CoChat
Sign Up
Cited by 713Artificial Intelligence in Healthcare and EducationMedical Image Segmentation TechniquesRadiomics and Machine Learning in Medical Imaging

AI in Medical Imaging Informatics: Current Challenges and Future Directions

Open access
Panayides, Andreas S., Amini, Amir, Filipovic, Nenad D., Sharma, Ashish, Tsaftaris, Sotirios A., Young, Alistair, Foran, David, Do, Nhan +8 more·2020·IEEE Journal of Biomedical and Health Informatics
DOI 10.1109/jbhi.2020.2991043PMC PMC8580417
Open this paper in CoChat — free accountOpen PDF

Abstract

This paper reviews state-of-the-art research solutions across the spectrum of medical imaging informatics, discusses clinical translation, and provides future directions for advancing clinical practice. More specifically, it summarizes advances in medical imaging acquisition technologies for different modalities, highlighting the necessity for efficient medical data management strategies in the context of AI in big healthcare data analytics. It then provides a synopsis of contemporary and emerging algorithmic methods for disease classification and organ/ tissue segmentation, focusing on AI and deep learning architectures that have already become the de facto approach. The clinical benefits of in-silico modelling advances linked with evolving 3D reconstruction and visualization applications are further documented. Concluding, integrative analytics approaches driven by associate research branches highlighted in this study promise to revolutionize imaging informatics as known today across the healthcare continuum for both radiology and digital pathology applications. The latter, is projected to enable informed, more accurate diagnosis, timely prognosis, and effective treatment planning, underpinning precision medicine.

AI snapshot

Written from the paper’s full text.

TL;DR

This review surveys AI and deep learning across medical imaging informatics, from acquisition and data repositories to radiology, digital pathology, visualization, in-silico modeling, and radiogenomics. It argues that integrative, standards-based, explainable AI is needed to translate these advances into precision medicine.

Key findings

  • Deep learning has become the de facto approach for medical image segmentation and classification; U-Net and 3D U-Net are highlighted as highly successful architectures, and transfer learning helps when medical training data are limited.
  • Data infrastructure is a central bottleneck: the paper calls for enterprise-wide clinical data repositories, interoperability standards (IHE, HL7 FHIR, DICOM, SNOMED CT, ICD), FAIR principles, and privacy-preserving publishing; TCIA/TCGA provide matched imaging, genetic, and clinical data for over 20 cancer types.
  • Radiogenomics examples include a breast phenotype group defining 38 radiomics features in six categories and the glioma VASARI schema with 30 descriptive features; deep-learning radiogenomics is described as still in its infancy.
  • Digital pathology has advanced to whole-slide imaging, with FDA approval in 2017 of a commercial digital pathology system; integrative image-plus-genomic analyses are reported to improve prognosis prediction, including in 86 breast cancer cases from the Genomics Data Commons.
  • Acquisition scale and performance are noted: 36 million MRI scans and about 74 million CT studies in the US in 2017; CT spatial resolution down to 0.25 mm and slice thickness down to 0.625 mm; X-ray resolution down to 100 microns; light microscopy up to 1,000x and approximately 0.2 μm resolution.

Methods

  • Review article; no primary data, sample, or experimental design are reported.
  • The paper narratively synthesizes literature across image acquisition, data management, radiology AI, digital pathology, 3D visualization, in-silico modeling, and integrative analytics.
  • Main analysis is a qualitative overview with selected examples, tables, and references; no systematic search strategy, quality appraisal, or meta-analysis is stated.

Limitations

  • Stated: challenges remain in explainable AI, generalization across populations, small training datasets, data heterogeneity and curation, batch effects in pathology, and clinical translation.
  • Stated: in-silico tumor models have reduced ability to simulate specific cellular factors such as cell-to-cell adhesion and subcellular-scale processes; many TMA analysis systems are closed, proprietary, and/or non-standard.
  • Evident: as a narrative review, it does not provide a systematic search protocol, risk-of-bias assessment, or quantitative synthesis, and its claims are not validated with primary data.

Model-written summary. Check it against the paper before relying on it.

Integrity signals

Registry

Crossref · 10.1109/jbhi.2020.2991043

Cites

262 of 262 references verified0 citation errors · 0 not checkable

Statistics

Nothing to recompute.

Claims the paper makes

0 of 82 cited sources checked · 0 supported · 0 flagged

  • Findingreasoning

    Heterogeneous healthcare data (imaging, genomic, clinical, environmental, physiological signals, etc.) cannot currently be used for integrative analysis and new hypothesis testing.

    “In its present form, heterogeneous healthcare data ranging from imaging, to genomic, to clinical data, that are further augmented by environmental data, physiological signals and other, cannot be used for integrative analysis (see Section VII) and new hypothesis testing.”
    Quote in the paperSubstantial claim
  • Findingnone given

    A significant challenge to establishing enterprise-wide EHRs is that a tremendous amount of clinical data are unstructured or semi-structured, with many reports generated at third-party laboratories.

    “One of the most significant challenges towards establishing enterprise-wide EHRs stems from the fact that a tremendous amount of clinical data are found in unstructured or semi-structured format with a significant number of reports generated at 3rd party laboratories.”
    Quote in the paperSubstantial claim
  • Findingcitation

    A recent analysis of challenge design showed that biases exist that question how easy it would be to translate methods to clinical practice.

    “however, a recent analysis of challenge design has showed that biases exist that questions how easy would be to translate methods to clinical practice [84].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [84] Maier-Hein L, “ BIAS: Transparent reporting of biomedical image analysis challenges,” 2019, arXiv:1910.04071v3. PMC7441980 32911207 source

      Not checked in the free run.

  • Findingnone given

    No single ad hoc (hand-crafted) feature analysis method has provided robust, cross-domain application solutions.

    “A number of ad hoc (or hand-crafted) feature analysis approaches have achieved some success in different applications, by explicitly defining a prior set of features and processing steps. However, no single method has provided robust, cross-domain application solutions.”
    Quote in the paperNot a substantial claim
  • Interpretationcitation

    Multi-institutional collaboration is becoming the vehicle driving the creation of well-curated and semantically annotated large cohorts enhanced with research methods and results metadata, underpinning reproducible, extensible, and explainable research.

