Galileo-an Artificial Intelligence tool for evaluating pre-implantation kidney biopsies.

BACKGROUND: Pre-transplant procurement biopsy interpretation is challenging, also because of the low number of renal pathology experts. Artificial intelligence (AI) can assist by aiding pathologists with kidney donor biopsy assessment. Herein we present the "Galileo" AI tool, designed specifically to assist the on-call pathologist with interpreting pre-implantation kidney biopsies. METHODS: A multicenter cohort of whole slide images acquired from core-needle and wedge biopsies of the kidney was collected. A deep learning algorithm was trained to detect the main findings evaluated in the pre-implantation setting (normal glomeruli, globally sclerosed glomeruli, ischemic glomeruli, arterioles and arteries). The model obtained on the Aiforia Create platform was validated on an external dataset by three independent pathologists to evaluate the performance of the algorithm. RESULTS: Galileo demonstrated a precision, sensitivity, F1 score and total area error of 81.96%, 94.39%, 87.74%, 2.81% and 74.05%, 71.03%, 72.5%, 2% in the training and validation sets, respectively. Galileo was significantly faster than pathologists, requiring 2 min overall in the validation phase (vs 25, 22 and 31 min by 3 separate human readers, p < 0.001). Galileo-assisted detection of renal structures and quantitative information was directly integrated in the final report. CONCLUSIONS: The Galileo AI-assisted tool shows promise in speeding up pre-implantation kidney biopsy interpretation, as well as in reducing inter-observer variability. This tool may represent a starting point for further improvements based on hard endpoints such as graft survival.

Linking Health Terminologies: A Unified Approach to the WHO Family of International Classifications.

This paper presents an effort by the World Health Organization (WHO) to integrate the reference classifications of the Family of International Classifications (ICD, ICF, and ICHI) into a unified digital framework. The integration was accomplished via an expanded Content Model and a single Foundation that hosts all entities from these classifications, allowing the traditional use cases of individual classifications to be retained while enhancing their combined use. The harmonized WHO-FIC Content Model and the unified Foundation has streamlined the content management, enhanced the web-based tool functionalities, and provided opportunities for linkage with external terminologies and ontologies. This integration promises reduced maintenance cost, seamless joint application, complete representation of health-related concepts while enabling better interoperability with other informatics infrastructures.

From EHR to Machine Learning: A Preliminary Report on an Ingestion Pipeline Based on JSON-LD.

In this paper, we present the preliminary experiments for the development of an ingestion mechanism to move data from Electronic Health Records to machine learning processes, based on the concept of Linked Data and the JSON-LD format.

Optimizing Vision Transformers for Histopathology: Pretraining and Normalization in Breast Cancer Classification.

This paper introduces a self-attention Vision Transformer model specifically developed for classifying breast cancer in histology images. We examine various training strategies and configurations, including pretraining, dimension resizing, data augmentation and color normalization strategies, patch overlap, and patch size configurations, in order to evaluate their impact on the effectiveness of the histology image classification. Additionally, we provide evidence for the increase in effectiveness gathered through geometric and color data augmentation techniques. We primarily utilize the BACH dataset to train and validate our methods and models, but we also test them on two additional datasets, BRACS and AIDPATH, to verify their generalization capabilities. Our model, developed from a transformer pretrained on ImageNet, achieves an accuracy rate of 0.91 on the BACH dataset, 0.74 on the BRACS dataset, and 0.92 on the AIDPATH dataset. Using a model based on the prostate small and prostate medium HistoEncoder models, we achieve accuracy rates of 0.89 and 0.86, respectively. Our results suggest that pretraining on large-scale general datasets like ImageNet is advantageous. We also show the potential benefits of using domain-specific pretraining datasets, such as extensive histopathological image collections as in HistoEncoder, though not yet with clear advantages.

Enhancing AI Research for Breast Cancer: A Comprehensive Review of Tumor-Infiltrating Lymphocyte Datasets.

The field of immunology is fundamental to our understanding of the intricate dynamics of the tumor microenvironment. In particular, tumor-infiltrating lymphocyte (TIL) assessment emerges as essential aspect in breast cancer cases. To gain comprehensive insights, the quantification of TILs through computer-assisted pathology (CAP) tools has become a prominent approach, employing advanced artificial intelligence models based on deep learning techniques. The successful recognition of TILs requires the models to be trained, a process that demands access to annotated datasets. Unfortunately, this task is hampered not only by the scarcity of such datasets, but also by the time-consuming nature of the annotation phase required to create them. Our review endeavors to examine publicly accessible datasets pertaining to the TIL domain and thereby become a valuable resource for the TIL community. The overall aim of the present review is thus to make it easier to train and validate current and upcoming CAP tools for TIL assessment by inspecting and evaluating existing publicly available online datasets.

