Digital Pathology Update

Transcription

1 Digital Pathology Update J. Mark Tuthill, MD Division Head, Pathology Informatics Department of Pathology & Laboratory Medicine Henry Ford Hospital Detroit, MI ASCT Webinar, January 2016

2 Objectives 1. Define digital pathology 2. Understand the variety of technologies that are available to accomplish digital pathology 3. Describe what the future holds for digital pathology J. Mark Tuthill, MD, Henry Ford Hospital 2

3 Widespread Uses of Digital Imaging in Pathology Documentation Medical Legal Education and Training Interesting cases, teaching sets, learning modules Clinical Communication Tumor boards, inter/intra departmental conferences, reporting Diagnostics Telepathology, automated pap screening J. Mark Tuthill, MD, Henry Ford Hospital 3

4 Digital Photography Advantages images immediately available for clinical use and assessment ease of storage and retrieval high availability, telepathology etc. no loss of images no degradation of quality readily available for multiple purposes easy editing diagnostic opportunities J. Mark Tuthill, MD, Henry Ford Hospital 4

5 Advantages of Digital Pathology Improved storage and retrieval More efficient workflow Error reduction Historical case retrieval and comparison More accurate timely analysis New analytic tools Image analysis Content based image retrieval Qualitative analysis Quantitative analysis Digital imaging based pathology tools J. Mark Tuthill, MD, Henry Ford Hospital 5

6 Virtual Microscopy Whole slide imaging (WSI) The ability to scan, store, retrieve and analyze entire pathology slides Storage, retrieval advantages compared to glass Large bandwidth and storage requirements data intensive for processing Image analysis opportunities Obvious integration with molecular technology Multi-spectral analysis Digital staining J. Mark Tuthill, MD, Henry Ford Hospital 6

7 Impact of Virtual Microscope on Digital Imaging Whole slide imaging With whole slide digitization there is a convergence of processes supporting: pathology digital imaging telemedicine applications Additional efficiencies and opportunities Education New diagnostic opportunities This will change how pathology is practiced! J. Mark Tuthill, MD, Henry Ford Hospital 7

8 Digital Pathology What is it? With the advent of practical whole slide imaging one can envision a new workflow with pathology Virtual microscope i.e. slide scanners Pathology slides will able to be fully represented digitally Rather than moving slides to diagnose, we will move digital image files Digital imaging tools will be used in tissue grossing Workflow analogous to current digital radiology Images moved and viewed on computer workstation J. Mark Tuthill, MD, Henry Ford Hospital 8

9 Digital Pathology Systems Digital pathology will provide diagnostic support abilities beyond the traditional uses of photo documentation systems These systems are becoming analogous to clinical pathology analyzers e.g. tissue imaging instrumentation Tissue analyzers This will allow for new diagnostic paradigms This technology is already used for primary cytopathology screening in some centers J. Mark Tuthill, MD, Henry Ford Hospital 9

10 What is Digital Pathology? Not just whole slide imaging! Distributed microscopic images Distributed EM Gross images Scanned documents Clinical lab images: gels, plates, etc Digital Pathology is: digitally capturing, storing, moving, analyzing, interpreting tissues submitted to the laboratory for diagnosis (and communicating those findings)

11 Aperio s

12 J. Mark Tuthill, MD, Henry Ford Hospital 12

13 Technology The Process of Digitization Technical Requirements Devices in Use at HFHS J. Mark Tuthill, MD, Henry Ford Hospital 13

14 Analog to digital... At the heart of any digital imaging system is some type of device that initially converts the analog images of everyday life to a realm of the numbers 0 and 1, the binary digital realm.

15 Analog to digital... With the advent of this deceptively simple conversion, perhaps seen by some as bordering on mystical, a new world of possibilities in imaging is within reach. -Ulysses J. Balis, MD Chair, CAP Informatics Committee Pathologist, Electrical Engineer

16 What is a CCD? The Heart of the digitizer CCD charge coupled device takes the place of film in the digital camera integrated circuit that uses the photoelectric effect with a analog to digital converter arrayed in a matrix or grid voltage is produced proportional to photons striking the charge coupled area each area corresponds to a pixel all CCD are black and white color is generated by in line filters

17 Traditional Photography

18 Digital Photography

19 The CCD Charge Coupled Device

20 CCD Function

21 CCD Function Black and White

22 CCD Function Color

23 Three Chip CCD

24 Pixel Resolution Resolution of an image refers to the detail that an image holds Equals image width x image height (measured in pixels) First number in pixel resolution equals the width (columns) and the second, the height (rows) Example: 1024 x 768 Can also be described by the total number of pixels in an image (megapixels) The higher the resolution, the more information the image contains

25 Pixel Density A digital image has no actual physical size until it is printed or displayed Once output, each printed/displayed image has a pixel density (number of pixels per unit length) Expressed as pixels per inch (PPI) or dots per inch (DPI)

26 Resolution Different images require different input resolutions for clarity (point discrimination) 1000X image 640 X 480 pixels 20X image 3000 X 3000 pixels higher the pixel depth more software zoom you can achieve more detail captured Images need to be input at high resolution viewing resolution will vary based on needs