    “Toward this direction, multi-institutional collaboration is quickly becoming the vehicle driving the creation of well-curated and semantically annotated large cohorts that are further enhanced with research methods and results metadata, underpinning reproducible, extensible, and explainable research [59], [60].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [59] Holzinger A, “ What do we need to build explainable AI systems for the medical domain?” 12. 2017, arXiv:1712.09923. source

      Not checked in the free run.

    • cites [60] Suinesiaputra A, “ Big heart data: Advancing health informatics through data sharing in cardiovascular imaging,” IEEE J. Biomed. Health Inform, vol. 19, no. 4, pp. 1283– 1290, 7. 2015 25415993 PMC4659497 source

      Not checked in the free run.

  • May combine two claims. 31
  • Classified as probably not a substantial claim. 41

  • May combine two claims.

    Findingreasoning

    Many clinical and research data management and exploration tools receive limited adoption because they require manual data entry and do not furnish tools for advanced queries.

    “Yet, many of them receive limited adoption from the clinical and research communities because they require manual data entry and do not furnish the necessary tools required to enable end-users to perform advanced queries.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingcitation

    Graph cut algorithms facilitate a global optimal solution.

    “On the other hand, graph cut algorithms facilitate a global optimal solution [86].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [86] Boykov Y and Funka-Lea G, “ Graph cuts and efficient ND image segmentation,” Int. J. Comput. Vis, pp. 109– 131, 11. 2006. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingnone given

    Active shape models are typically iterative and may get stuck in a local minimum.

    “These methods are typically iterative and may thus get stuck in a local minimum.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Findingnone given

    Machine learning methods are often difficult to engineer because they can fail in unpredictable ways and are subject to bias or spurious feature identification due to limitations in the training dataset.

    “However, these methods are often difficult to engineer since they can fail in unpredictable ways and are subject to bias or spurious feature identification due to limitations in the training dataset.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingcitation

    A CNN was used for lung nodule detection in chest x-rays as early as 1995.

    “One of the earliest applications of convolutional neural networks (CNN, the currently most common form of deep learning) has appeared as early as 1995, where a CNN was used for lung nodule detection in chest x-rays [100].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [100] Lo SCB, “ Artificial convolution neural network techniques and applications for lung nodule detection,” IEEE Trans. Med. Imag, vol. 14, no. 4, pp. 711– 718, 12. 1995. 18215875 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Medical image analysis has been hampered by a lack of theoretical understanding on how to optimally choose and process visual features.

    “To date, medical image analysis has been hampered by a lack of theoretical understanding on how to optimally choose and process visual features.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Machine learning approaches have provided good results in a wide range of applications.

    “The recent advent of machine learning approaches has provided good results in a wide range of applications.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingreasoning

    The current unusability of heterogeneous healthcare data for integrative analysis is attributed to data being scattered across and within institutions in a poorly indexed fashion, not openly available to the research community, not well-curated nor semantically annotated, and semi- or unstructured.

    “The latter is attributed to a number of factors, a non-exhaustive list extending to the data being scattered across and within institutions in a poorly indexed fashion, not being openly-available to the research community, and not being well-curated nor semantically annotated. Additionally, these data are typically semi- or un- structured, adding a significant computational burden for constituting”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Findingcitation

    Deep learning algorithms have been extensively used for disease classification or screening and have resulted in excellent performance in many tasks.

    “Deep learning algorithms have been extensively used for disease classification, or screening, and have resulted in excellent performance in many tasks (see Table II).”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Findingcitation

    Deep learning based segmentation of anatomy and pathology has witnessed a revolution, and for some tasks human level performance is now observed.

    “Since then, fueled by the revolutionary results of AlexNet [101] and incarnations of patch-based adaptations of Deep Boltzmann Machines and stacked autoencoders, deep learning based segmentation of anatomy and pathology has witnessed a revolution (see also Table II), where for some tasks now we observe human level performance [102].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [101] Krizhevsky A, Ilya S, and Geoffrey EH, “ Imagenet classification with deep convolutional neural networks.” in Proc. Adv. Neural Inf. Process. Syst, 2012, pp. 1097– 1105. source

      Not checked in the free run.

    • cites [102] Kamnitsas K, “ Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation,” Med. Image Anal, vol. 36, pp. 61– 78, 2017. 27865153 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Findingcitation

    U-Net is one of the most successful architectures for medical image segmentation.

    “The limitations of such approaches for medical image segmentation were quickly realised and led to the discovery of U-Net [104], which is even today one of the most successful architectures for medical image segmentation.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [104] Ronneberger O, Fischer P, and Brox T, “ U-net: Convolutional networks for biomedical image segmentation,” in Proc. Medical Image Comput. Comp.-Assis. Interv. –MICCAI, Navab N, Hornegger J, Wells W, Frangi A eds. Lecture Notes in Computer Science, vol. 9351, Cham: Springer, 2015. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Findingcitation

    The 3D U-net allowed full volumetric processing of imaging data while maintaining the same principles as the original U-net.

    “While the original U-Net was 2D, in 2016, the 3D U-net was proposed that allowed full volumetric processing of imaging data [105], maintaining the same principles of the original U-net.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [105] Çiçek Ö, “ 3D U-net: Learning dense volumetric segmentation from sparse annotation,” in Proc. Med. Image Comput. Comput.-Assisted Intervention, 2016, pp. 424– 432.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Findingnone given

    Classification tasks in medical imaging are dominated by some formulation of a CNN, often with fully-connected layers at the end to perform the final classification.

    “Overall, irrespective of the training strategy used, classification tasks in medical imaging are dominated by some formulation of a CNN – often with fully-connected layers at the end to perform the final classification.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Premisecitation

    AI and deep learning architectures have already become the de facto approach for disease classification and organ/tissue segmentation.

