Digital Pathology The Swedish Experience. Anna Bodén MD, Pathologist and Project leader

Transcription

1 Digital Pathology The Swedish Experience Anna Bodén MD, Pathologist and Project leader

2 Linköping, Sweden

3 Department of Clinical Pathology Number of Requests/year Histopathology Cyt Autopsy MO Screening cyt Sweden: appr. 10 milj habit. Area: km² (22,07 habit. /km² ) 23 pathology departments About 260 pathologists Östergötland: Area: 9979 km² (44 habitants/km²) 17 senior pathologists 6 training pathologists

4 The making of a pathology report

5 Introduction Background and deployment Work and integration Diagnostic work Future plans

6 Linköping University Hospital Whole slide scanning and diagnostic in routine > slides > 440 TB 27 digital workstations All histopathological slides scanned Combination diagnostic Reviews Multidisciplinary team meetings Distant consultations Cytology (ROI) for MDT meetings Supporting molecular workflow 2017 first part: primary digitally diagnosed (30%)

7 Timeline digital system implementation Linköping st digital system Lab integration 10 WS Digipat Digipat3 2 nd digital system Lab workflow 20 WS Defined Management and prioritized working area

8 Digital implementation, timeline Define the digtal system; scanners + share + viewer + workstations Investments Technical specification System adaptation and innovation Scanning routine Laboratory changes User aspects + medical assurance Training Verifying Validating Pathologists in validation as a fraction of total SOFT TISSUE SKIN LYMFOMA HEMATOLOGY LIVER AND PANCREAS ALIMENTARY TRACT PROSTATE OCH URO ENDOCRINE LUNG PLACENTA AND PERIN HEAD AND NECK NEURO GYN BREAST Adaptation + innovation New workprocedure Diagnostic criteria Image analysis Decision support in diagnostic

9 Digital system, experience 2010 Aperio scanners Image scope 10 workstations 4MP screen and administration screen Local share, spectrum database Software solution for management of nonradiologic images; Picsara 2015 Aperio and Hamamatsu scanners Sectra pathology PACS, (viewer, informationssystem, share structure) 27 workstations 8 MP medical technical screens Connection through LIS

10 Level of integration 2017

11 Shopping and to-do list going digital To buy Microscopic scanners Computers Data storage and structure Good network Screens Computer devices Integration with LIS Software to image the virtual slides and to assist your diagnostic work To resourse Pathologists Laboratory commitment Scanning personel IT people Medical technical knowledge Image analysis knowledge Project management Vendor commitment Interest from leaders, politicians, researchers etc

12 You need a team! Pathologist Laboratory technician IT technician

13 Digital pathology workflow and whole slide imaging Reduced transport and assorting Work-flow efficiency Ergonomic friendly Digitally supported documentation and traceability Computer aided diagnostic (CAD) tools Integration to LIS, RIS; reduce error rates related to specimen-patient mismatching Secondary viewing, sign-out confirmation, QA, MDT Education Access to pathology services in under sourced locations Existing LIS and digitally immature support implementation throughout laboratory and administrative workprocesses Investment in technical equipment Different image and data standards IT infrastructure and regulations Virtual slide image quality Image acquisition Implementation costs Digital system management and data storage costs Regulatory approval Pathologists confidence and acceptance

14 Workstation Specification Work station Monitor (Primary diagnostic) Model Typ Resolution Eizo Radiforce RX850 Color TFT LCD Panel (IPS) 4096 x 2160 (17:9 viewing rate) Monitor2 (Administrative screen) Other Standard keyboard Standard mouse 3D-mouse Spacemouse Pro 3DConnection Computer Model: HP Z640 CPU: Intel (R) Xeon (R) CPU E GHz 2.40 GHz RAM: 12gb DISK: 128gb SSD Graphic card: Nvidia Quadro K4200

15 Monitor size, pixels, color Work station 4 Mp 6 Mp 4 Mp

16 2016 workstations for both diagnostic modalities

17 Introduction Background and deployment Work and integration Diagnostic work Future plans

18 Sten Thorstenson, Jesper Molin, Claes Lundström. Implementation of large-scale routine diagnostics using whole slide imaging in Sweden: Digital pathology experiences J pathol Inform :14

