Decision Support in Pathology:  
Franklin AI Digital Prostate

Book a DEMO

Franklin.ai is on a mission to offer advanced Decision Support AI solutions,
designed to assist pathologists in improving daignostic consistency and supporting workflow efficiency within computational pathology.

Blank space

Predictions indicate that the number of new cases of cancer will reach 35 million by 2050 (77% increase)*

With pathologists responsible for diagnosing 100% of all cancers, their workloads are only set to increase.

*Global cancer statistics 2022: GLOBOCAN estimates of incidence
and mortality worldwide for 36 cancers in 185 countries

Advanced Decision Support AI solutions for digital pathology.

Franklin.ai Digital offers advanced Decision Support AI solutions for digital pathology. Franklin.ai Digital enhances diagnostic accuracy and workflow efficiency in computational pathology. By pre-analysing cases, our Decision Support AI flags the most severe specimens, allowing pathologists to navigate through cases with unparalleled efficiency.

Once new cases have been processed through a whole slide image scanner in at your laboratory, Franklin AI Digital Prostate can be configured to pick up those new cases as requiring analysis. The AI model will then analyze each slide, looking to identify various findings and calculate measurements before finally pre-opulating the synoptic reports with Gleason grading and other prognostic information.

This is all designed to happen before a pathologist starts work for the day. This process can then assist in helping them decide which cases to start with, and then which specimens within a case to look at first.

Blank space

Helping digital pathologists do more with their time

With rising caseloads and mounting pressure on pathology services, Franklin.ai is helping pathologists do more with their time, by streamlining routine tasks, supporting diagnostic decisions, and improving overall workflow efficiency. Our goal is to help unlock healthcare capacity, without compromising quality or control.

We envision a world where more patients are on the right treatment path sooner, enabled by faster, more consistent diagnostic support.

In fact, our research shows that pathologists produce more accurate Prostate PCB and TURP reports when assisted by Franklin AI Digital Prostate.

  • Diagnostic support for both Prostate Core Biopsies and Transurethral Resection of the Prostate (TURP) specimens
  • Detailed ontology, covering 45 findings enabling malignancies, prognostic factors and benign findings to be detected and displayed for pathologist review
  • Pre-filled reports including involvement and grading quantification exportable into downstream lab information systems

Supporting lab directors looking at capacity solutions

When building pathology AI, Franklin.ai is committed to providing labs with reliable and robust solutions. Here are some quick highlights:

  • Franklin AI Digital Prostate performs across various slide scanners
  • Franklin AI Digital Prostate performs across different labs
  • Pathologists using Franklin AI Digital Prostate report that it supports their clinical decision-making, helping them make more confident and informed diagnoses.
  • Pathologists using our platform saw an average ~30% improvement in efficiency compared to working without AI assistance
  • Pathologists using Franklin.ai Digital Prostate have reported workflow improvements and greater efficiency, helping them dedicate more time to patient care.
  • Franklin AI Digital Prostate is robust even in the presence of technical artefacts

Changing healthcare for the better with Decision Support AI

At Franklin.ai, we believe collaboration is key to revolutionising healthcare. By partnering with clinicians, laboratories, hospitals, technicians, and the IVD and medical device community, we aim to empower pathologists to perform at their best, ensuring patients receive the timely and accurate care they deserve.

Learn more about our Pathology AI Solutions.

‍

FAQs

How will stain variation affect performance? Is on-site calibration or training needed? How generalisable is this (e.g., to different staining methods, operators, etc.)?
Franklin Digital was trained using a large and diverse dataset with varying H&E stain characteristics to ensure it is robust against lab-to-lab variability. The model has been optimised to handle different staining techniques, ensuring consistency in results across various labs. While the AI is designed to adapt to different conditions, on-site calibration may still be beneficial for certain labs to ensure the highest accuracy, particularly when using non-standard staining protocols. We offer support for on-site validation and training to help your lab optimise performance with Franklin Digital, ensuring it performs accurately across different operators and lab conditions.
How do I know the AI will work with my lab's specific conditions?
To ensure compatibility with your lab’s specific conditions, Franklin Digital is highly flexible and can be tailored to fit your workflows and sample types. We provide a pre-deployment validation process, where we run slides from your lab through the AI model to assess its accuracy and reliability. This step ensures that the product is suited to your lab’s unique staining methods, slide preparation techniques, and equipment. If adjustments are necessary, we provide full support to ensure that Franklin Digital delivers optimal performance in your lab environment.
Is this a local app or browser-based? What integrations does Franklin AI Digital Prostate support?
Franklin Digital will be browser based, which can be launched either directly or with contexet from a 3rd party IMS (e.g. Sectra). It integrates with major IMS and LISHL7, FHIR, and custom interfaces. We also support a variety of file formats such as .svs and .dicom.
How will the AI know to differentiate between different cores vs. levels?
Franklin Digital uses a segmentation model to differentiate between cores and levels on a slide. This model is specifically trained to recognise and segment tissue at different specimen levels. This ensures accurate classification and quantification, allowing the AI to provide precise findings for each core or level of the tissue sample. The segmentation process is integral to ensuring that histopathological findings are reported at the correct specimen level, giving pathologists a clear understanding of the tissue being analysed.
Can your AI detect a specific finding? Can your AI detect rare cancers?
Franklin Digital has been trained on a comprehensive dataset that includes a wide range of histological findings, including rare cancers. The AI uses deep learning techniques to identify and classify malignant and benign tissue types, including rarer cancers that pathologists may encounter. If you are working with particularly rare cancer types, we encourage collaboration with our team to ensure the AI meets your needs. Get in touch with us if you’d like to know more about the full ontology of findings we support.
How will the AI know to differentiate between different cores vs. levels?
The AI uses a segmentation model to differentiate between cores and levels on a slide. This model is specifically trained to recognise and segment tissue at different specimen levels. This ensures accurate classification and quantification, allowing the AI to provide precise findings for each core or level of the tissue sample. The segmentation process is integral to ensuring that histopathological findings are reported at the correct specimen level, giving pathologists a clear understanding of the tissue being analysed.
Why do you show findings like “Acinar Adenocarcinoma”?
Franklin Digital will also show findings from what we call our ontology tree, which categorises findings from broadest (the ‘trunk’), down the ‘branch’ and finally ending at the ‘leaf’. An example of this, is ‘adenocarcinoma, acinar’ -> ‘acinar gleason pattern 4’ -> ‘cribriform’, The reason we show all findings that the AI may identify this condition with a high degree of confidence, and we want to ensure that pathologists have all relevant findings available for review. By highlighting these findings, we allow pathologists to confirm and categorise the malignancy using an intuitive workflow, ensuring that no potential malignancy is missed during the review process. This goes towards improving the explainability of our model which we know is important to healthcare practitioners.
Your vendor neutral partner

We understand laboratories want choices in the vendors they work with. We partner and integrate with your suppliers of choice to ensure a seamless digital pathology solution.

Company logo of SectraCompany logo of Path-aiCompany logo of LeicaCompany logo of PhilipsCompany logo of 3DHistechCompany logo of HammamatsuCompany log of Amazon AWS