System and method for processing backlogs of medical images

Abstract

A method for enhancing diagnostic accuracy and efficiency in radiology using an Al- driven platform, the method comprising: receiving a plurality of medical imaging request and determining relative urgency through a pre-determined set of key protocols; de-identifying the patient's medical images to ensure confidentiality and converting the images into a format suitable for analysis; enhancing image quality for improved diagnostic review; analyzing the enhanced medical images using an Al- powered application that employs machine learning algorithms to detect anomalies, patterns, and diagnostic indicators; generating a clinical report and a patient-specific report using natural language processing techniques based on the analysis conducted by the application; facilitating a collaborative review process among healthcare providers via a secure communication interface; enabling patient access to their medical images and reports through secure login mechanisms.

Description

TITLESystem and Method for Processing Backlogs of Medical ImagesBackground

[0001] The present invention pertains to the technical field of medical imaging and radiology diagnostics. More specifically, it relates to an Artificial Intelligence (Al) driven platform designed to enhance diagnostic accuracy and efficiency in radiology through advanced image processing, analysis, and collaborative review mechanisms.

[0002] Recent advancements in medical imaging technologies and Al have significantly impacted the field of radiology, leading to the development of systems like the "Cognitive Collaboration with Neurosynaptic Imaging Networks, Augmented Medical Intelligence and Cybernetic Workflow Streams" (Publication number: 20190355483) and technologies for "Extracting Fine Grain Labels from Medical Imaging Reports” (Publication number: 20220108070). These innovations enable improved communication, collaboration, and cognitive enrichment through the use of neurosynaptic networks, as well as enhanced medical report analysis using Natural Language Processing (NLP) techniques. Despite these advancements, there remains a critical need for integrated solutions that can further streamline radiology workflows where there are large backlogs of images, improve diagnostic accuracy, and facilitate secure and efficient collaboration among healthcare providers.

[0003] The current landscape of medical imaging and diagnostics faces several challenges, including but not limited to, the handling of large volumes ofimaging data, maintaining patient confidentiality while ensuring data accessibility, and the integration of Al tools in a manner that complements the diagnostic process without replacing the critical role of human expertise. These challenges underscore the need for an innovative solution that not only leverages the strengths of Al and machine learning (ML) in image analysis but also addresses large backlogs of medical images and enhances the collaborative diagnostic process among healthcare professionals as well as improves patient engagement in their care journey.Summary

[0004] In view of the circumstances outlined above, aspects of the present described invention systems and methods for the implementation of method for enhancing diagnostic accuracy and efficiency in radiology by tackling a backlog of medical images using an Al-driven platform, the method comprising: receiving a medical imaging request and determining its urgency through a pre-determined set of key protocols; de- identify! ng the patient’s medical images to ensure confidentiality and converting the images into a format suitable for analysis; enhancing image quality for improved diagnostic review; analyzing the enhanced medical images using an Al-powered application that employs machine learning algorithms to detect anomalies, patterns, and diagnostic indicators; generating a clinical report and a patientspecific report using natural language processing techniques based on the analysis conducted by the application; facilitating a collaborative review process among healthcare providers via a secure communication interface; and enabling patient access to their medical images and reports through secure login mechanisms.

[0005] The objectives of the present invention are to provide a comprehensive Al-driven platform that: efficiently manages and prioritizes medical imaging requests: ensures patient confidentiality through de-identification of medical images; applies advanced image processing techniques for enhanced diagnostic reviews; utilizes Al and ML algorithms for accurate anomaly detection and diagnosis; generates detailed clinical and patient-specific reports using NLP; and facilitates seamless collaboration among healthcare providers and patients through secure communication interfaces and access controls.

[0006] The proposed invention offers significant improvements over existing solutions by advanced algorithmic insights, reducing diagnostic imaging backlogs, and enhancing patient outcomes through a comprehensive approach that integrates Al and ML technologies for image analysis with a collaborative platform for healthcare providers and patients. This integration not only improves diagnostic accuracy and efficiency but also promotes transparency and informed decision-making, thereby addressing the critical gaps in the current technology landscape.Brief Description of the Drawings

[0007] FIG 1 illustrates technical architecture for embodiments of the present invention known as Imagini.