    “focusing on AI and deep learning architectures that have already become the de facto approach.”
    • Also stated in: § Deep Learning for Segmentation · boosted
      “fueled by the revolutionary results of AlexNet [101] and incarnations of patch-based adaptations of Deep Boltzmann Machines and stacked autoencoders, deep learning based segmentation of anatomy and pathology has witnessed a revolution (see also Table II), where for some tasks now we observe human level performance [102].”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [101] Krizhevsky A, Ilya S, and Geoffrey EH, “ Imagenet classification with deep convolutional neural networks.” in Proc. Adv. Neural Inf. Process. Syst, 2012, pp. 1097– 1105. source

      Not checked in the free run.

    • cites [102] Kamnitsas K, “ Efficient multi-scale 3D CNN with fully connected CRF for accurate brain lesion segmentation,” Med. Image Anal, vol. 36, pp. 61– 78, 2017. 27865153 source

      Not checked in the free run.

  • May combine two claims.

    Premisereasoning

    Automated extraction, transformation and load (ETL) interfaces can accommodate the full spectrum of clinical information, imaging studies and genomic information, enabling systematic interrogation of multi-modal data.

    “ETLs can accommodate the full spectrum of clinical information, imaging studies and genomic information.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationreasoning

    Multi-atlas segmentation adds robustness because errors associated with a particular atlas are averaged to form a maximum likelihood consensus.

    “This procedure adds robustness since errors associated with a particular atlas are averaged to form a maximum likelihood consensus.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Metadata underpin medical dataset sharing by providing descriptive information that characterizes the underlying data.

    “Metadata underpin medical dataset sharing by providing descriptive information that characterize the underlying data.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [50] Pezoulas V, Exarchos T, and Fotiadis DI, Medical Data Sharing, Harmonization and Analytics. New York, NY, USA: Academic Press, 2020.

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisereasoning

    Deep learning methods optimize simultaneously for the decision (classification or regression) while identifying and learning suitable input features, replacing feature engineering with a data-driven approach.

    “Deep learning methods instead optimize simultaneously for the decision (classification or regression) whilst identifying and learning suitable input features.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim.

    Interpretationcitation

    To achieve semantic interoperability, it is necessary to develop and optimize software that endorses and relies on interoperability profiles and standards (IHE, HL7 FHIR, DICOM) and clinical terminology coding (SNOMED CT, ICD).

    “To effectively incorporate this information into the EHRs and achieve semantic interoperability, it is necessary to develop and optimize software that endorses and relies on interoperability profiles and standards.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [48] Panayides AS, Pattichis MS, Pantziaris M, Constantinides AG, and Pattichis CS, “ The battle of the video codecs in the healthcare domain - a comparative performance evaluation study leveraging VVC and AV1,” IEEE Access, vol. 8, pp. 11469– 11481, 2020. source

      Not checked in the free run.

    • cites [49] Integrating the Healthcare Enterprise. [ Online]. Available: https://www.ihe.net/. Accessed: Feb. 2018.

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    Early machine learning approaches relied on hand-crafted features to reduce data dimensionality and induce necessary invariances and covariances.

    “To reduce data dimensionality and induce necessary invariances and covariances (e.g. robustness to intensity changes or scale) early machine learning approaches relied on hand-crafted features to represent data.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationcitation

    Sharing privacy-preserving data can increase the likelihood of novel findings and replication of existing research results.

    “Sharing such data can increase the likelihood of novel findings and replication of existing research results [54].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [54] Aristodimou A, Antoniades A, and Pattichis CS, “ Privacy preserving data publishing of categorical data through k-anonymity and feature selection,” Healthcare Technol. Lett, vol. 3, no. 1, pp. 16– 21, 2016. PMC4814843 27222728 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based). May combine two claims.

    Interpretationnone given

    Standardized data entry (acquisition) underpins extensibility, which in turn results in increased statistical power of research studies relying on larger cohorts.

    “Importantly, (new) data entry (acquisition) in a standardized fashion underpins extensibility that in turns results in increased statistical power of research studies relying on larger cohorts.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Interpretationnone given

    Open access challenges in which participants benchmark methods on standardized datasets are an important mechanism for advancing medical image analysis.

    “An important mechanism for advancing the field is by open access challenges in which participants can benchmark methods on standardized datasets.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisenone given

    Deep CNN interpretability is an emerging area of machine learning research targeting a better understanding of what the network has learned and how it derives its classification decisions.

    “Deep CNN interpretability is an emerging area of machine learning research targeting a better understanding of what the network has learned and how it derives its classification decisions.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Interpretationnone given

    The major draw of deep learning and convolutional architectures is the ability to learn suitable features and decision functions in tandem.

    “The major draw of deep learning and convolutional architectures is the ability to learn suitable features and decision functions in tandem.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationnone given

    U-Net's skip connections allow the model to be trained even with few input data and offer highly accurate segmentation boundaries.

    “The skip connections allow the model to be trained even with few input data and offer highly accurate segmentation boundaries, albeit perhaps at the “loss” of a clearly determined latent space.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    CycleGAN allows mapping of one image domain to another even without paired images.

    “CycleGAN allows mapping of one image domain to another image domain even without having pairs of images.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [107] Zhu J, “ Unpaired image-to-image translation using cycle-consistent adversarial networks,” in Proc. IEEE Int. Conf. Comput. Vision, 2017, pp. 2242– 2251.

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Application of imaging informatics methodologies in the cardiovascular system has produced important new knowledge and has improved understanding of normal function as well as of pathophysiology, diagnosis and treatment of cardiovascular disorders.

    “Moreover, application of imaging informatics methodologies in the cardiovascular system has produced important new knowledge and has improved our understanding of normal function as well as of pathophysiology, diagnosis and treatment of cardiovascular disorders [144].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [144] Cardiovascular Computing: Methodologies and Clinical Applications, Golemati S, Nikita KS Eds., Vienna, Austria: Springer, 2019.

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • May combine two claims.

    Premisenone given

    Deep CNNs have achieved extremely high accuracy but are still black-box functions with multiple layers of nonlinearities.

