Summary of the National Cancer Institute 2023 Virtual Workshop on Medical Image De-identification-Part 1: Report of the MIDI Task Group - Best Practices and Recommendations, Tools for Conventional Approaches to De-identification, International Approaches to De-identification, and Industry Panel on Image De-identification.

De-identification of medical images intended for research is a core requirement for data-sharing initiatives, particularly as the demand for data for artificial intelligence (AI) applications grows. The Center for Biomedical Informatics and Information Technology (CBIIT) of the US National Cancer Institute (NCI) convened a virtual workshop with the intent of summarizing the state of the art in de-identification technology and processes and exploring interesting aspects of the subject. This paper summarizes the highlights of the first day of the workshop, the recordings, and presentations of which are publicly available for review. The topics covered included the report of the Medical Image De-Identification Initiative (MIDI) Task Group on best practices and recommendations, tools for conventional approaches to de-identification, international approaches to de-identification, and an industry panel.

Documenting the de-identification process of clinical and imaging data for AI for health imaging projects.

Artificial intelligence (AI) is revolutionizing the field of medical imaging, holding the potential to shift medicine from a reactive "sick-care" approach to a proactive focus on healthcare and prevention. The successful development of AI in this domain relies on access to large, comprehensive, and standardized real-world datasets that accurately represent diverse populations and diseases. However, images and data are sensitive, and as such, before using them in any way the data needs to be modified to protect the privacy of the patients. This paper explores the approaches in the domain of five EU projects working on the creation of ethically compliant and GDPR-regulated European medical imaging platforms, focused on cancer-related data. It presents the individual approaches to the de-identification of imaging data, and describes the problems and the solutions adopted in each case. Further, lessons learned are provided, enabling future projects to optimally handle the problem of data de-identification. CRITICAL RELEVANCE STATEMENT: This paper presents key approaches from five flagship EU projects for the de-identification of imaging and clinical data offering valuable insights and guidelines in the domain. KEY POINTS: ΑΙ models for health imaging require access to large amounts of data. Access to large imaging datasets requires an appropriate de-identification process. This paper provides de-identification guidelines from the AI for health imaging (AI4HI) projects.

Discovery and preclinical development of a therapeutically active nanobody-based chimeric antigen receptor targeting human CD22.

Chimeric antigen receptor (CAR) T cell therapies targeting B cell-restricted antigens CD19, CD20, or CD22 can produce potent clinical responses for some B cell malignancies, but relapse remains common. Camelid single-domain antibodies (sdAbs or nanobodies) are smaller, simpler, and easier to recombine than single-chain variable fragments (scFvs) used in most CARs, but fewer sdAb-CARs have been reported. Thus, we sought to identify a therapeutically active sdAb-CAR targeting human CD22. Immunization of an adult Llama glama with CD22 protein, sdAb-cDNA library construction, and phage panning yielded >20 sdAbs with diverse epitope and binding properties. Expressing CD22-sdAb-CAR in Jurkat cells drove varying CD22-specific reactivity not correlated with antibody affinity. Changing CD28- to CD8-transmembrane design increased CAR persistence and expression in vitro. CD22-sdAb-CAR candidates showed similar CD22-dependent CAR-T expansion in vitro, although only membrane-proximal epitope targeting CD22-sdAb-CARs activated direct cytolytic killing and extended survival in a lymphoma xenograft model. Based on enhanced survival in blinded xenograft studies, a lead CD22sdCAR-T was selected, achieving comparable complete responses to a benchmark short linker m971-scFv CAR-T in high-dose experiments. Finally, immunohistochemistry and flow cytometry confirm tissue and cellular-level specificity of the lead CD22-sdAb. This presents a complete report on preclinical development of a novel CD22sdCAR therapeutic.

medigan: a Python library of pretrained generative models for medical image synthesis.