[0008] FIG 2 illustrates image capture and storage generating a backlog for input into the present invention.

[0009] FIG 3 illustrates processing a backlog of images and providing an output.

[0010] FIG 4 illustrates reporting and review of results.Detailed Description of the Drawings

[0011] Backlogs of medical images waiting for review presents a significant problem in the health care system. Radiology delays have a cascading effect on downstream healthcare and can lead to increased morbidity and mortality from a variety of conditions.

[0012] A hospital or other health care provider may have thousands or tens of thousands of medical images waiting for review for patients experiencing conditions such as cancer, heart disease, and many others.

[0013] The present invention which is known as Imagini's RapidReview: Backlog Buster interacts with this backlog, to autonomously scan and prioritize imaging reviews.

[0014] FIG 1 illustrates technical architecture for embodiments of the present invention known as Imagini.

[0015] The present invention can integrate with electronic health records platforms like EPIC, Veradigm, and others. A connection can be made through a Fire API on FHIR standard.

[0016] Public endpoint API access can be attained or in the alternative Secure File Transfer Protocol (SFTP), a network protocol for securely accessing, transferring and managing large files and sensitive data, can be used.

[0017] Lambda compute, a serverless, event-driven compute service, can run code for the application without provisioning or managing servers.

[0018] A Relational Database Service contains DB instances with various storage types. DB instance storage comes in the following types: GeneralPurpose (SSD), Provisioned IOPS (PIOPS), and Magnetic, each with different rates of retrieval.

[0019] A Dynamo Database uses hashing and B-trees to manage data. Upon entry, data is first distributed into different partitions by hashing on the partition key.

[0020] User pool token handling and management is provided on the client side through Cognito SDKs.

[0021] CloudFront is used as a content delivery network (CDN) service in the embodiment. The CDN improves efficiency by introducing intermediary servers between the client servers. These CDN servers manage some of the client-server communications. They decrease web traffic to the web server, reduce bandwidth consumption, and improve the user experience of your applications.

[0022] FIG 2 illustrates image capture and storage generating a backlog for input into the present invention.

[0023] A patient and health care provider decide to order imaging either in response to a patient’s complaint or as a screening tool. An image is obtained and stored in a picture archiving and communication system (PACS), a computerized means replacing the roles of conventional radioiogical film. Providers of cloud-based PACS often use a hybrid cioud system in which primary images are stored on the customer's premises and backups are kept in the cioud. Additional types of storage architectures may be configured and attached to the PACS server, such as direct-attached storage (DAS), network- attached storage (NAS) or a storage area network (SAN), each allowing forupgradeability, connectivity, improved protection against failure and added security.

[0024] DICOM provides a standard file format and network protocol to enable the storage and retrieval of medical images in a PACS. However, PACS vendors employ various syntaxes within DiCOM, which can make it difficult to use data from one system in another medical system. Because of the lack of a consistent standard, vendor neutral archive (VNA) technology has replaced PACS in some healthcare settings and integrated with PACS in others. VNA enables data integration by deconstructing data from an originating PACS and then migrating the data to the new system with the proper syntax. For the purposes of the present invention, both PACS and VNA can be used.

[0025] FIG 3 illustrates processing a backlog of images and providing an output.

[0026] Images are generally obtained in raw format i.e. , a DICOM format. The system of the present invention can parse data from the meta data in the DICOM file such as name, birthday, age, gender, which helps to verify the patient’s identity and track images prior to de-identifying of data.

[0027] Then, image data is brought into an S3 cloud object store, where it goes through an ETL process to ingest the store file. Quality checks are performed and the data is enriched for use downstream in engine itself.

[0028] Data can be formatted into parquet file format structured by row, with every separate column independently accessible from the rest. Since the data in each column is expected to be of the same type, the parquet file format makes encoding, compressing and optimizing data storage possible.