Supporting Fair and Efficient Emergency Medical Services in a Large Heterogeneous Region.

Emergency Medical Services (EMS) are crucial in delivering timely and effective medical care to patients in need. However, the complex and dynamic nature of operations poses challenges for decision-making processes at strategic, tactical, and operational levels. This paper proposes an action-driven strategy for EMS management, employing a multi-objective optimizer and a simulator to evaluate potential outcomes of decisions. The approach combines historical data with dynamic simulations and multi-objective optimization techniques to inform decision-makers and improve the overall performance of the system. The research focuses on the Friuli Venezia Giulia region in north-eastern Italy. The region encompasses various landscapes and demographic situations that challenge fairness and equity in service access. Similar challenges are faced in other regions with comparable characteristics. The Decision Support System developed in this work accurately models the real-world system and provides valuable feedback and suggestions to EMS professionals, enabling them to make informed decisions and enhance the efficiency and fairness of the system.

Digital pathology world tour.

Objective: Digital pathology (DP) is currently in the spotlight and is rapidly gaining ground, even though the history of this field spans decades. Despite great technological progress, the adoption of DP for routine clinical diagnostic use remains limited. Methods: A systematic search was conducted in the electronic databases Pubmed-MEDLINE and Embase. Inclusion criteria were all published studies that encompassed any application of DP. Results: Of 4888 articles retrieved, 4041 were included. Relevant articles were categorized as "diagnostic" (147/4041, 4%) where DP was utilized for routine diagnostic workflow and "non-diagnostic" (3894/4041, 96%) for all other applications. The "non-diagnostic" articles were further categorized according to DP application including "artificial intelligence" (33%), "education" (5%), "narrative" (17%) for reviews and editorials, and "technical" (45%) for pure research publications. Conclusion: This manuscript provided temporal and geographical insight into the global adoption of DP by analyzing the published scientific literature.

Harmonization of ICF Body Structures and ICD-11 Anatomic Detail: One foundation for multiple classifications.

The Family of International Classifications of the World Health Organization (WHO-FIC) currently includes three reference classifications, namely International Classification of Diseases (ICD), International Classification of Functioning, Disability, and Health (ICF), and International Classification of Health Interventions (ICHI). Recently, the three classifications have been incorporated into a single WHO-FIC Foundation that serves as the repository of all concepts in the classifications. Each classification serves a specific classification need. However, they share some common concepts that are present, in different forms, in two or all of them. For the WHO-FIC Foundation to be a logically consistent repository without duplicates, these common concepts must be reconciled. One important set of shared concepts is the representation of human anatomy entities, which are not always modeled in the same way and with the same level of detail. To understand the relationships among the three anatomical representations, an effort is needed to compare them, identifying common areas, gaps, and compatible and incompatible modeling. The work presented here contributes to this effort, focusing on the anatomy representations in ICF and ICD-11. For this aim, three experts were asked to identify, for each entity in the ICF Body Structures, one or more entities in the ICD-11 Anatomic Detail that could be considered identical, broader or narrower. To do this, they used a specifically developed web application, which also automatically identified the most obvious equivalences. A total of 631 maps were independently identified by the three mappers for 218 ICF Body Structures, with an interobserver agreement of 93.5%. Together with 113 maps identified by the software, they were then consolidated into 434 relations. The results highlight some differences between the two classifications: in general, ICF is less detailed than ICD-11; ICF favors lumping of structures; in very few cases, the two classifications follow different anatomic models. For these issues, solutions have to be found that are compliant with the WHO approach to classification modeling and maintenance.

The slow-paced digital evolution of pathology: lights and shadows from a multifaceted board.

