Key Features You Should Look For In Pathology Image Analysis Software

22 Dec, 2021 | Blog posts, Prisma

Advanced pathology image analysis software transforms biomedical research practice. To provide you with a brief overview of the bioimage analysis tools in digital pathology, we have reviewed the most popular types of software solutions currently available. Uncover the numerous benefits of integrating these tools into your laboratory workflow.

The rise of chronic diseases worldwide and the ever-increasing governmental emphasis on research activities for new drug discovery and disease treatment drive business growth in the digital pathology market. This sector is booming, with an expected market growth of 7.5% by 2030.

The pathology image analysis market
The pathology image analysis market

Pathology software products can fall into several major segments based on their area of use:  

  • Lab information management software (LIMS), 
  • Laboratory information systems (LIS),
  • Pathology training software,  
  • Whole slide imaging  (WSI) software, 
  • Pathology image analysis software.   

The image analysis software segment currently dominates the digital pathology market. The growth in this sector is in line with the increasing demand for advanced integrated solutions and the rising awareness of pathologists on the possibilities of healthcare IT. Expansion possibilities primarily revolve around new business models using AI technology in bioimage analysis. Let’s explore how state-of-the-art solutions will help you achieve unprecedented quality and reproducibility of results faster and easier than ever.

A closer look at the benefits of pathology image analysis software    

Whole slide image analysis plays a vital role in contemporary life science research. Current digital pathology practice involves the scanning (slide imaging) and processing of large and complex WSI datasets. Advanced image analysis solutions allow pathologists to quickly and efficiently analyze digital whole slide images while generating reliable quantitative results.

Enhancing pathology image analysis outcomes with the help of AI 

State-of-the-art bioimage analysis solutions use advanced artificial intelligence methods. These methods significantly speed up research processes and reduce the amount of resources needed. Software companies introduce smart bioimage analysis solutions that can learn to autonomously analyze image data, recognize patterns, and generate predictions using diverse input images.

AI-backed applications mimic the workflow used by pathologists to perform image analysis. Object and pattern recognition relies on the consensus among experts (ground truth) who have drawn annotations on the input images. It helps standardize the analytical workflow and rule out intra- and interobserver variability.

Did you know?

Interobserver: Variability of the measurements between different observers.

Intraobserver: Variability in repeated measurements by the same observer.

Why use AI-powered software for analyzing bioimage data?

  • Increase productivity
  • Improve result quality
  • Gather reproducible results
  • Arrive at results faster
  • Standardize workflows
  • Save resources  
  • Collaborate remotely  

Standardize your image analysis processes using specialized AI-based software.

Choosing the right pathology image analysis tool for your research

Artificial intelligence capabilities in pathology image analysis can be put into a hierarchical order depending on task complexity. Primary cell and tissue structure quantification tasks usually lie at the bottom of this hierarchy. Researchers view the automated identification of tissue structures and the spatial relationships between them as more complex functionalities (Tosun et al. 2020). 

Pathology Image Analysis Software Capabilities
Hierarchy of image analysis software capabilities based on Tosun et al. (2020).

Our team has reviewed the features available in modern bioimage analysis solutions to assist you in selecting the most suitable tools for your research project. 

Computational quantification of tissue structures    

AI-driven bioimage analysis software allows you to detect and count particular cells and morphological objects within digital slides. These can be, for example, particular atypical features of the tumor microenvironment that you view as hallmarks of cancer, including tumor cells, stroma cells, and lymphocytes. Apart from feature counting, AI-powered tools allow you to make objective quantitative measurements on several parameters to assess the nature of cell morphology. Such parameters usually depend on your research design, but often include the size and roundness of cells and cell nuclei.

Human prostate sample slide viewed in the image analysis software IKOSA
Human prostate sample slide viewed in IKOSA. (Image source: https://doi.org/10.5281/zenodo.3875786)

In histopathology, automated biomarker quantification plays a central role. Popular staining techniques used in IHC imaging include simultaneous multimarker staining and virtual multi-staining by computational methods (Koezel et al. 2019). 

AI-driven microscopy image analysis software can be employed to detect particular biomarkers identified through fluorescent staining or to accurately quantify signals from dyes like hematoxylin and eosin (H&E) as well as IHC staining. 

If existing analysis applications don’t perfectly serve the needs of your research project, you can also develop your own deep learning applications using IKOSA AI without the need for any coding. This approach is known as application training. App training provides you with exceptional flexibility, allowing you to customize the application to perfectly suit your requirements. 

Automating complex detection, segmentation, and classification tasks

State-of-the-art pathology image analysis solutions support the automation of complex analytical tasks like bioimage detection, segmentation, and classification.

Did you know?