[0029] Athena allows for query of the data from parquet files stored in the S3 cloud object store. Athena is an interactive query service that makes it easy to analyze data directly from S3 using standard SQL. Athena is serverless, so there is no infrastructure to set up or manage and you can start analyzing your data immediately. Athena works directly with data stored in S3.

[0030] The present invention’s RapidReview artificial intelligence system goes through the S3 cloud object store and processes the backlog of image stored therein. If it finds, e.g., large cancer mass, it can bump up the priority of an image.

[0031] The client organization can set parameters for the RapidReview of backlogs by utilizing controls with sliders on a graphical user interface (GUI), i.e., a list of sliders for various diseases that a client organization can use to move particular conditions and findings up and down in priority. For example, an organizations can set thresholds for e.g., size of probable tumor mass, to prioritize.

[0032] Organizations can also set a variety of alert and notification protocols based on the same types of findings, e.g., a tumor mass of a specified size found on an image.

[0033] FIG 4 illustrates reporting and review of results.

[0034] A radiologist can be notified of results ready for review and can log into the Imagini system, then navigate to the RapidReview feature. The RapidReview system can present a list of imaging sorted by priority based on Imagini's prior scan.

[0035] A radiologist can then review images and text generated by Imagini’s Natural Language Processing (NLP) module configured to execute NLP algorithms to generate clinical and patient-specific reports. The Natural Language Processing module can be configured for Natural Language Generation, otherwise known as NLG, a software process driven by artificial intelligence that produces natural written or spoken language from structured and unstructured data, i.e., the images. The Imagini NLP module is able to understand how the medical images relate to actual text based on the training of the model to associate image findings with textual reports.

[0036] The radiologist and the health care provider ordering the imaging can collaborate to review, note discrepancies with the report, and communicate securely to resolve them.

[0037] Finally, the patient can be notified of the report on the imaging and review the report.

[0038] The illustrations of embodiments described herein are intended to provide a general understanding of the structure of various embodiments, and they are not intended to serve as a complete description of all the elements and features of apparatus and systems that might make use of the structures described herein. Many other embodiments will be apparent to those of skill in the art upon reviewing the above description. Other embodiments may be utilized and derived therefrom, such that structural and logical substitutions and changes may be made without departing from the scope of this disclosure. Figures are also merely representational and may not be drawn to scale. Certain proportions thereof may be exaggerated, while others may be minimized. Accordingly, the specification and drawings are to be regarded inan illustrative rather than a restrictive sense. Thus, although specific embodiments have been illustrated and described herein, it should be appreciated that any arrangement calculated to achieve the same purpose may be substituted for the specific embodiments shown. This disclosure is intended to cover any and all adaptations or variations of various embodiments. Combinations of the above embodiments, and other embodiments not specifically described herein, will be apparent to those of skill in the art upon reviewing the above description. Therefore, it is intended that the disclosure not be limited to the particular embodiments ) disclosed.

Claims

What is claimed is:1 . An Al-driven diagnostic enhancement system specifically designed for processing medical imaging data, comprising:• a cloud-based platform configured to receive, store, and manage medical imaging data securely;• an Al analysis module integrated within the platform, designed to analyze medical imaging data using ML algorithms to identify diagnostic indicators, anomalies, and patterns indicative of medical conditions;• an image processing module for de-identifying, converting, and enhancing the quality of medical images for analysis;• a module of Large Language Model configured to execute NLP algorithms to generate clinical and patient-specific reports;• a collaboration interface facilitating secure communication and collaborative review of diagnostic data and reports among healthcare providers;• and patient interface features enabling patients access to medical images and reports, wherein:• the system receives a backlog comprising a plurality of medical images and prioritizes the backlog for processing the received medical imaging data based on urgency, wherein urgency is assessed based on predefined criteria comprising patient symptoms, type of request, and healthcare provider's assessment and the criteria are used for deeplearning to identify patterns, anomalies, and diagnostic indicators within complex medical images.

2. The system according to claim 1 , wherein the image processing module is further configured to remove all personally identifiable information (Pll).