Objective: The digital revolution in pathology represents an invaluable resource fto optimise costs, reduce the risk of error and improve patient care, even though it is still adopted in a minority of laboratories. Barriers include concerns about initial costs, lack of confidence in using whole slide images for primary diagnosis, and lack of guidance on transition. To address these challenges and develop a programme to facilitate the introduction of digital pathology (DP) in Italian pathology departments, a panel discussion was set up to identify the key points to be considered. Methods: On 21 July 2022, an initial conference call was held on Zoom to identify the main issues to be discussed during the face-to-face meeting. The final summit was divided into four different sessions: (I) the definition of DP, (II) practical applications of DP, (III) the use of AI in DP, (IV) DP and education. Results: Essential requirements for the implementation of DP are a fully tracked and automated workflow, selection of the appropriate scanner based on the specific needs of each department, and a strong commitment combined with coordinated teamwork (pathologists, technicians, biologists, IT service and industries). This could reduce human error, leading to the application of AI tools for diagnosis, prognosis and prediction. Open challenges are the lack of specific regulations for virtual slide storage and the optimal storage solution for large volumes of slides. Conclusion: Teamwork is key to DP transition, including close collaboration with industry. This will ease the transition and help bridge the gap that currently exists between many labs and full digitisation. The ultimate goal is to improve patient care.

Automated ICF Coding of Rehabilitation Notes for Low-Resource Languages via Continual Training of Language Models.

The coding of medical documents and in particular of rehabilitation notes using the International Classification of Functioning, Disability and Health (ICF) is a difficult task showing low agreement among experts. Such difficulty is mainly caused by the specific terminology that needs to be used for the task. In this paper, we address the task developing a model based on a large language model, BERT. By leveraging continual training of such a model using ICF textual descriptions, we are able to effectively encode rehabilitation notes expressed in Italian, an under-resourced language.

Natural Language Processing to extract SNOMED-CT codes from pathological reports.

Objective: The use of standardized structured reports (SSR) and suitable terminologies like SNOMED-CT can enhance data retrieval and analysis, fostering large-scale studies and collaboration. However, the still large prevalence of narrative reports in our laboratories warrants alternative and automated labeling approaches. In this project, natural language processing (NLP) methods were used to associate SNOMED-CT codes to structured and unstructured reports from an Italian Digital Pathology Department. Methods: Two NLP-based automatic coding systems (support vector machine, SVM, and long-short term memory, LSTM) were trained and applied to a series of narrative reports. Results: The 1163 cases were tested with both algorithms, showing good performances in terms of accuracy, precision, recall, and F1 score, with SVM showing slightly better performances as compared to LSTM (0.84, 0.87, 0.83, 0.82 vs 0.83, 0.85, 0.83, 0.82, respectively). The integration of an explainability allowed identification of terms and groups of words of importance, enabling fine-tuning, balancing semantic meaning and model performance. Conclusions: AI tools allow the automatic SNOMED-CT labeling of the pathology archives, providing a retrospective fix to the large lack of organization of narrative reports.

HEROHE Challenge: Predicting HER2 Status in Breast Cancer from Hematoxylin-Eosin Whole-Slide Imaging.

Breast cancer is the most common malignancy in women worldwide, and is responsible for more than half a million deaths each year. The appropriate therapy depends on the evaluation of the expression of various biomarkers, such as the human epidermal growth factor receptor 2 (HER2) transmembrane protein, through specialized techniques, such as immunohistochemistry or in situ hybridization. In this work, we present the HER2 on hematoxylin and eosin (HEROHE) challenge, a parallel event of the 16th European Congress on Digital Pathology, which aimed to predict the HER2 status in breast cancer based only on hematoxylin-eosin-stained tissue samples, thus avoiding specialized techniques. The challenge consisted of a large, annotated, whole-slide images dataset (509), specifically collected for the challenge. Models for predicting HER2 status were presented by 21 teams worldwide. The best-performing models are presented by detailing the network architectures and key parameters. Methods are compared and approaches, core methodologies, and software choices contrasted. Different evaluation metrics are discussed, as well as the performance of the presented models for each of these metrics. Potential differences in ranking that would result from different choices of evaluation metrics highlight the need for careful consideration at the time of their selection, as the results show that some metrics may misrepresent the true potential of a model to solve the problem for which it was developed. The HEROHE dataset remains publicly available to promote advances in the field of computational pathology.

Preliminary Validation of a Rule-Based System for Mortality Coding Using ICD-11.

A crucial process for world-level mortality statistics is the capability to identify the underlying cause of death from death certificates. Currently such certificates are coded using ICD-10. The selection of the underlying cause is done by means of semi-automated rule-based systems. However, starting from 2022, countries should begin to adopt ICD-11, for which no system is already available. The present paper describes the architecture of a novel system for automated UC selection, with classification-independent rules, and its preliminary validation on two sets of death certificates coded with ICD-10 and ICD-11.