Segmentation is a prerequisite to performing more complex spatial modeling.

Image segmentation is a computer vision technique for WSI analysis used by pathologists to identify and locate tissue components like cancer cells, cell nuclei, glands, or entire tumor regions. Automated slide segmentation enables researchers to locate cancer-related processes such as inflammation, fibrosis, and mitosis (Wang et al. 2019a). 

Cell nuclei belong to the most often segmented morphological features in pathology literature. An effective method for precise cell and nuclei segmentation is using advanced AI technology. With IKOSA AI, you can train a customized image analysis application that suits your specific requirements, all without the need to code. By utilizing Regions of Interest (ROIs), you can concentrate on a specific area within an image containing structures relevant to your research. Using the wide range of readily available annotation tools you can mark the objects of your interest. The segmentation of WSI slides with specialized pathology image analysis software also involves completing several preprocessing and processing steps. These include image normalization, color augmentation, model selection, and creation. Properly conducting each of these steps results in improved segmentation performance (Wang et al. 2019a).

Annotations of cell nuclei assigned to the labels positive or negative in the IKOSA Platform
Annotations of cell nuclei assigned to the labels “positive” or “negative” in the IKOSA Platform. (Image source: https://doi.org/10.5281/zenodo.3875786)

By employing classification, you can also group morphological objects from the same type by different labels. Thus, you can assign individual cell nuclei in histology images to categories such as “epithelial,” “inflammatory” or “fibroblast” nuclei (Madabhushi & Lee 2016).

Did you know?

Automated classification models assist in cancer grading.

The automated classification of tissue structures can help assign WSI images and Regions of Interest (ROIs) to a cancer grade category such as “benign,” “atypical,” “high-risk,” “low-risk”, “malignant,” “ADH,” “invasive carcinoma,” depending on the feature pattern detected (Tosun et al. 2020). 

User-friendly graphic interface and zero coding to analyze your pathology images. 

Advanced Region of Interest (ROI) feature 

Using the Region of Interest (ROI) feature provided in current pathology image analysis applications, you can define specific areas within your slides that you want your analysis to focus on.

Did you know?

Regions of Interest are the main diagnostic areas within the pathology images.   

Sometimes, it makes sense to concentrate only on particular parts of a WSI slide for the purposes of segmentation tasks. These are usually the main malignant or the most diagnostically impactful areas. In some cases, it is reasonable to check ambiguous or borderline areas within the images (Wang et al. 2019; Tosun et al. 2020). In such cases, the ROI feature provided in your analysis tool can help you.

Using the versatile ROI features in IKOSA allows you to focus your analysis on particular sections within pathology images.

Current bioimage analysis software applications support features like ROI segmentation and generating quantitative data about tissue structures within the ROIs.

State-of-the-art spatial analysis options    

Digital pathology image analysis software is designed to gather high-content data about the spatial organization of tissue samples, which is otherwise very hard to acquire with traditional manual assessment methods. Adding a spatial context to your data can help you identify the distribution of specific cell populations (e.g., T-cells) and various inter- and intracellular signaling pathways.

Nuclei segmentation and nearest neighbor analysis within a multichannel image with our ready-to-use Sparkfinder App
Nuclei segmentation and nearest neighbor analysis within a multichannel image with our ready-to-use Sparkfinder App. The generated analysis report includes the X and Y coordinates of the detected objects, allowing you to align your data into a spatial context.

Furthermore, you can access reliable quantitative measurements on spatial metrics like density, coexpression (phenotyping), and colocalization (proximity) (Baxi et al. 2022).

Did you know?

The spatial arrangement of nuclei or glands in tissue images can help you categorize cancer grades.

Spatial biology techniques enables you to locate tumor-infiltrating lymphocytes by identifying intra-tumor lymphocytes (ITLs), adjacent-tumor lymphocytes (ATLs), and distal-tumor lymphocytes (DTL) (Yuan 2015).

Use existing AI apps or create your own.

Open-source pathology image analysis solutions for starters 

Some open-source software tools enable researchers to automate digital slide analysis. Most are desktop tools available for download, which users can enhance by installing additional plugins according to their needs.

QuPath is the most commonly cited open-source image analysis tool in pathology literature. The Centre for Cancer Research at the Queen’s University Belfast developed it as a quantitative pathology and bioimage analysis system. QuPath provides capabilities like WSI viewing, annotation and ROI drawing, cell counting, classification and segmentation of objects and pixels, automated tumor and biomarker detection, estimation of stain intensity, batch processing, and data export. It also allows data exchange with other analytical applications like ImageJ and MATLAB. Among the obvious benefits of these open-source image analysis tools are their accessibility, affordability, and constant development (Humphries et al. 2021).