    “Although Deep CNNs have achieved extremely high accuracy, they are still black-box functions with multiple layers of nonlinearities.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Saliency maps and guided backpropagation are common approaches used to shed light on the predictions of Deep CNN.

    “Another common approach that is used to shed light on the predictions of Deep CNN is based on creating saliency maps [132] and guided backpropagation [133], [134].”
    Quote in the paperNot a substantial claim0 of 3 sources checked
    • cites [132] Simonyan K, Vedaldi A, and Zisserman A, “ Deep inside convolutional networks: Visualizing image classification models and saliency maps,” 2014, arXiv:1312.6034.v2. source

      Not checked in the free run.

    • cites [133] Zeiler MD and Fergus R, “ Visualizing and understanding convolutional networks,” in Proc. Eur. Conf. Comput. Vision, 9. 2014, pp. 818– 833. source

      Not checked in the free run.

    • cites [134] Springerberg JT, “ Striving for simplicity: The all convolutional net,” 2014, arXiv:1412.6806.v3. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The CAMELYON challenges target methods for algorithmic detection and classification of breast cancer metastases in H&E whole slide lymph node sections.

    “The CAMELYON challenges target methods for algorithmic detection and classification of breast cancer metastases in H&E whole slide lymph node sections [171].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [171] Bandi P, “ From detection of individual metastases to classification of lymph node status at the patient level: The CAMELYON17 challenge,” IEEE Trans. Med. Imag, vol. 38, no. 2, pp. 550– 560, 2. 2019. 30716025 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based).

    Premisenone given

    Whole slide images can contain hundreds of thousands or more cells and nuclei.

    “Whole slide images can contain hundreds of thousands or more cells and nuclei.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    In recent years, deep learning methods have rapidly grown in importance in pathology image analysis.

    “In recent years, deep learning methods have rapidly grown in importance in pathology image analysis [160].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [160] Janowczyk A and Madabhushi A, “ Deep learning for digital pathology image analysis: A comprehensive tutorial with selected use cases,” J. Pathol. Informat, vol. 7, no. 29, 2016. PMC4977982 27563488

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S4 (based).

    Premisenone given

    Tissue images contain a much denser amount of information than many other imaging modalities, encoded at multiple scales.

    “Tissue images contain a much denser amount of information than many other imaging modalities, encoded at multiple scales (pixels, objects such as nuclei and cells, and regions such as tumor and stromal tissue areas).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The FDA approved in 2017 the use of a commercial digital pathology system in clinical settings.

    “In light of advances in imaging instruments and software, the FDA approved in 2017 the use of a commercial digital pathology system in clinical settings [158].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [158] Evans AJ, “ US food and drug Administration approval of whole slide imaging for primary diagnosis: A key milestone is reached and new questions are raised,” Archives Pathol. Lab. Med, vol. 142, no. 11, pp. 1383– 1387, 2018. 29708429 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Routine availability of digitized pathology images, coupled with well-known issues associated with inter-observer variability in how pathologists interpret studies, has led to increased interest in computer-assisted decision support systems.

    “Routine availability of digitized pathology images, coupled with well-known issues associated with inter-observer variability in how pathologists interpret studies [159], has led to increased interest in computer-assisted decision support systems.”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [159] Matysiak BE, “ Simple, inexpensive method for automating tissue microarray production provides enhanced microarray reproducibility,” Appl Immunohistochem. Mol. Morphol, vol. 11, no. 3, pp. 269– 273, 2003. 12966355 source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    These datasets can be correlated to molecular tumor characteristics and can be used to quantitatively characterize tissue at multiple spatial scales to create biomarkers that predict outcome and treatment response.

    “These datasets can be correlated to molecular tumor characteristics and can be used to quantitatively characterize tissue at multiple spatial scales to create biomarkers that predict outcome and treatment response [150], [152]–[154].”
    Quote in the paperSubstantial claim0 of 3 sources checked
    • cites [150] Mobadersany P, “ Predicting cancer outcomes from histology and genomics using convolutional networks,” in Proc. Nat. Acad. Sci, 2018, pp. E2970– E2979 29531073 PMC5879673 source

      Not checked in the free run.

    • cites [152] Saltz J, “ Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images,” Cell Rep, vol. 23, no. 1, pp. 181– 193, 2018. 29617659 PMC5943714 source

      Not checked in the free run.

    • cites [153] Cheng J, “ Identification of topological features in renal tumor microenvironment associated with patient survival,” Bioinformatics, vol. 34, no. 6, pp. 1024– 1030, 2018. 29136101 PMC7263397 source

      Not checked in the free run.

  • Classified as probably not a substantial claim. May combine two claims.

    Premisenone given

    In pathology, classifications can hinge on the fraction of the specimen found to have one or another pattern of tissue abnormality, and in such cases the assessment of abnormality and the estimate of tissue area are both subjective.

    “Classifications can hinge of the fraction of the specimen found to have one or another pattern of tissue abnormality. In such cases, the assessment of abnormality and the estimate of tissue area are both subjective.”
    Quote in the paperSubstantial claim

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    To maximize the flexibility and utility of computational imaging tools, it will be necessary to address the challenge of batch effect, which arises due to histopathology tissue slides from different institutions showing heterogeneous appearances as a result of differences in tissue preparation and staining procedures.

    “To maximize the flexibility and utility of the computational imaging tools that are being developed, it will be necessary to address the challenge of batch affect, which arises due to the fact that histopathology tissue slides from different institutions show heterogeneous appearances as a result of differences in tissue preparation and staining procedures.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    Feature-based registration appears more suitable in the case of vessels' contours and centerline, while intensity-based registration can be effectively used for image slices depicting abnormal morphologies such as tumors.

    “To that respect, feature-based registration appears more suitable in the case of vessels’ contours and centerline [196], while the intensity-based registration can be effectively used for image slices depicting abnormal morphologies such as tumors [197].”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [196] Athanasiou L, “ Three-dimensional reconstruction of coronary arteries and plaque morphology using CT angiography–comparison and registration with IVUS,” BMC Med. Imag, vol. 16, no. 1, p. 9, 2016. PMC4719213 26785613 source

      Not checked in the free run.