27 Resolution Screen Captures

28 Digital Color All images on screen are comprised of variations of red, green and blue (RGB) Color depth (dynamic range) number colors used to represent an image number of shades used determines image color quality Number of bits of information used per color for a single pixel allows for depth A digital image with thousands or millions of colors provide life like or true color

29 Digital Image 1000 s of Colors

30 Digital Image 256 Colors

31 Gross Pathology Imaging Instrumentation Vended solution versus locally deployed Industrial grade Supported Computer based measurements Non destructive image annotation File versioning Workflow integration Bar code enabled, LIS integration J. Mark Tuthill, MD, Henry Ford Hospital 31

32 Milestone MacroPathD Gross Pathology Instrumentation

33 Other image Other image

34 Gross Workstation with Imaging

35 Microscopic Input Station

36 Gross Telepathology Station

37 Robotic Telemicroscopy

38 Current Digital Pathology Storage and Network Requirements Storage issues for 2 dimensional whole slide imaging (using compression) 1 slide in x, y 500MB 200,000 slides per year 1.0 x 10 (8) MB or 100 Terabytes Network requirements 8 x 10(8) Mbits At 100 mbs 93 days to transmit! (8 million seconds) At 1 gbs 9.3 At 10 gbs per.9 days to transmit J. Mark Tuthill, MD, Henry Ford Hospital 38

39 Digital Pathology Storage Requirements Storage issues for 3 dimension whole slide imaging 1 slide in x, y and z axis (11 planes) ~10 gigabytes (uncompressed) 200,000 slides per year 2,000,000 gigs 2 petabytes+ 2 Petabytes is a lot of hamburger! Current data center has less than two hundred terabytes Compression and other tricks may decrease this raw number but even at 100:1 this is a whole new world in medical data storage J. Mark Tuthill, MD, Henry Ford Hospital 39

40 Image Storage PACS Like Systems Computer file server Ample disk space, fast hard drive Files and folders work initially, but ultimately fail Image databases software input and catalog images search retrieve Ideally needs to be integrated with clinical databases for maximum utilization critical component of a clinically useful system very few out there that work well difference between cottage and turn key imaging systems J. Mark Tuthill, MD, Henry Ford Hospital 40

41 Image Storage PACS Like Systems Image databases software input and catalog images search retrieve Ideally needs to be integrated with clinical databases for maximum utilization critical component of a clinically useful system very few out there that work well difference between cottage and turn key imaging systems

42 Pathology PACS in your institution Apollo s PathPACS Solutions integrates into your LIS/HIS Systems Microscope Pathologist Remote Access Apollo PathPACS software Apollo Digital Pathology Solutions Enterprise PACS Acquisition Devices Information Systems Robotic Microscope Whole Slide Scanner Grossing Station Apollo PathPACS client software on Desktop PC PathPACS Image Management Workflow Data Access HL7 LIS Radiology Information System Hospital & Clinical Information Systems IS System Interfaces Electron Microscope Document Scanner, Other PathPACS Pathology Image Storage DICOM PACS Radiology Image Storage

43

44 TECHNOLOGY File mover services leveraged Grossing stations Pathology PACS Autopsy suite Autopsy suite Specimen X-rays Robotic Telepathology Document Scanners

45 Viewing Images in Apollo

46 Viewing Images in Apollo

47

48 CoPathPlus Attachment Edit/Entry

49 Image in CoPath Report

50 Barriers to Adoption of Digital Pathology Quality Z axis Time Infrastructure requirements Storage Network Pathology workstation Implementation LIS integration Analysis tool sets Systems design and management J. Mark Tuthill, MD, Henry Ford Hospital 50

51 Telepathology Network Henry Ford Hospital West Bloomfield Hospital Datacenter Rochester, Michigan Macomb Hospital Wyandotte Hospital

52 Methods of Accomplishing Telepathology Store and forward: asynchronous Live dynamic: synchronous Robotic telemicroscopy: both Whole Slide imaging (virtual microscopy) : both

53 Method Application Diagnosis Advanced QA Conferences Educ. Static Imaging (Lowest cost) Live Dynamic Robotic Microscopy Virtual Microscopy (Highest cost) Least Effective ++++ Most Effective

54 Robotic Telemicroscopy Remote control of a microscope + live dynamic video Synchronous or asynchronous Bandwidth similar to live dynamic Has integrated store and forward type technology useful for documentation User experience becomes nearly equivalent to microscopy

55 Data Center

56 Henry Ford Utilization of Telemicroscopy Carl Zeiss Mirax AT System 4 slide system; no robotics Location The system is located at West Bloomfield Hospital frozen section laboratory Most consults come from there to the other hospitals usually to the core laboratory at Henry Ford Hospital Case mix Majority cases are neuropathology or bone cases Occasional difficult case or intradepartmental consults (these numbers are not documented)

57 Digital Pathology and the Future Image Analysis and Diagnostic Support 57

58 Opportunities in Digital Pathology Business opportunities Diagnostic telepathology Primary cytopathology screening Machine Vision and Image Analysis provide improved care Quantitative analysis Brown stain counting Reflex testing Qualitative Image retrieval based on content Rare cell detection DNA analysis Quality Control Was the whole slide analyzed? Diagnostic algorithms Ease of retrieval for review New Digital staining Multispectral analysis Gross image systems Include measurement and annotation tools J. Mark Tuthill, MD, