Digital pathology and biobanks

02-05-2023 pag. 1

How digital pathology can add value to your biobank

Yves Sucaet, PhDDiabetes Research Center

CMI

02-05-2023 pag. 2

Who am I?

Education• 2000: Hogeschool Gent

(BE)– BS Computer Sciences

• 2001-2005: Troy State University (US)– Exchange program– MS Biological Sciences

• 2005-2010: Iowa State University– PhD Bioinformatics &

Computational Biology

Professional

• 2000-2001: Becton Dickinson

• 2010-2013: HistoGeneX• Data Management• Bioinformatics• Section head

• 2014-2016: VUB• Digital Pathology Manager

02-05-2023 pag. 3

Presentation outline

• Background– What is Whole Slide Imaging (WSI)– What is Digital Pathology

• What is a biobank?• Adding value to your biobank– By adding slide imaging data– By adding (semi)automated image analysis

• Things to consider

Innoviris portal introductieBackground

02-05-2023 pag. 5

What is Whole Slide Imaging (WSI)?

• The technique of converting glass-mounted microscopy material into a digital representation / image.

02-05-2023 pag. 6

Free University of Brussels setup

02-05-2023 pag. 7

Implementing WSI for end-users

Enhanced network infrastructure

I gave you slides; how do I get to

see these?We’ll send you a link that you can open with

your webbrowserCool!

02-05-2023 pag. 8

Am I doing digital pathology yet?

• Whole Slide Imaging– A technique– Get a scanner– Generate LOTS of data– Easy

• Digital (histo)pathology– A method, • a way of thinking

– Workflow management, integration

– USE the data • valorization

– Hard

02-05-2023 pag. 9

So you want to do digital pathology?

Education

Biobanking

Second opinion

Primary diagnosis

Research

QA

Start small; select one use case; implement it RIGHT; scale and build

on your success

Innoviris portal introductieWhat is a biobank?

02-05-2023 pag. 11

Why do we build biobanks?

– Show-case for patients and samples.

– Facilitates patient selection for studies.

– Permanent record of cases allows retrospective validation of studies.

02-05-2023 pag. 12

Flemish Biobank (CMI – BBMRI. vl)

• Minimal clinical data (OECD)• Extended epidemiological dataset• HLA, auto-immune data• Images

02-05-2023 pag. 13

Source material: slides and clinical data

Innoviris portal introductieAdding value to your biobank

02-05-2023 pag. 15

Opportunities for digital pathology

• Without digital histopathology: • With digital histopathology:

I’m looking for breast tumor tissue

Does the sample contain cancer cells

Is it in situ or invasive?

Is it hormone-responsive?

I’m looking for breast tumor tissue

Does the sample contain cancer cells

Is it in situ or invasive?

Is it hormone-responsive?

Sure; glad to help

Just trust us

Just trust us

We don’t have that information

Sure; glad to help

Have a look at the HE

Have a look at the HE and the immunostains (p63 – calponine)

Have a look at the immunostains and the fluo

data (FISH)

Fidelity increases, confident about requests, investments pay off

02-05-2023 pag. 16

Valorization through digital pathology

As you add more layers of data and move further down the pyramid, the value of the initial sample in the biobank increases and the biobank as a whole is likely to prove more useful.

02-05-2023 pag. 17

Step 1

02-05-2023 pag. 18

Step 1: adding imaging datahttp://histosrv.vub.ac.be/dbb

02-05-2023 pag. 19

Step 1: adding imaging datahttp://histosrv.vub.ac.be/dbb

02-05-2023 pag. 20

Step 1: adding imaging datahttp://histosrv.vub.ac.be/dbb

02-05-2023 pag. 21

Biobank “behind the scenes”

Custom ASP.net website http://histosrv.vub.ac.be/dbb

02-05-2023 pag. 22

Pathomation software architecture

Custom built

Public portal

02-05-2023 pag. 23

Step 2

02-05-2023 pag. 24

Step 2: image analysis

Step 1: find tissue

(Based on work by J. Watson, Pathology Visions meeting 2012)

02-05-2023 pag. 25

Step 2: image analysis

Step 1: find tissue

Step 2: locate the islets

(Based on work by J. Watson, Pathology Visions meeting 2012)

02-05-2023 pag. 26

Step 2: image analysis

Step 1: find tissue

Step 2: locate the islets

Step 3: quantitate insulin

(Based on work by J. Watson, Pathology Visions meeting 2012)

02-05-2023 pag. 27

The result: intelligent querying

I’m looking for pancreatic tissue from a patient with

recent onset diabetes and a susceptible HLA-DQ

genotype. At least y% of the islets still have to contain

insulin-producing cells

Here you go

Innoviris portal introductieThank you for taking the time to review these materialsContact me at [email protected] to continue the conversation

02-05-2023 pag. 29

Acknowledgements

• Diabetes Research Center, VUB– Yves Sucaet– Silke Smeets– Stijn Piessens– Peter In’t Veld

• Dept of Pathology, UZ Brussel– Wim Waelput– Ramses Forsyth

CMI