    • cites [197] Angelopoulou A, “ 3D reconstruction of medical images from slices automatically landmarked with growing neural models,” Neurocomputing, vol. 150, pp. 16– 25, 2. 2015. source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    H&E is one of the main tissue stains and is the most commonly used stain in histopathology.

    “H&E is one of the main tissue stains and is most commonly used stain in histopathology.”
    Quote in the paperNot a substantial claim

  • May combine two claims.

    Interpretationreasoning

    These studies demonstrate the value of whole slide tissue imaging in producing quantitative evaluations of sub-cellular data and opportunities for richer correlative studies.

    “These studies demonstrate the value of whole slide tissue imaging in producing quantitative evaluations of sub-cellular data and opportunities for richer correlative studies.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    These studies showed there are correlations between characterizations of TIL patterns, as analyzed by computerized algorithms, and patient survival rates and groupings of patients based on subclasses of immunotypes.

    “These studies showed there are correlations between characterizations of TIL patterns, as analyzed by computerized algorithms, and patient survival rates and groupings of patients based on subclasses of immunotypes.”
    Quote in the paperSubstantial claim0 of 2 sources checked
    • cites [152] Saltz J, “ Spatial organization and molecular correlation of tumor-infiltrating lymphocytes using deep learning on pathology images,” Cell Rep, vol. 23, no. 1, pp. 181– 193, 2018. 29617659 PMC5943714 source

      Not checked in the free run.

    • cites [181] Amgad M, “ Structured crowdsourcing enables convolutional segmentation of histology images,” Bioinformatics, vol. 35, no. 18, pp. 3461– 3467, 2019. 30726865 PMC6748796 source

      Not checked in the free run.

  • Classified as probably not a substantial claim.

    Interpretationreasoning

    For larger numbers of patient-specific cases, automatic segmentation and reconstruction systems are needed.

    “For those situations, automatic segmentation and reconstruction systems are needed.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisereasoning

    The reconstruction process necessitates expert knowledge and guidance and is particularly time consuming and hence not applicable in the analysis of larger numbers of patient-specific cases.

    “The reconstruction process necessitates expert knowledge and guidance. However, this is particularly time consuming and hence not applicable in the analysis of larger numbers of patient-specific cases.”
    Quote in the paperSubstantial claim

  • May combine two claims.

    Premisecitation

    3D visualization can contribute to the design of aggressive surgery and radiotherapy strategies, with realistic testing and verification, with extensive applications in spinal surgery, joint replacement, neuro-interventions, as well as coronary and aortic stenting.

    “By enabling precise position and orientation of the patient’s anatomy, 3D visualization can contribute to the design of aggressive surgery and radiotherapy strategies, with realistic testing and verification, with extensive applications in spinal surgery, joint replacement, neuro-interventions, as well as coronary and aortic stenting [193].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [193] Siddeshappa N, “ Multiresolution image registration for multimodal brain images and fusion for better neurosurgical planning,” Biomed. J, vol. 40, no. 6, pp. 329– 338, 12. 2017. 29433836 PMC6138619

      Not checked in the free run.

  • May combine two claims.

    Premisenone given

    Most systems for automated assessment of TMA images are limited by being closed and proprietary, not exploiting advanced computer vision techniques, and/or not conforming with emerging data standards.

    “Although there has been some progress made in the development of automated methods for assessing TMA images, most of systems are limited by the fact that they are closed and proprietary; do not exploit the potential of advanced computer vision techniques; and/or do not conform with emerging data standards.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    There are some efforts towards the adoption of DICOM in digital pathology, including the availability of tools such as the Orthanc DICOMizer that can convert a pyramidal tiled tiff file into a DICOM pathology file.

    “There are some efforts towards the adoption of DICOM in digital pathology, including the availability of tools such as the Orthanc DICOMizer [207] that can convert a pyramidal tiled tiff file into a DICOM pathology file.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [207] Jodogne S, “ The Orthanc ecosystem for medical imaging,” J. Digit. Imag, vol. 31, no. 3, pp. 341– 352, 2018. PMC5959835 29725964 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    A dense literature demonstrates the potentially transforming impact of radiomics for different disease staging such as cancer, neurodegenerative, and cardiovascular diseases.

    “A dense literature demonstrates the potentially transforming impact of radiomics for different disease staging such as cancer, neurodegenerative, and cardiovascular diseases [224]–[228].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [224]–[228]

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    Breast phenotype group studies performed relatively well in predicting clinical receptor status, multigene assay recurrence scores (poor vs good prognosis), and molecular subtyping.

    “Moreover, they performed relatively well in predicting clinical receptor status, multigene assay recurrence scores (poor vs good prognosis), and molecular subtyping.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [236]–[239]

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Interpretationcitation

    Since the digital twin can incorporate modelling at different resolutions, from organ structure to cellular and genomic level, it may enable complex simulations with the use of AI tools to integrate huge amounts of data and knowledge aiming at improved diagnostics and therapeutic treatments, without harming the patient.

    “Since the digital twin can incorporate modelling at different resolutions, from organ structure to cellular and genomic level, it may enable complex simulations [223] with the use of AI tools to integrate huge amounts of data and knowledge aiming at improved diagnostics and therapeutic treatments, without harming the patient.”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [223] Malik AA and Bilberg A, “ Digital twins of human robot collaboration in a production setting,” Procedia Manuf, vol. 17, pp. 278– 285. 2018.

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing). May combine two claims.

    Premisenone given

    Processing of digital pathology images is a challenging activity, in part due to the size of whole-slide images, but also because of an abundance of image formats and the frequent need for human guidance and intervention during processing.

    “The processing of digital pathology images is a challenging activity, in part due to the size of whole-slide images, but also because of an abundance of image formats and the frequent need for human guidance and intervention during processing.”
    Quote in the paperNot a substantial claim

  • Classified as probably not a substantial claim. May combine two claims.

    Premisereasoning

    Digital pathology requires interactive systems that can query the underlying data and feature management systems, as well as support interactive visualizations.

