)
ADMANI: Annotated Digital Mammograms and Associated Non-Image Datasets
“Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review ...
- 1 March 2023
- Published by Radiological Society of North America (RSNA) in Radiology: Artificial Intelligence
- Vol. 5 (2)
Open AccessADMANI: Annotated Digital Mammograms and Associated Non-Image Datasets
“Just Accepted” papers have undergone full peer review and have been accepted for publication in Radiology: Artificial Intelligence. This article will undergo copyediting, layout, and proof review ...
- Open Access
Comparison of AI-integrated pathways with human-AI interaction for population mammographic screening
Background Artificial intelligence (AI) readers, derived from applying deep learning models to medical image analysis, hold great promise for improving population breast cancer screening. However, previous evaluations of AI readers for breast cancer screening have mostly been conducted using cancer-enriched cohorts and have lacked assessment of the potential use of AI readers alongside radiologists in multi-reader screening programs. Here, we present a new AI reader for detecting breast cancer from mammograms in a large-scale population screening setting, and a novel analysis of the potential for human-AI reader collaboration in a well-established, high-performing population screening program. We evaluated the performance of our AI reader and AI-integrated screening scenarios using a two-year, real-world, population dataset from Victoria, Australia, a screening program in which two radiologists independently assess each episode and disagreements are arbitrated by a third radiologist. Methods We used a retrospective full-field digital mammography image and non-image dataset comprising 808,318 episodes, 577,576 clients and 3,404,326 images in the period 2013 to 2019. Screening episodes from 2016, 2017 and 2018 were sequential population cohorts containing 752,609 episodes, 565,087 clients and 3,169,322 images. The dataset was split by screening date into training, development, and testing sets. All episodes from 2017 and 2018 were allocated to the testing set (509,109 episodes; 3,651 screen-detected cancer episodes). Eight distinct AI models were trained on subsets of the training set (which included a validation set) and combined into our ensemble AI reader. Operating points were selected using the development set. We evaluated our AI reader on our testing set and on external datasets previously unseen by our models. Findings The AI reader outperformed the mean individual radiologist on this large retrospective testing dataset with an area under the receiver operator characteristic curve of 0.92 (95% CI 0.91, 0.92). The AI reader generalised well across screening round, client demographics, device manufacturer and cancer type, and achieved state-of-the-art performance on external datasets compared to recently published AI readers. Our simulations of AI-integrated screening scenarios demonstrated that a reader-replacement human-AI collaborative system could achieve better sensitivity and specificity (82.6%, 96.1%) compared to the current two-reader consensus system (79.9%, 96.0%), with reduced human reading workload and cost. Our band-pass AI-integrated scenario also enabled both higher sensitivity and specificity (80.6%, 96.2%) with larger reductions in human reading workload and cost. Interpretation This study demonstrated that human-AI collaboration in a population breast cancer screening program has potential to improve accuracy and lower radiologist workload and costs in real world screening programs. The next stage of validation is to undertake prospective studies that can also assess the effects of the AI systems on human performance and behaviour.
