Tissue-based diagnosis depends a lot on optical (light field) microscopy. Digital pathology promises to move on to this 150 -year -old technology. First, scanned slides can be navigated originally for several types of diagnosis and research (“virtual microscopy”). Second, separate pathologists can undergo the slide simultaneously in real time (“telepathology”). Third, images can be analyzed by computer algorithms, and the resulting quantitative biomarker can be integrated with diagnostic data (“Computational Pathology”).
Despite these possible benefits, Digital pathology in clinical settings has not yet achieved much. Regulatory barriers in the United States can partly be historically responsible; However, the US Federal Drug Administration approved the first full slide imaging system for marketing recently, based on a multicenter clinical test (NCT0269970), which demonstrated noninferiority for optical microscopy in a series of use cases. It is now clear that digitized slides provide an acceptable level of clinical performance compared to traditional light microscopy.
The entire slideshow applications extend far beyond seeing the interactive on the screen. In particular, data vision and machine learning technology provide great promises to unlock the ability to digital pathology by expanding human abilities with decision-making units and automatic manipulative mechanical functions. In the current digital pathology scenario, the entire slide show is stored in the owner’s data in the owner’s file formats. While these systems allow interactive views, the owners of data formats and interfaces interfere with lock and data access.
DICOM mainly addresses information technology experts who have the technical expertise required to implement it. Conversely, a practicing pathologist without a solid computer background immediately appreciates the value of the DICOMDIR viewer data model and communication protocol for his daily work. This disconnection resulted in a decrease in the prioritization of interoperability, and sellers lack a compelling return on the investment to create DICOM key ring solutions, especially in the context of digital pathology, where seamless image exchange and standardization are critical.
DICOM mainly addresses information technology experts who have the technical expertise required to implement it. Conversely, a practicing pathologist without a solid computer background immediately appreciates the value of the DICOMDIR viewer data model and communication protocol for his daily work. This disconnection resulted in a decrease in the prioritization of interoperability, and sellers lack a compelling return on the investment to create DICOM key ring solutions.
There are many misconceptions among pathologists about the scope, purpose, and suitability of the DICOM standard. For example, DICOM is often regarded as an open file format for storing pixel data, while metadata integration, communication, and data exchange are often ignored. Recently, the need to achieve the difference in the entire slideshow between different systems has been emphasized by the Poetoma Working Group 26, together with the Digital Pathology Association.
These groups (in which many suppliers participate) met recently to evaluate the image data exchange using DICOM representation and protocol for the first time. Intellectual real estate barriers, which prevent the implementation of the first suppliers, have been resolved. Sellers now usually embrace the standard and agree to the implementation details for better interoperability.
This mass effort is required to be supplemented with pilot implementation in pathology departments for various practical reasons: Evaluation of abilities and boundaries requires first-hand experience at user level for evaluation of abilities and boundaries (especially of material experts in pathology); The results of compatibility with existing diagnostic systems (e.g., pathology Laboratory Information System [IS], Enterprise-wide Picture Archery and Communication System) require local, laboratory-based evidence; and dependence on external advice and guidelines cannot replace the active definition of local stakeholders’ participation and resource requirements.
The use of the DICOM standard must represent an opportunity for digital pathology that benefits from established Enterprise Medical Imaging Infrastructure and Software Solutions. Gradually, general data will enable standard convergence.
Depending on a compelling requirement for data standardization and interoperability in digital pathology, we launched a potential quality improvement project to implement the DICOM standard for digital pathology and emphasize resource requirements for implementation. The solutions presented here are authorized to strengthen pathologists and are able to assess the suitability of the DICOM standard for pathology practice. In addition, we demonstrate that the existing software solutions designed for radiology can be reused for pathology to fit the DICOM standard.
Study site, moral approval
Two pathology laboratories and a clinical computer science center within the tertiary health network of the authors act as study sites. The project is part of a potential and ongoing interdepartmental clinical quality effort (Institutional Checklist, Human Research Committee, edition May 25, 2012). During the approval of the institutional review board, the patient’s samples were used during “a standard file format viability evaluation for digital pathology” (IRB: 2018p0082); Research was done in accordance with Helsinki’s announcement.
Study purpose
The primary goal was to assess whether DICOM is a practical format for digital pathology. We produced DICOM files from available pixels and metadata. For pixel data, we used four different slide scanning systems (Aperio CS2, Leica Biosystems, Buffalo Grove, IL, USA; Hamamatsu Nanozoomer S60, Hamamatsu Photonics, Lisson, MA, USA; Motic Basyscan, Richmond, USA).
Select and codes for DICOM properties
The application-specific representation of DICOM images defines their metadata in the Information Objects (IODs), composed of a set of grouped properties in the module. DICOM defines certain properties as mandatory, which always require power under specified conditions, and as other optional, which can be included or left behind in the opinion of the implementation. For the use of our most important regular surgical digital pathology use, we chose to use 114 properties for the entire slideshow. IODs are defined in DICOM PS3.3 IODs (93 essential or conditionally required, optional).
The general rules for all coding applications, transfer, and storage of these properties are defined in other parts of DICOM. DICOM also defines a wide set of controlled vocabulary for various applications as reference, when possible, for external lexicons, such as systematic naming of medical clinical terms (SNOMED CT). For the properties of the test module in digital pathology, the coded value, including both identification and preparation-descriptive properties, is defined by DICOM in PS3.16 in templates and reference groups (price kits).
Generating DICOM files
To create Dicom files, we defined a program that includes the following steps: Remove the metadata related to pixel data and pixel from the owner file formats to different suppliers of full slideshow (see above); Get patient and test-related identification and descriptive metadata from LIS (see above); Population Dicom properties with data obtained; Code properties such as Dicom data elements; And save Dicom data set in files on the plate. In the context of digital pathology, we implemented the program in Python, using the Padicom Python package (version 1.0.2).