Ontological modeling of the International Classification of Functioning, Disabilities and Health (ICF): Activities&Participation and Environmental Factors components.

BACKGROUND: The International Classification of Functioning, Disability and Health (ICF) is a classification of health and health-related states developed by the World Health Organization (WHO) to provide a standard and unified language to be used as a reference model for the description of health and health-related states. The concept of functioning on which ICF is based is that of a "dynamic interaction between a person's health condition, environmental factors and personal factors". This overall model has been translated into a classification covering all the main components of functioning. However, the practical use of ICF has highlighted some formal problems, mainly concerning conceptual clarity and ontological coherence. METHODS: In the present work, we propose an initial ontological formalization of ICF beyond its current status, focusing specifically on the interaction between activities and participation and environmental factors. The formalization has been based on ontology engineering methods to drive goal and scope definition, knowledge acquisition, selection of an upper ontology for mapping, conceptual model definition and evaluation, and finally representation using the Ontology Web Language (OWL). RESULTS: A conceptual model has been defined in a graphical language that included 202 entities, when possible mapped to the SUMO upper ontology. The conceptual model has been validated against 60 case studies from the literature, plus 6 ad-hoc case studies. The model has been then represented using OWL. CONCLUSIONS: This formalization might provide the basis for a revision of the ICF classification in line with current efforts made by WHO on the International Classification of Diseases and on the International Classification of Health Interventions.

Best Practice Recommendations for the Implementation of a Digital Pathology Workflow in the Anatomic Pathology Laboratory by the European Society of Digital and Integrative Pathology (ESDIP).

The interest in implementing digital pathology (DP) workflows to obtain whole slide image (WSI) files for diagnostic purposes has increased in the last few years. The increasing performance of technical components and the Food and Drug Administration (FDA) approval of systems for primary diagnosis led to increased interest in applying DP workflows. However, despite this revolutionary transition, real world data suggest that a fully digital approach to the histological workflow has been implemented in only a minority of pathology laboratories. The objective of this study is to facilitate the implementation of DP workflows in pathology laboratories, helping those involved in this process of transformation to identify: (a) the scope and the boundaries of the DP transformation; (b) how to introduce automation to reduce errors; (c) how to introduce appropriate quality control to guarantee the safety of the process and (d) the hardware and software needed to implement DP systems inside the pathology laboratory. The European Society of Digital and Integrative Pathology (ESDIP) provided consensus-based recommendations developed through discussion among members of the Scientific Committee. The recommendations are thus based on the expertise of the panel members and on the agreement obtained after virtual meetings. Prior to publication, the recommendations were reviewed by members of the ESDIP Board. The recommendations comprehensively cover every step of the implementation of the digital workflow in the anatomic pathology department, emphasizing the importance of interoperability, automation and tracking of the entire process before the introduction of a scanning facility. Compared to the available national and international guidelines, the present document represents a practical, handy reference for the correct implementation of the digital workflow in Europe.

Logical Rules and a Preliminary Prototype for Translating Mortality Coding Rules from ICD-10 to ICD-11.

Iris is a system for coding multiple causes of death in ICD-10 and for the selection of the underlying cause of death, based on a knowledge base composed by a large number of rules. With the adoption of ICD-11, those rules need translation to ICD-11. A pre-project has been carried out to evaluate feasibility of transition to ICD-11, which included the analysis of the logical meta-rules needed for rule translation and development of a prototype support system for the expert that will translate the coding rules.

Classification of Histologic Images Using a Single Staining: Experiments with Deep Learning on Deconvolved Images.

The automated analysis of digitized immunohistochemistry microscope slides is usually a challenging task, because markers should be analysed on the tumor area only. Tumor areas could be recognized on a different slide, stained with Haematoxylin-Eosin. The basic idea of the present poster is to evaluate how well deep learning methods perform on the single haematoxylin component of staining, with the prospective possibility of developing a classifier able to recognize tumor areas on IHC slides on their haematoxylin component only. In a preliminary experiment, single stain images obtained by H-E color deconvolution showed an accuracy of 0.808 and 0.812 for Hematoxilyn and Eosin components, respectively.

A Communication Infrastructure for the Health and Social Care Internet of Things: Proof-of-Concept Study.