Open AccessComparison of AI-integrated pathways with human-AI interaction for population mammographic screening
Background Artificial intelligence (AI) readers, derived from applying deep learning models to medical image analysis, hold great promise for improving population breast cancer screening. However, previous evaluations of AI readers for breast cancer screening have mostly been conducted using cancer-enriched cohorts and have lacked assessment of the potential use of AI readers alongside radiologists in multi-reader screening programs. Here, we present a new AI reader for detecting breast cancer from mammograms in a large-scale population screening setting, and a novel analysis of the potential for human-AI reader collaboration in a well-established, high-performing population screening program. We evaluated the performance of our AI reader and AI-integrated screening scenarios using a two-year, real-world, population dataset from Victoria, Australia, a screening program in which two radiologists independently assess each episode and disagreements are arbitrated by a third radiologist. Methods We used a retrospective full-field digital mammography image and non-image dataset comprising 808,318 episodes, 577,576 clients and 3,404,326 images in the period 2013 to 2019. Screening episodes from 2016, 2017 and 2018 were sequential population cohorts containing 752,609 episodes, 565,087 clients and 3,169,322 images. The dataset was split by screening date into training, development, and testing sets. All episodes from 2017 and 2018 were allocated to the testing set (509,109 episodes; 3,651 screen-detected cancer episodes). Eight distinct AI models were trained on subsets of the training set (which included a validation set) and combined into our ensemble AI reader. Operating points were selected using the development set. We evaluated our AI reader on our testing set and on external datasets previously unseen by our models. Findings The AI reader outperformed the mean individual radiologist on this large retrospective testing dataset with an area under the receiver operator characteristic curve of 0.92 (95% CI 0.91, 0.92). The AI reader generalised well across screening round, client demographics, device manufacturer and cancer type, and achieved state-of-the-art performance on external datasets compared to recently published AI readers. Our simulations of AI-integrated screening scenarios demonstrated that a reader-replacement human-AI collaborative system could achieve better sensitivity and specificity (82.6%, 96.1%) compared to the current two-reader consensus system (79.9%, 96.0%), with reduced human reading workload and cost. Our band-pass AI-integrated scenario also enabled both higher sensitivity and specificity (80.6%, 96.2%) with larger reductions in human reading workload and cost. Interpretation This study demonstrated that human-AI collaboration in a population breast cancer screening program has potential to improve accuracy and lower radiologist workload and costs in real world screening programs. The next stage of validation is to undertake prospective studies that can also assess the effects of the AI systems on human performance and behaviour.
- 1 November 2020
- Published by Elsevier in Forensic Science International
- Vol. 316, 110538
Findings from machine learning in clinical medical imaging applications – Lessons for translation to the forensic setting
- 1 November 2020
- Published by Elsevier in Forensic Science International
- Vol. 316, 110538
Findings from machine learning in clinical medical imaging applications – Lessons for translation to the forensic setting
- 1 July 2020
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 122, 103797
Semi-supervised labelling of the femur in a whole-body post-mortem CT database using deep learning
- 1 July 2020
- Published by Elsevier in Computers in Biology and Medicine
- Vol. 122, 103797
Semi-supervised labelling of the femur in a whole-body post-mortem CT database using deep learning
- 1 April 2019
- Published by Wiley in Medical Physics
- Vol. 46 (4) , 1766-1776
Development of a simple numerical model for trabecular bone structures
PurposeAdvances in additive manufacturing processes are enabling the fabrication of surrogate bone structures for applications including use in high-resolution anthropomorphic phantoms. In this research, a simple numerical model is proposed that enables the generation of microarchitecture with similar statistical distribution to trabecular bone. MethodsA human humerus, radius, ulna, and several vertebrae were scanned on the Imaging and Medical beamline at the Australian Synchrotron and the proposed numerical model was developed through the definition of two complex functions that encode the trabecular thickness and position-dependant spacing to generate volumetric surrogate trabecular structures. The structures reproduced those observed at 19 separate axial locations through the experimental bone volumes. The applicability of the model when incorporating a two-material approximation to absorption- and phase-contrast CT was also investigated through simulation. ResultsThe synthetic structures, when compared with the real trabecular microarchitecture, yielded an average mean thickness error of 2m, and a mean difference in standard deviation of 33m for the humerus, 24m for the ulna and radius, and 15m for the vertebrae. Simulated absorption- and propagation-based phase contrast CT projection data were generated and reconstructed using the derived mathematical simplifications from the two-material approximation, and the phase-contrast effects were successfully demonstrated. ConclusionsThe presented model reproduced trabecular distributions that could be used to generate phantoms for quality assurance and validation processes. The implication of utilizing a two-material approximation results in simplification of the additive manufacturing process and the generation of synthetic data that could be used for training of machine learning applications.