Most open-source pathology image analysis solutions offer advanced scripting features that enable users to develop and share their automated image analysis models. A highly engaged user community usually actively participates in the software development process.

Yet, some of these open-source applications have a few limitations in their functionalities, such as few image modification options, lack of support for annotations and multimodality images, and missing compatibility with some slide scanning devices. 

Open-source applications for whole-slide image analysis primarily aim to engage users in software development, allowing them to draft their plugins, scripts, and workflows. Yet, this requires advanced coding skills and the dedication of additional IT resources. Therefore, if you intend to use an open-source image analysis solution in your lab, you should be prepared for a steep learning curve and acquire some new technical skills.

Benefits of commercial pathology image analysis solutions to consider   

Using commercial pathology image analysis software brings additional benefits. Commercial cell image analysis products support a broader range of imaging modalities. These are generally more user-friendly as they offer a range of practice-proven image analysis modules developed through the joint effort of computer vision- and pathology experts.

Investing in commercial bioimage analysis software is an important decision for the success of your research project. It is essential to assess the features and analytical packages your lab requires. Here are a few pressing reasons to choose a commercial pathology image analysis solution.

  • User-friendliness and flexibility 
  • Constant development in dialogue with experts
  • High precision achievable 
  • Tailored analysis applications
  • Integration with slide scanning systems 
  • Resource efficiency
  • Learning materials and tutorials 
  • Fast user- and technical support   
Work collaboratively on your research projects using software with AI assistive features and accelerate your progress.

Integrated pathology image analysis solutions support broad-ranging functionalities such as image data viewing, augmentation, annotation, management, and sharing, which results in a faster discovery process. The IKOSA Platform is a holistic microscopy image analysis solution spanning AI-powered analysis and application  training features. The IKOSA product portfolio also includes a collection of ready-made pathology image analysis solutions combined with smart bioimage data management and user collaboration

Another aspect that sets the IKOSA Platform apart is its browser-based interface, which can be a game-changer for many users. Unlike some desktop solutions that demand substantial investments in additional IT infrastructure, IKOSA provides a seamless experience directly in your web browser. This accessibility opens doors for a wide range of users, from individual researchers to smaller organizations, to harness the power of image analysis without the burden of extensive IT investments. Users can access their projects and image analysis results from different devices at any time. This browser-based approach democratizes image analysis, making it more accessible and cost-effective for a broader audience.

Pathology image analysis becomes easier with assistive AI technologies.

Our authors:

KML Vision Team Benjamin Obexer Lead Content Writer

Benjamin Obexer

Lead content writer, life science professional, and simply a passionate person about technology in healthcare

KML Vision Team Elisa Opriessnig Content writer

Elisa Opriessnig

Content writer focused on the technological advancements in healthcare such as digital health literacy and telemedicine.

KML Vision Team Fanny Dobrenova Marketing Specialist

Fanny Dobrenova

Health communications and marketing expert dedicated to delivering the latest topics in life science technology to healthcare professionals.

References

Baxi, V., Edwards, R., Montalto, M., & Saha, S. (2022). Digital pathology and artificial intelligence in translational medicine and clinical practice. Modern Pathology, 35(1), 23-32. 

Humphries, M. P., Maxwell, P., & Salto-Tellez, M. (2021). QuPath: The global impact of an open source digital pathology system. Computational and Structural Biotechnology Journal, 19, 852-859. 

Koelzer, V. H., Sirinukunwattana, K., Rittscher, J., & Mertz, K. D. (2019). Precision immunoprofiling by image analysis and artificial intelligence. Virchows Archiv, 474, 511-522.  

Madabhushi, A., & Lee, G. (2016). Image analysis and machine learning in digital pathology: Challenges and opportunities. Medical image analysis, 33, 170-175. 

Tosun, A. B., Pullara, F., Becich, M. J., Taylor, D., Fine, J. L., & Chennubhotla, S. C. (2020). Explainable AI (xAI) for anatomic pathology. Advances in Anatomic Pathology, 27(4), 241-250.   

Wang, Shidan, et al. “Pathology image analysis using segmentation deep learning algorithms.” The American journal of pathology 189.9 (2019a): 1686-1698.  

Wang, S., Wang, T., Yang, L., Yang, D. M., Fujimoto, J., Yi, F., … & Xiao, G. (2019b). ConvPath: A software tool for lung adenocarcinoma digital pathological image analysis aided by a convolutional neural network. EBioMedicine, 50, 103-110.

Yuan, Y. (2015). Modelling the spatial heterogeneity and molecular correlates of lymphocytic infiltration in triple-negative breast cancer. Journal of The Royal Society Interface, 12(103), 20141153.

Categories

Join our newsletter