    “It requires interactive systems that can query the underlying data and feature management systems, as well as support interactive visualizations.”
    Quote in the paperSubstantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    QuIP is an open-source system that uses the caMicroscope viewer to support the interactive visualization of images, image annotations, and segmentation results as overlays of heatmaps or polygons.

    “QuIP is an open-source system that uses the caMicroscope viewer [202] to support the interactive visualization of images, image annotations, and segmentation results as overlays of heatmaps or polygons.”
    Quote in the paperNot a substantial claim0 of 2 sources checked
    • cites [201] Saltz J, “ A containerized software system for generation, management, and exploration of features from whole slide tissue images,” Cancer Res, vol. 77, no. 21, pp. e79– e82, 2017. 29092946 PMC5987533 source

      Not checked in the free run.

    • cites [202] Sharma A, “ Framework for data management and visualization of the national lung screening trial pathology images,” Pathol. Informat. Summit, pp. 13– 16, 5 2014.

      Couldn’t be matched to a paper in Crossref, so it can’t be checked.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisereasoning

    There is growing interest in the use of cloud computing for digital pathology, driven by the rapid decline in costs, making them increasingly cost-effective solutions for large-scale computing.

    “There is also growing interest in the use of cloud computing for digital pathology, driven by the rapid decline in costs, making them increasingly cost-effective solutions for large-scale computing.”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing), S4 (based). May combine two claims.

    Premisenone given

    Applications of in-silico models evolve drastically in early diagnosis and prognosis, with personalized therapy planning, noninvasive and invasive interactive treatment, as well as planning of pre-operative stages, chemotherapy and radiotherapy.

    “Applications of in-silico models evolve drastically in early diagnosis and prognosis, with personalized therapy planning, noninvasive and invasive interactive treatment, as well as planning of pre-operative stages, chemotherapy and radiotherapy (see Fig. 2).”
    Quote in the paperNot a substantial claim

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    The digital equivalent of a complex human functional system enables the consideration of event dynamics, such as tumour growth or information transfer in epilepsy network, as well as a systemic response to therapy, such as response to pharmacogenomics or targeted radiotherapy.

    “In this form, the digital equivalent of a complex human functional system enables the consideration of event dynamics, such as tumour growth or information transfer in epilepsy network, as well as a systemic response to therapy, such as response to pharmacogenomics or targeted radiotherapy [222].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [222] Bruynseels K, De Sio FS, and van den Hoven J, “ Digital twins in health care: Ethical implications of an emerging engineering paradigm,” Frontiers Genetics, vol. 9, p. 31– 2018. PMC5816748 29487613 source

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Radiogenomics methods extend radiomics approaches by investigating the correlation between a tumor's quantitative imaging features and its molecular and genetic profiling.

    “Going one-step further, radiogenomics methods extend radiomics approaches by investigating the correlation between, for example, a tumor’s characteristics in terms of quantitative imaging features and its molecular and genetic profiling [68].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [68] Panayides AS, “ Radiogenomics for precision medicine with a big data analytics perspective,” IEEE J. Biomed. Health Inform, vol. 23, no. 5, pp. 2063– 2079, 9. 2019. 30596591

      Not checked in the free run.

    • Rests on an assumption: During the transformation from a benign to malignant state and throughout disease progression, changes occur in underlying molecular, histologic and protein expression patterns, with each contributing a different perspective and complementary strength.

  • Probably not a substantial claim: may not pass S2 (checkable), S3 (load-bearing).

    Premisecitation

    Precision medicine is an emerging approach for administering healthcare that aims to improve the accuracy with which clinical decisions are rendered towards improving the delivery of personalized treatment and therapy planning for patients.

    “Precision medicine is an emerging approach for administering healthcare that aims to improve the accuracy with which clinical decisions are rendered towards improving the delivery of personalized treatment and therapy planning for patients as depicted in Fig. 3 [67].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [67] Pattichis CS, “ Guest Editorial on the special issue on integrating informatics and technology for precision medicine,” IEEE J. Biomed. Health Inform, vol. 23, no. 1, pp. 12– 13, 1. 2019. source

      Not checked in the free run.

  • May combine two claims.

    Premisecitation

    One of the main challenges in digital pathology data management is managing relatively large-scale, multi-dimensional data sets that continue to expand over time, because exhaustively comparing query data with each sample in a high-dimensional database is unreasonable due to practical storage and computational bottlenecks.

    “One of the main challenges is how to manage relatively large-scale, multi-dimensional data sets that will continue to expand over time since it is unreasonable to exhaustively compare the query data with each sample in a high-dimensional database due to practical storage and computational bottlenecks [255].”
    Quote in the paperSubstantial claim0 of 1 sources checked
    • cites [255] Zhang Z, “ High-throughput histopathological image analysis via robust cell segmentation and hashing,” Med. Image. Anal, vol. 26, no. 1, pp. 306– 315, 2015. 26599156 PMC4679540

      Not checked in the free run.

  • Probably not a substantial claim: may not pass S3 (load-bearing).

    Premisecitation

    Hashing-based retrieval approaches are gaining popularity in the medical imaging community due to their exceptional efficiency and scalability.

    “Hashing-based retrieval approaches are gaining popularity in the medical imaging community due to their exceptional efficiency and scalability [255].”
    Quote in the paperNot a substantial claim0 of 1 sources checked
    • cites [255] Zhang Z, “ High-throughput histopathological image analysis via robust cell segmentation and hashing,” Med. Image. Anal, vol. 26, no. 1, pp. 306– 315, 2015. 26599156 PMC4679540

      Not checked in the free run.

    • Rests on an assumption: CBR systems have to deal with very large and high-dimensional datasets, the complexity of which can easily render simple feature concatenation inefficient and insufficiently robust.
61 of 61

  • The paper relies on this without support.

    Variations in imaging parameters (e.g., in-plane resolution, slice thickness) may have strong impacts on image analysis.

    “Variations in imaging parameters (e.g. in-plane resolution, slice thickness, etc.) – which were not discussed – may have strong impacts on image analysis and should be considered during algorithm development.”