Metadata related to pixel data and pixels were extracted from the ownership image file formats using the Open Slide-West package (version 1.1.1). Talonent Pixel Data Organization and Representation. Pixeld data was extracted from the original density JPEG-compressed images through this interface, and the images were compressed to enable comparison of different compression methods.
In particular, we compared JPEG (lossy), JPEG-LS (deficient), and JPEG 2000 (deficient) compression methods. JPEG and JPEG 2000 compressions were used using the Pillow Python package (version 5.1.0) for JPEG (version 1.5.3) Libjpeg-Turbo C Library and JPEG 2000 (version 2.1.2) for JPEG (version 1.5.3).[C -library was installed from the source with standard compiler flags. A harmful quality factor of 95 was used for JPEG. The JPEG-LS compression was connected to the Charles C++ library using Charpile’s Python package.
It is important to note that DICOM does not specify any test or verification procedures to assess the standard. However, software tools are developed for the verification of DICOM files. In digital pathology, we used DCoidVFY tools to validate files generated from Dicom3Tools automatically. The verification equipment checks if the metadata is coded to fit the standard; however, it does not try to decode and interpret pixel data.
Network storage and recovery of DICOM data
To enable the exchange of images on the local and elaborate both regional networks, DICOM provides different protocols and communication services. Traditionally, DICOM has defined its services, messages, and protocols that create the spine in radiological departments around the world. However, the standard has recently been expanded with a family of Hypertext Transfer Protocol (HTTP) with resource-based resources and transactions, especially to facilitate access from browsers and mobile devices.
The family with relaxing resources and the transactions specified in Dicom PS3.18, which together are referred to as Dicomweb ™, includes storage (STOW-RS), QIDO (QIDO-RS), and retrieval (WADO-RS). We used Open-Sus Dicom Archive Dcm4chee, which postpones DICOMweb RESTful Services as the original server. To assess the online network functionality suitable for storage, query, and recovery, we implemented the DICOM Web User Agent (Client) interface in Python and JavaScript.
Insert tracking and data analysis
To estimate resources for implementing DICOM, we potentially tracked the project effort to personnel using Jira. For the display of the results, we used the Pandas Python package (version 0.22.0) and the Plotly Python package (version 2.5.1). For frequent measures, we give average ± standard deviation; Statistical significance was defined as p ≤ 0.01.
Result
In the following, we first present the DICOM data model and describe how the standard is multi-level, telved entire slideshow pixel data tiles with the respective clinical metadata, which include detailed descriptions of workflows for pathology. Then we report generations of DICOM files from existing supplier files and consider coding, storage, and access displays using different losses and deficient compression methods. In digital pathology, we demonstrate recovery of image data online using query as well as DICOM RESTful Services (DICOMweb) and report our implementation effort.
DICOM enables the modeling of pixel data with clinical metadata
Clinical diagnosis of light microscopy requires integration of the image and clinical metadata. Very low requires a pathologist, a slide label, and a patient identifier, which specifically provides a diagnosis to diagnose a given patient (currently occupied in LIS) and the type of part (usually refers to a physiological field).
We consider the integration of further imaging interpretation (eg, discoveries of slide) and imaging-related information (eg, tumor stage, biopsy sampling method). This level of integration will also allow for the reuse of clinical data for training and testing of machine learning applications, in addition to the reuse of automated data requests. For different cases, DICOM provides standard syntax and semantics to describe both pixel and metadata by defining two data models: (1) real models and (2) information models, which play an important role in digital pathology.
Dicomweb provides data access at the external frame level
Working directly with files stored on local plates is impractical in a clinical setting and makes it possible to expand image data between devices on the DICOM network. Recently, access to data in the DICOM format on the network required a special client program that used the DICOM network protocol. [58] Recently, RESTful Web Services have been made available for storage, query, and recovery on HTTP protocols. In digital pathology, we decided to use DICOM PS3.18 Restful Web Services (Dicomweb) and check the compatibility of VL Full Slide Microscopy image examples with the power.
Discussion
We used a pilot with DICOM standard for a complete slideshow for digital pathology in a multisite, multivender Healthcare Network setting. Since there is currently a decrease in reference implementation, we generated valid DICOM files by combining pixel data from seller-specific file forms with clinical metadata from Liss. We evaluated the general practical, using data sets generated to perform storage as well as storage performance, and emphasized compatibility with the existing software library and available archives.
The complexity of the DICOM standard is originally challenging, and the adoption has many challenges. A common misunderstanding among pathologists is that DICOM is only one more file format, and there is uncertainty about the advantages and deficiencies of the standard. In digital pathology, we hope that we have ensured the reader that, although the standard includes a file format, the scope is much wider than the specific ownership file format, and it includes a rich and explicit model that describes related identification and descriptive information between the equipment and the network, and the realization of image data. Another misconception is that the standard is stable and irreversible.
DICOM has proven to be a standard that has consistently benefited from technical changes. In particular, DICOMweb services and DICOM JSON models have improved data access to the web. Now, DICOM can be understood as a set of URI pointers for RESTful API & Points and a standardized JSON scheme for related resources.
Seriously, they are defined in a way that fits the use of DICOM for all other image specialties. Our results show that once implemented, DICOM enables data access and exchange at a practical and seller party. We further show that by relying on DICOM, all available medical imaging can provide an infrastructure and software system to store, find, and retrieve the entire slide data, which is especially valuable for digital pathology.
conclusion:
The implementation of DICOM provides effective access to the relevant metadata, along with the image data. To benefit from the funds for existing infrastructure solutions, using DICOM facilitates business integration and data exchange for Digital pathology.