3. The system according to ciaim 1 , wherein the image processing module supports multiple medical imaging formats, comprising DICOM, JPEG, PNG, and TIFF.

4. The system according to claim 1 , wherein the image processing module applies advanced image processing techniques, comprising noise reduction, contrast enhancement, and edge sharpening.

5. The system according to claim 1 , wherein the Al analysis module utilizes deep learning algorithms specifically trained to identify particular types of anomalies, patterns, and diagnostic indicators relevant to a range of medical conditions, comprising tumors, fractures, and cardiovascular diseases.

6. The system according to claim 1 , further configured to undergo selfimprovement by retraining its data-driven machine learning algorithms.

7. The system according to claim 1 , wherein the application is further configured to incorporate visual aids, including annotated images and graphs, in the clinical and patient-specific reports.

8. The system according to claim 1 , further comprising an integrated feedback mechanism, designed to capture and incorporate input from healthcare providers and patients regarding the accuracy and clarity of the generated reports.

9. The system according to claim 1 , wherein the collaboration interface supports multimedia messaging, comprising text, voice, and video.

10. The system according to claim 1 , further comprising integration with electronic health record (EHR) systems to automatically update patient records with the generated reports and / or any subsequent diagnostic conclusions.11.The system according to claim 1 , further comprising a module within the application for tracking and managing the diagnostic imaging backlog, including real-time updates on imaging request status.

12. The method for enhancing diagnostic accuracy and efficiency in radiology using an Al-driven platform, the method comprising:• receiving a backlog containing plurality of medical images and determining relative urgency through a pre-determined set of key protocols;• de-identifying the patient's medical images and metadata to ensure confidentiality and converting the images into a format suitable for analysis;• enhancing image quality for improved diagnostic review;• analyzing the enhanced medical images using an Al-powered application that employs machine learning algorithms to detect anomalies, patterns, and diagnostic indicators;• generating a clinical report and a patient-specific report using natural language processing techniques based on the analysis conducted by the application;• facilitating a collaborative review process among healthcare providers via a secure communication interface;• enabling patient access to their medicai images and reports through secure login mechanisms.

13. The method of claim 12, further comprising automatically prioritizing the medical imaging requests based on the determined urgency, wherein the urgency is determined based on predefined criteria comprising patient symptoms, type of request, and healthcare provider's assessment.

14. The method of claim 12, wherein the converting step includes supporting multiple medical imaging formats, including DICOM, JPEG, PNG, and TIFF.

15. The method of claim 12, further comprising applying advanced image processing techniques during the image quality enhancement step,comprising noise reduction, contrast enhancement, and edge sharpening techniques.

16. The method of claim 12, wherein the analyzing step involves the use of deep learning algorithms specifically trained to identify particular types of anomalies, patterns, and diagnostic indicators relevant to a range of medical conditions, comprising tumors, fractures, and cardiovascular diseases.

17. The method of claim 12, wherein the generation of the clinical report and patient-specific report is enhanced by the inclusion of visual aids designed to cater to the needs of visually impaired users. These aids comprise carefully annotated images and descriptive graphs, each tailored with accessibility features such as high-contrast markings, tactile graphics, and audio descriptions, facilitating a comprehensive understanding of the diagnostic findings for all patients, including those with visual impairments. This approach ensures that the reports are not only detailed and informative for healthcare professionals but also inclusively designed to support patient engagement and understanding across a broad spectrum of visual capabilities.

18. The method of claim 12, further comprising a feedback mechanism for healthcare providers and patients to provide input on the accuracy and clarity of the reports generated.

19. The method of claim 12, further comprising integration with EHR systems to automatically update patient records with the generated reports and any subsequent diagnostic conclusions from the collaborative review process.

20. The method of claim 12, wherein enabling patient access includes providing educational materials tailored to the patient's diagnosed condition(s), using the advanced NLP techniques to generate easily understandable content based on the analysis conducted by the application, sending the content electronically to the radiologist for transmission to the ordering physician, receiving a notice of a discrepancy electronically from the ordering physician, editing the content, and providing the content to the patient electronically.

Images