Purpose: Deep learning has shown great promise as the backbone of clinical decision support systems. Synthetic data generated by generative models can enhance the performance and capabilities of data-hungry deep learning models. However, there is (1) limited availability of (synthetic) datasets and (2) generative models are complex to train, which hinders their adoption in research and clinical applications. To reduce this entry barrier, we explore generative model sharing to allow more researchers to access, generate, and benefit from synthetic data. Approach: We propose medigan, a one-stop shop for pretrained generative models implemented as an open-source framework-agnostic Python library. After gathering end-user requirements, design decisions based on usability, technical feasibility, and scalability are formulated. Subsequently, we implement medigan based on modular components for generative model (i) execution, (ii) visualization, (iii) search & ranking, and (iv) contribution. We integrate pretrained models with applications across modalities such as mammography, endoscopy, x-ray, and MRI. Results: The scalability and design of the library are demonstrated by its growing number of integrated and readily-usable pretrained generative models, which include 21 models utilizing nine different generative adversarial network architectures trained on 11 different datasets. We further analyze three medigan applications, which include (a) enabling community-wide sharing of restricted data, (b) investigating generative model evaluation metrics, and (c) improving clinical downstream tasks. In (b), we extract Fréchet inception distances (FID) demonstrating FID variability based on image normalization and radiology-specific feature extractors. Conclusion: medigan allows researchers and developers to create, increase, and domain-adapt their training data in just a few lines of code. Capable of enriching and accelerating the development of clinical machine learning models, we show medigan's viability as platform for generative model sharing. Our multimodel synthetic data experiments uncover standards for assessing and reporting metrics, such as FID, in image synthesis studies.

Analysis of Caregiver Burden Expressed in Social Media Discussions.

Almost 40% of US adults provide informal caregiving, yet research gaps remain around what burdens affect informal caregivers. This study uses a novel social media site, Reddit, to mine and better understand what online communities focus on as their caregiving burdens. These forums were accessed using an application programming interface, a machine learning classifier was developed to remove low information posts, and topic modeling was applied to the corpus. An expert panel summarized the forums' themes into ten categories. The largest theme extracted from Reddit's forums discussed the personal emotional toll of being a caregiver. This was followed by logistic issues while caregiving and caring for parents who have cancer. Smaller themes included approaches to end-of-life care, physical equipment needs when caregiving, and the use of wearables or technology to help monitor care recipients. The platform often discusses caregiving for parents which may reflect the age of Reddit's users. This study confirms that Reddit forums are used for caregivers to discuss the burdens associated with their role and the types of stress that can result from informal caregiving.

Patient survey examining the experience of care of a hospital-based opt-out tobacco dependency treatment service (the CURE Project).

INTRODUCTION: Treating tobacco dependency in patients admitted to hospital is a key priority in the National Health Service long-term plan. This service evaluation assessed the perception, needs and experience of care within an opt-out hospital-based tobacco dependency treatment service (the Conversation, Understand, Replace, Experts and Evidence Base (CURE) team) in North-West England. METHODS: A survey was offered to all eligible patients between 1 July 2020 and 30 September 2020. Eligibility criteria were adult patients identified as an active smoker being approached by the CURE team as part of the standard opt-out service model, on a non-covid ward without a high suspicion of COVID-19 infection and able to read and write in English. RESULTS: 106 completed surveys were evaluated. Participants demonstrated high levels of tobacco dependency with an average of 37 years smoking history and 66% describing the onset of cravings within 30 min of hospital admission. The average number quit attempts in the previous 12 months was 1.3 but only 9% had used the most effective National Institute for Health and Care Excellence (NICE) recommended treatments. 100% felt the opt-out service model was appropriate and 96% stated the treatment and support they had received had prompted them to consider a further quit attempt. 82% of participants rated their experience of care as 9/10 or 10/10. Participants wanted a broad range of support post discharge with the most popular option being with their general practitioner. 66% and 65% of participants would have been interested in a vaping kit as stop smoking intervention and support vaping-friendly hospital grounds respectively. CONCLUSION: These results suggest this hospital-based, opt-out tobacco dependency treatment service delivers high-quality experience of care and meets the needs of the patients it serves. It also highlights the opportunity to enhance outcomes by providing access to NICE recommended most-effective interventions (varenicline, vaping and combination nicotine replacement therapy) and providing flexible, individualised discharge pathways.

Plasma cells expression from smouldering myeloma to myeloma reveals the importance of the PRC2 complex, cell cycle progression, and the divergent evolutionary pathways within the different molecular subgroups.

Sequencing studies have shed some light on the pathogenesis of progression from smouldering multiple myeloma (SMM) and symptomatic multiple myeloma (MM). Given the scarcity of smouldering samples, little data are available to determine which translational programmes are dysregulated and whether the mechanisms of progression are uniform across the main molecular subgroups. In this work, we investigated 223 SMM and 1348 MM samples from the University of Arkansas for Medical Sciences (UAMS) for which we had gene expression profiling (GEP). Patients were analysed by TC-7 subgroup for gene expression changes between SMM and MM. Among the commonly dysregulated genes in each subgroup, PHF19 and EZH2 highlight the importance of the PRC2.1 complex. We show that subgroup specific differences exist even at the SMM stage of disease with different biological features driving progression within each TC molecular subgroup. These data suggest that MMSET SMM has already transformed, but that the other precursor diseases are distinct clinical entities from their symptomatic counterpart.