19 Lab- scan issues

20 Laboratory adjustments = image quality issues! Less tissue on each slide and in each block Tissue position on the slide Reduce space on slide Coverslide, mounting media, airbubbles Label = connection to LIS Dried slides necessary, 60 o C for 1-2 hours

21 Mount and airbubbles 2. Folds. 3. Unknown 4. Scanning area outside coverslide 4 2

22 Scanning routine workflow Labscan team Loading the slides into the trays/racks Loading and unloading the scanner Confirmation that all tissue on the slide is captured Other QA image issues Signing out the slide/case Confirmation that the case is complete according to LIS information Real slides +Virtual slides

23 Scanning routine

24 Scanner Ewerlöv and Boklöv Hamamatsu NanoZoomer-XR 320 slides Hasselblad ; Hamamatsu NanoZoomer-XRL, 160 large slides Aspelöv and Lindelöv ; Aperio Scanscope, 400 slides About 1000 slides/day (mean 6 slides/case) About slides/year > slides have been scanned so far Turn around time is increased in the initial laboratory workprocess Rescanning first and second less than 1% (?)

25

26 Diagnostic image quality Diagnos routine 1. Dehydration 2. Thick tissue slide 3. Fixation artefacts 4. Material outside cover slide 5. Same slide different scanners Tissue and stain variation Scanner Compression Focus points Scanning x20/x40/x60 Layers Colour Viewer Monitor Size and resolution

27 Same scanner brand, different cameras 2. Same slide different focus points 3. Different scanner brands

28 Validation digital system, scanner, monitor Microscopic slide/digital image Fixed monitor Same test slides Color Texture Area and focus Impact on diagnostic fidelity?

29 Introduction Background and deployment Work and integration Diagnostic work Future plans

30 Macro-Micro

31 Annotations

32 Q

33 Mitotic count 10 HPF = 2mm² = 0,50 mm field diameter

34 Mitotic count 10 HPF = 2mm² = 0,50 mm field diameter

35 Image analysis Automation vs human controlled Stain variation, scanner colour variation, overlapping nucleus, multiple slides. IHC stain quantfication Segmentation of desired regions or objects in the slide Classification Prognosis Increase reproducablility Effectivity

36 Integrating pathology and radiology

37 Introduction Background and deployment Workintegration Diagnostic work Future plans

38 Diagnostic network and collaborations

39 Increasing quality Digital archive=second opinion and review Integrated diagnostic=efficient MDT and QA 12-20% major discrepancies Second opinion can statistically significantly improve diagnostic accuracy of interpretation of breast histopathology Second opinion improve but do not completely eliminate diagnostic variability Evaluation of 12 strategies for obtaining second opinions to improve intepretation of breast histopathology: simultation study, Elmore et al, BMJ 2016;353:3069

40 Digital tools: Scale stain

41 15% increased effeciency, retained accuracy

42 Collaboration - you need a team 2! University Industry CMIV, center of medical imaging and visualisation AIDA- Analytic Imaging Diagnostics Arena (AIDA) (aim to create a national arena for research and innovation analytical imaging) DROID -Diagnostic Reference Oncology Imaging Database annotated images

43 Introduction Background and deployment Workintegration Diagnostic work Future plans

44 DigITal position 2017 in Linköping We see the advantages of having digital pathology integrated at the laboratory not only for primary diagnostic issues Solid technical system for share, store and viewing is the core Collaboration between IT and pathology knowledge, as well as laboratory personal, is crucial Research and development of diagnostic tools is good for collaboration (hospital-university-industry) Need of established (virtual)networks for facilitating second opinion

45 Acknowledgements Darren Treanor Martin Hallbeck Claes Lundström Jesper Molin Bethany Williams Maria Vrang Robert Ring, Ann-Charlotte Mengel Helén Richard, Marie Ahnström-Walterson Carl Clasén Karin Skoglund, Arrigo Capitanio, Stina Garvin, Fredrik Bäckström Sten Thorstenson

46 Contacts Project leader Digital Pathology Anna Bodén, M.D. Head of department, Anatomical Pathology Martin Hallbeck, Associate Professor, M.D. Ph.D.