  • The paper relies on this without support.

    The prodigious amount of imaging data generated causes a significant need for informatics in storage, transmission, analysis, and automated interpretation.

    “Moreover, the prodigious amount of imaging data generated causes a significant need for informatics in the storage and transmission as well as in the analysis and automated interpretation of the data, underpinning the use of big data science in improved utilization and diagnosis.”

  • The paper relies on this without support.

    There is variability in the agreement between experts in delineation tasks.

    “We highlight that all the above works treat expert delineations as ground truth, whereas our community is well aware of the variability in the agreement between experts in delineation tasks.”

  • The paper relies on this without support.

    Deep learning methods generally suffer with limited training data.

    “With bountiful training data, CNNs can often achieve state-of-the-art performance; however, deep learning methods generally suffer with limited training data.”

  • The paper relies on this without support.

    Spurious correlations in training data can be incorrectly learned by CNN systems in the medical domain.

    “Importantly, due diligence must be done during the design of CNN systems in the medical domain to ensure spurious correlations in the training data are not incorrectly learned.”

  • The paper relies on this without support.

    Digital pathology is an inherently interactive human-guided activity.

    “Digital pathology is an inherently interactive human-guided activity.”

  • The paper relies on this without support.

    Interactivity is a prerequisite to wide-scale adoption of digital pathology in imaging informatics applications.

    “Such interactivity is a prerequisite to wide-scale adoption of digital pathology in imaging informatics applications.”

  • The paper relies on this without support.

    The continued availability of large, open datasets of medical images will play a big part in strengthening classification tasks in the medical domain.

    “As discussed, transfer learning has been beneficial in coping with scant data, but the continued availability of large, open datasets of medical images will play a big part in strengthening classification tasks in the medical domain.”

  • The paper relies on this without support.

    During the transformation from a benign to malignant state and throughout disease progression, changes occur in underlying molecular, histologic and protein expression patterns, with each contributing a different perspective and complementary strength.

    “During the transformation from a benign to malignant state and throughout the course of disease progression, changes occur in the underlying molecular, histologic and protein expression patterns, with each contributing a different perspective and complementary strength.”

  • The paper relies on this without support.

    TCIA and TCGA repositories provide matched imaging, genetic and clinical data for over 20 different cancer types.

    “Such example is the TCIA and the TCGA repositories, which provide matched imaging, genetic and clinical data for over 20 different cancer types.”

  • The paper relies on this without support.

    Incorporating longitudinal and multi-modality radiology and pathology image features further enhances the discriminatory power of radiomics predictive models.

    “Incorporating longitudinal and multi-modality radiology and pathology (see also Section VII.C) image features further enhances the discriminatory power of these models.”

  • The paper relies on this without support.

    In the medical domain, matching the size of natural image datasets is not readily possible.

    “This paradigm is rather worrying as in the medical domain matching that size is not readily possible.”

  • The paper relies on this without support.

    CBR systems have to deal with very large and high-dimensional datasets, the complexity of which can easily render simple feature concatenation inefficient and insufficiently robust.

    “However, CBR systems have to deal with very large and high-dimensional datasets, the complexity of which can easily render simple feature concatenation inefficient and insufficiently robust.”

  • The paper relies on this without support.

    Deep learning-based computer vision tasks have been developed on enormous data of natural images that go beyond ImageNet.

    “Deep learning-based computer vision tasks have been developed on “enormous” data of natural images that go beyond ImageNet (see for example the efforts of Google, and Facebook).”
14 of 14

Claims other papers make about this one

Each is checked against this paper’s own text: does it say what the citing paper says it does?

  • A major application of AI in medicine is image analysis in radiology, and AI is increasingly adept at gathering information from images and finding underlying patterns that even humans can't identify.

    “A major application of AI in medicine is image analysis in radiology, and AI being increasingly adept at gathering information from images and finding underlying patterns that even humans can't identify has become a very powerful tool in this field [31].”

    From A Review of Trustworthy and Explainable Artificial Intelligence (XAI) (2023) · cites this paper as [31]

    The Concluding Remarks & Future Directions describe AI applications that analyze radiology images, and Processing, Analysis, and Understanding in Radiology says machine-learning methods learn image features from examples. Our copy does not establish the stronger claim that AI finds patterns humans cannot identify. (LLM, conf=0.86)

    Experimental — accuracy not yet measured

  • Many machine learning and deep learning systems for predicting sepsis onset and patient outcomes function as black boxes, generating predictions without explanations.

    “many of these systems function as “black boxes,” generating predictions without providing explanations for their reasoning.”

    From Explainable AI for critical care: a systematic review of interpretable models for sepsis and ICU mortality prediction (2026) · cites this paper as [1]

    The “CNN Interpretability” section describes deep CNNs as black-box functions and says researchers need to understand how they derive classification decisions. That supports the general point about unexplained predictions, but the discussion concerns medical imaging, not systems predicting sepsis onset or patient outcomes. The passages found for sepsis and patient outcomes do not establish that narrower application. (LLM, conf=0.87)

    Experimental — accuracy not yet measured

  • Much current research prioritizes improving prediction accuracy while giving less attention to whether explainable AI explanations are truly understandable and useful for clinicians.

    “In addition, much of the current research prioritizes improving prediction accuracy while giving less attention to whether the explanations produced by XAI models are truly understandable and useful for clinicians [1, 2, 12].”

    From Explainable AI for critical care: a systematic review of interpretable models for sepsis and ICU mortality prediction (2026) · cites this paper as [1]

    The CNN Interpretability section (page unavailable) says deep CNNs have achieved high accuracy but remain black boxes, and describes interpretability as an emerging research area. It does not say that much current research prioritizes accuracy over evaluating whether XAI explanations are understandable and useful to clinicians. (LLM, conf=0.82)

    Experimental — accuracy not yet measured

  • In critical care, lack of transparency in AI tools can reduce clinician confidence, delay necessary interventions, and limit practical adoption.

    “In critical care, where timely decisions can directly affect patient survival, this lack of transparency can reduce clinician confidence, delay necessary interventions, and limit the practical adoption of AI tools [1, 2].”