Health economic analysis for the 'CURE Project' pilot: a hospital-based tobacco dependency treatment service in Greater Manchester.

INTRODUCTION: Treating tobacco dependency in patients admitted to acute care National Health Service (NHS) trusts is a key priority in the NHS 10-year plan. This paper sets out the results of a health economic analysis for 'The CURE Project' pilot; a new hospital-based tobacco dependency service. METHODS: A health economic analysis to understand the costs of the intervention (both for the inpatient service and postdischarge costs), the return on investment (ROI) and the cost per quality-adjusted life year (QALY) of the CURE Project pilot in Greater Manchester. ROI and cost per QALY were calculated using the European Study on Quantifying Utility of Investment in Protection from Tobacco and Greater Manchester Cost Benefit Analysis Tools. RESULTS: The total intervention costs for the inpatient service in the 6-month CURE pilot were £96 224 with a cost per patient who smokes of £40.21. The estimated average cost per patient who was discharged on pharmacotherapy was £97.40. The cost per quit (22% quit rate for smokers at 12 weeks post discharge) was £475. The gross financial ROI ratio was £2.12 return per £1 invested with a payback period of 4 years. The cashable financial ROI ratio was £1.06 return per £1 invested with a payback period of 10 years. The public value ROI ratio was £30.49 per £1 invested. The cost per QALY for this programme was £487. DISCUSSION: The CURE Project pilot has been shown to be exceptionally cost-effective with highly significant ROI in this health economic analysis. This supports the NHS priority to embed high-quality tobacco addiction treatment services in acute NHS trusts, and the CURE Project provides a blueprint and framework to achieve this.

A DICOM dataset for evaluation of medical image de-identification.

We developed a DICOM dataset that can be used to evaluate the performance of de-identification algorithms. DICOM objects (a total of 1,693 CT, MRI, PET, and digital X-ray images) were selected from datasets published in the Cancer Imaging Archive (TCIA). Synthetic Protected Health Information (PHI) was generated and inserted into selected DICOM Attributes to mimic typical clinical imaging exams. The DICOM Standard and TCIA curation audit logs guided the insertion of synthetic PHI into standard and non-standard DICOM data elements. A TCIA curation team tested the utility of the evaluation dataset. With this publication, the evaluation dataset (containing synthetic PHI) and de-identified evaluation dataset (the result of TCIA curation) are released on TCIA in advance of a competition, sponsored by the National Cancer Institute (NCI), for algorithmic de-identification of medical image datasets. The competition will use a much larger evaluation dataset constructed in the same manner. This paper describes the creation of the evaluation datasets and guidelines for their use.

Chest imaging representing a COVID-19 positive rural U.S. population.

As the COVID-19 pandemic unfolds, radiology imaging is playing an increasingly vital role in determining therapeutic options, patient management, and research directions. Publicly available data are essential to drive new research into disease etiology, early detection, and response to therapy. In response to the COVID-19 crisis, the National Cancer Institute (NCI) has extended the Cancer Imaging Archive (TCIA) to include COVID-19 related images. Rural populations are one population at risk for underrepresentation in such public repositories. We have published in TCIA a collection of radiographic and CT imaging studies for patients who tested positive for COVID-19 in the state of Arkansas. A set of clinical data describes each patient including demographics, comorbidities, selected lab data and key radiology findings. These data are cross-linked to SARS-COV-2 cDNA sequence data extracted from clinical isolates from the same population, uploaded to the GenBank repository. We believe this collection will help to address population imbalance in COVID-19 data by providing samples from this normally underrepresented population.

TarPan: an easily adaptable targeted sequencing panel viewer for research and clinical use.