BACKGROUND: Increasing life expectancy and reducing birth rates indicate that the world population is becoming older, with many challenges related to quality of life for old and fragile people, as well as their informal caregivers. In the last few years, novel information and communication technology techniques generally known as the Internet of Things (IoT) have been developed, and they are centered around the provision of computation and communication capabilities to objects. The IoT may provide older people with devices that enable their functional independence in daily life by either extending their own capacity or facilitating the efforts of their caregivers. LoRa is a proprietary wireless transmission protocol optimized for long-range, low-power, low-data-rate applications. LoRaWAN is an open stack built upon LoRa. OBJECTIVE: This paper describes an infrastructure designed and experimentally developed to support IoT deployment in a health care setup, and the management of patients with Alzheimer's disease and dementia has been chosen for a proof-of-concept study. The peculiarity of the proposed approach is that it is based on the LoRaWAN protocol stack, which exploits unlicensed frequencies and allows for the use of very low-power radio devices, making it a rational choice for IoT communication. METHODS: A complete LoRaWAN-based infrastructure was designed, with features partly decided in agreement with caregivers, including outdoor patient tracking to control wandering; fall recognition; and capability of collecting data for further clinical studies. Further features suggested by caregivers were night motion surveillance and indoor tracking for large residential structures. Implementation involved a prototype node with tracking and fall recognition capabilities, a middle layer based on an existing network server, and a Web application for overall management of patients and caregivers. Tests were performed to investigate indoor and outdoor capabilities in a real-world setting and study the applicability of LoRaWAN in health and social care scenarios. RESULTS: Three experiments were carried out. One aimed to test the technical functionality of the infrastructure, another assessed indoor features, and the last assessed outdoor features. The only critical issue was fall recognition, because a slip was not always easy to recognize. CONCLUSIONS: The project allowed the identification of some advantages and restrictions of the LoRaWAN technology when applied to the health and social care sectors. Free installation allows the development of services that reach ranges comparable to those available with cellular telephony, but without running costs like telephony fees. However, there are technological limitations, which restrict the scenarios in which LoRaWAN is applicable, although there is room for many applications. We believe that setting up low-weight infrastructure and carefully determining whether applications can be concretely implemented within LoRaWAN limits might help in optimizing community care activities while not adding much burden and cost in information technology management.

Marker controlled superpixel nuclei segmentation and automatic counting on immunohistochemistry staining images.

MOTIVATION: For the diagnosis of cancer, manually counting nuclei on massive histopathological images is tedious and the counting results might vary due to the subjective nature of the operation. RESULTS: This paper presents a new segmentation and counting method for nuclei, which can automatically provide nucleus counting results. This method segments nuclei with detected nuclei seed markers through a modified simple one-pass superpixel segmentation method. Rather than using a single pixel as a seed, we created a superseed for each nucleus to involve more information for improved segmentation results. Nucleus pixels are extracted by a newly proposed fusing method to reduce stain variations and preserve nucleus contour information. By evaluating segmentation results, the proposed method was compared to five existing methods on a dataset with 52 immunohistochemically (IHC) stained images. Our proposed method produced the highest mean F1-score of 0.668. By evaluating the counting results, another dataset with more than 30 000 IHC stained nuclei in 88 images were prepared. The correlation between automatically generated nucleus counting results and manual nucleus counting results was up to R2 = 0.901 (P < 0.001). By evaluating segmentation results of proposed method-based tool, we tested on a 2018 Data Science Bowl (DSB) competition dataset, three users obtained DSB score of 0.331 ± 0.006. AVAILABILITY AND IMPLEMENTATION: The proposed method has been implemented as a plugin tool in ImageJ and the source code can be freely downloaded. SUPPLEMENTARY INFORMATION: Supplementary data are available at Bioinformatics online.

Detection of HER2 from Haematoxylin-Eosin Slides Through a Cascade of Deep Learning Classifiers via Multi-Instance Learning.

Breast cancer is the most frequently diagnosed cancer in woman. The correct identification of the HER2 receptor is a matter of major importance when dealing with breast cancer: an over-expression of HER2 is associated with aggressive clinical behaviour; moreover, HER2 targeted therapy results in a significant improvement in the overall survival rate. In this work, we employ a pipeline based on a cascade of deep neural network classifiers and multi-instance learning to detect the presence of HER2 from Haematoxylin-Eosin slides, which partly mimics the pathologist's behaviour by first recognizing cancer and then evaluating HER2. Our results show that the proposed system presents a good overall effectiveness. Furthermore, the system design is prone to further improvements that can be easily deployed in order to increase the effectiveness score.