- 1 April 2019
- Published by Wiley in Medical Physics
- Vol. 46 (4) , 1766-1776
Development of a simple numerical model for trabecular bone structures
PurposeAdvances in additive manufacturing processes are enabling the fabrication of surrogate bone structures for applications including use in high-resolution anthropomorphic phantoms. In this research, a simple numerical model is proposed that enables the generation of microarchitecture with similar statistical distribution to trabecular bone. MethodsA human humerus, radius, ulna, and several vertebrae were scanned on the Imaging and Medical beamline at the Australian Synchrotron and the proposed numerical model was developed through the definition of two complex functions that encode the trabecular thickness and position-dependant spacing to generate volumetric surrogate trabecular structures. The structures reproduced those observed at 19 separate axial locations through the experimental bone volumes. The applicability of the model when incorporating a two-material approximation to absorption- and phase-contrast CT was also investigated through simulation. ResultsThe synthetic structures, when compared with the real trabecular microarchitecture, yielded an average mean thickness error of 2m, and a mean difference in standard deviation of 33m for the humerus, 24m for the ulna and radius, and 15m for the vertebrae. Simulated absorption- and propagation-based phase contrast CT projection data were generated and reconstructed using the derived mathematical simplifications from the two-material approximation, and the phase-contrast effects were successfully demonstrated. ConclusionsThe presented model reproduced trabecular distributions that could be used to generate phantoms for quality assurance and validation processes. The implication of utilizing a two-material approximation results in simplification of the additive manufacturing process and the generation of synthetic data that could be used for training of machine learning applications.
- Open Access
- 15 September 2018
- Published by Jaypee Brothers Medical Publishing in Strategies in Trauma and Limb Reconstruction
- Vol. 13 (3) , 137-149
Effect of external fixation rod coupling in computed tomography
External fixation is a common tool in the treatment of complex fractures, correction of limb deformity, and salvage arthrodesis. These devices typically incorporate radio-opaque metal rods/struts connected at varying distances and orientations between rings. Whilst the predominant imaging modality is plain film radiology, computed tomography (CT) may be performed in order for the surgeon to make a more confident clinical decision (e.g. timing of frame removal, assessment of degree of arthrodesis). We used a fractured sheep leg to systematically assess CT imaging performance with a Discovery CT750 HD CT scanner (GE Healthcare) to show how rod coupling in both traditional Ilizarov and hexapod frames distorts images. We also investigated the role of dual-energy CT (DECT) and metal artefact reduction software (MARS) on the visualisation of the fractured leg. Whilst mechanical reasons predominantly dictate the rod/strut configurations when building a circular frame, rod coupling in CT can be minimised. Firstly, ideally, all or all but one rod can be removed during imaging resulting in no rod coupling. If this is not possible, strategies for configuring the rods to minimise the effect of the rod coupling on the region of interest are demonstrated, e.g., in the case of a four-rod construct, switching the two anterior rods to a more central single one will achieve this goal without particularly jeopardising mechanical strength for a short period. It is also shown that the addition of DECT and MARS results in a reduction of artefacts, but also affects tissue and bone differentiation.
- 15 September 2018
- Published by Jaypee Brothers Medical Publishing in Strategies in Trauma and Limb Reconstruction
- Vol. 13 (3) , 137-149
Open AccessEffect of external fixation rod coupling in computed tomography
External fixation is a common tool in the treatment of complex fractures, correction of limb deformity, and salvage arthrodesis. These devices typically incorporate radio-opaque metal rods/struts connected at varying distances and orientations between rings. Whilst the predominant imaging modality is plain film radiology, computed tomography (CT) may be performed in order for the surgeon to make a more confident clinical decision (e.g. timing of frame removal, assessment of degree of arthrodesis). We used a fractured sheep leg to systematically assess CT imaging performance with a Discovery CT750 HD CT scanner (GE Healthcare) to show how rod coupling in both traditional Ilizarov and hexapod frames distorts images. We also investigated the role of dual-energy CT (DECT) and metal artefact reduction software (MARS) on the visualisation of the fractured leg. Whilst mechanical reasons predominantly dictate the rod/strut configurations when building a circular frame, rod coupling in CT can be minimised. Firstly, ideally, all or all but one rod can be removed during imaging resulting in no rod coupling. If this is not possible, strategies for configuring the rods to minimise the effect of the rod coupling on the region of interest are demonstrated, e.g., in the case of a four-rod construct, switching the two anterior rods to a more central single one will achieve this goal without particularly jeopardising mechanical strength for a short period. It is also shown that the addition of DECT and MARS results in a reduction of artefacts, but also affects tissue and bone differentiation.