BACKGROUND: The study of cancer genomics continually matures as the number of patient samples sequenced increases. As more data is generated, oncogenic drivers for specific cancer types are discovered along with their associated risks. This in turn leads to potential treatment strategies that pave the way to precision medicine. However, significant financial and analytical barriers make it infeasible to sequence the entire genome of every patient. In contrast, targeted sequencing panels give reliable information on relevant portions of the genome at a fiscally responsible cost. Therefore, we have created the Targeted Panel (TarPan) Viewer, a software tool, to investigate this type of data. RESULTS: TarPan Viewer helps investigators understand data from targeted sequencing data by displaying the information through a web browser interface. Through this interface, investigators can easily observe copy number changes, mutations, and structural events in cancer samples. The viewer runs in R Shiny with a robust SQLite backend and its input is generated from bioinformatic algorithms reliably described in the literature. Here we show the results from using TarPan Viewer on publicly available follicular lymphoma, breast cancer, and multiple myeloma data. In addition, we have tested and utilized the viewer internally, and this data has been used in high-impact peer-reviewed publications. CONCLUSIONS: We have designed a flexible, simple to setup viewer that is easily adaptable to any type of cancer targeted sequencing, and has already proven its use in a research laboratory environment. Further, we believe with deeper sequencing and/or more targeted application it could be of use in the clinic in conjunction with an appropriate targeted sequencing panel as a cost-effective diagnostic test, especially in cancers such as acute leukemia or diffuse large B-cell lymphoma that require rapid interventions.

Feasibility, uptake and impact of a hospital-wide tobacco addiction treatment pathway: Results from the CURE project pilot.

INTRODUCTION: Providing comprehensive tobacco addiction treatment to smokers admitted to acute care settings represents an opportunity to realise major health resource savings and population health improvements. METHODS: The CURE project is a hospital-wide tobacco addiction treatment service piloted in Wythenshawe Hospital, Manchester, UK. The core components of the project are electronic screening of all patients to identify smokers; the provision of brief advice and pharmacotherapy by frontline staff; opt-out referral of smokers to a specialist team for inpatient behavioural interventions; and continued support after discharge. RESULTS: From 01 October 2018 to 31 March 2019, 92% (13,515/14,690) of adult admissions were screened for smoking status, identifying 2,393 current smokers. Of these, 96% were given brief advice to quit by the admitting team. Through the automated 'opt-out' referral process, 61% patients completed inpatient behavioural interventions with a specialist cessation practitioner (69% within the first 48 hours of admission). Overall, 66% of smokers were prescribed pharmacotherapy. Over one in five of all smokers admitted during this pilot reported that they were abstinent from smoking 12 weeks after discharge (22%) at a cost £183 per quit. DISCUSSION: National implementation of this cost-effective programme would be likely to generate substantial benefits to public health.

BRAF and DIS3 Mutations Associate with Adverse Outcome in a Long-term Follow-up of Patients with Multiple Myeloma.

PURPOSE: Copy-number changes and translocations have been studied extensively in many datasets with long-term follow-up. The impact of mutations remains debated given the short time to follow-up of most datasets. EXPERIMENTAL DESIGN: We performed targeted panel sequencing covering 125 myeloma-specific genes and the loci involved in translocations in 223 newly diagnosed myeloma samples recruited into one of the total therapy trials. RESULTS: As expected, the most commonly mutated genes were NRAS, KRAS, and BRAF, making up 44% of patients. Double-Hit and BRAF and DIS3 mutations had an impact on outcome alongside classical risk factors in the context of an intensive treatment approach. We were able to identify both V600E and non-V600E BRAF mutations, 58% of which were predicted to be hypoactive or kinase dead. Interestingly, 44% of the hypoactive/kinase dead BRAF-mutated patients showed co-occurring alterations in KRAS, NRAS, or activating BRAF mutations, suggesting that they play a role in the oncogenesis of multiple myeloma by facilitating MAPK activation and may lead to chemoresistance. CONCLUSIONS: Overall, these data highlight the importance of mutational screening to better understand newly diagnosed multiple myeloma and may lead to patient-specific mutation-driven treatment approaches.

Personalized medicine: CCO's vision, accomplishments and future plans.

Personalized medicine is a rapidly expanding field, with the potential to improve patient care. Its benefits include increasing efficiency in cancer screening, diagnosis and treatment through early detection, targeted therapy and identifying individuals with an underlying genetic risk for cancer or adverse outcomes. Through the work of Cancer Care Ontario (CCO)'s Pathology and Laboratory Medicine Program, a number of initiatives have been undertaken to support developments in personalized medicine. In keeping with the momentum of recent accomplishments, CCO has led the formation of the Personalized Medicine Steering Committee to develop a comprehensive provincial genetics strategy for the future of cancer care.