    From Explainable AI for critical care: a systematic review of interpretable models for sepsis and ICU mortality prediction (2026) · cites this paper as [1]

    The “CNN Interpretability” section (page unavailable) describes deep CNNs as black boxes and says it is essential to trust and verify their predictions, lending support to the transparency concern. It does not address clinician confidence specifically, or say that lack of transparency delays interventions or limits AI adoption in critical care. Its discussion of limited adoption elsewhere concerns data-management tools and different causes. (LLM, conf=0.79)

    Experimental — accuracy not yet measured

  • Deep learning algorithms can sift through electronic health records to predict patient outcomes.

    “In addition, the keen eye of deep learning algorithms sifts through electronic health records, unearthing valuable insights that pave the way for predicting patient outcomes [31,32].”

    From Edge AI for Early Detection of Chronic Diseases and the Spread of Infectious Diseases: Opportunities, Challenges, and Future Directions (2023) · cites this paper as [32]

Self-check

Percentages, sums and reported differences re-computed from the paper’s own numbers.

Each contradiction between two passages listed here was confirmed by an independent check.

  • Lung-nodule autoencoder claim cites a melanoma study

    Minor

    The text attributes lung-nodule feature learning with a 2D autoencoder to reference [128], but the paper's own entry for [128] describes melanoma recognition using deep residual networks.

1 of 1

Cites

Flagged first, then by citations

Research works printed but not linked: 6

  1. zaldana, 2018, Analysis of lung adenocarcinoma based on nuclear features and WHO subtype classification using deep convolutional neural networks on histopathology images, Proc Amer Assoc Cancer Res Special Conf Convergence Artif Intell Big Data Prediction Cancer
  2. 10.1109/TBME.2009.2035102 · DOI · Paper page
  3. 10.4103/jpi.jpi_69_18 · DOI · Paper page
  4. zheng, 2007, The survey of medical image 3D reconstruction, Proc PIBM
  5. sharma, 2014, Framework for data management and visualization of the national lung screening trial pathology images, Pathol Informat Summit, 13
  6. kohl, 2018, A probabilistic U-net for segmentation of ambiguous images, Advances Neural Information Processing Systems, 6965

Cited by

Sorted by citations

Related

Authors
Panayides, A. S., Amini, A., Filipovic, N. D., Sharma, A., Tsaftaris, S. A., Young, A., Foran, D., Do, N., Golemati, S., Kurc, T., Huang, K., Nikita, K. S., Veasey, B. P., Zervakis, M., Saltz, J. H., & Pattichis, C. S.
Affiliations
  • University of Cyprus (CY)
  • University of Louisville (US)
  • Emory University (US)
  • The Alan Turing Institute (GB)
  • University of Auckland (NZ)
  • Rutgers, The State University of New Jersey (US)
  • United States Department of Veterans Affairs (US)
  • National and Kapodistrian University of Athens (GR)
  • Regenstrief Institute (US)
  • National Technical University of Athens (GR)
  • Technical University of Crete (GR)
  • Stony Brook University (US)
ORCID
Venue
IEEE Journal of Biomedical and Health Informatics
Publisher
Institute of Electrical and Electronics Engineers (IEEE)
Published
2020-07
Volume
24
Issue
7
Pages
1837-1857
ISSN
  • 2168-2194 (ISSN-L)
  • 2168-2208
DOI
10.1109/jbhi.2020.2991043
PMC
PMC8580417
OpenAlex
W3030790048
Cited by
713
References
262
Citations by year
2026: 103 · 2025: 219 · 2024: 172 · 2023: 113 · 2022: 68 · 2021: 35 · 2020: 3
FWCI
16.12
Type
journal article
Language
en
Open access
Licence
Locations
  • IEEE Journal of Biomedical and Health Informatics (publishedVersion, cc-by)
  • PubMed (publishedVersion)
  • Europe PMC (PubMed Central) (acceptedVersion, cc-by)
  • PubMed Central (acceptedVersion, cc-by)
  • IUScholarWorks (Indiana University) (submittedVersion, other-oa)
  • Edinburgh Research Explorer (University of Edinburgh) (submittedVersion, cc-by)
Funders
  • National Institutes of Health (UL1TR003017)
  • Engineering and Physical Sciences Research Council (EP/P022928/1)
  • National Cancer Institute (UG3CA225021, P30CA072720)
Topics
  • Health Sciences › Medicine › Health Informatics › Artificial Intelligence in Healthcare and Education
  • Physical Sciences › Computer Science › Computer Vision and Pattern Recognition › Medical Image Segmentation Techniques
  • Health Sciences › Medicine › Radiology, Nuclear Medicine and Imaging › Radiomics and Machine Learning in Medical Imaging
Keywords
artificial intelligence in medical imaging, medical imaging informatics, health data management, disease classification, deep learning architectures, in silico modeling, 3D reconstruction, 3D visualization, digital pathology, radiology, precision medicine, clinical translation
MeSH terms
Big Data, Artificial Intelligence, Diagnostic Imaging, Humans, Image Interpretation, Computer-Assisted, Image Processing, Computer-Assisted, Medical Informatics, Precision Medicine

About these checks

Registry, retraction, Cites and statistics results are computed, not judged, with one exception: a printed reference that no index lists is read once by a model before it counts as a citation error. Claims are extracted by a model and kept only when their quote appears verbatim in the text; each cited source is then checked, with a model’s help, for whether it says what the claim relies on. Those verdicts can be wrong — read the claim’s quote and the cited source before relying on one.

Claims read by deepseek-v4.1-flash under definition version 3 (extract-v3) on 2026-10-09.

Classified by jev-1.13-20260917 (gold-guideline-v0.2) on 2026-10-09.

Statistics and arithmetic are recomputed, not judged.

Confidence bands are shown only for fields calibrated on our annotated test set. None is calibrated yet, so no band is shown.

What counts as a claim ›How we measure confidence ›How we check this ›

Are you an author, or do you see a problem with this page? Request a review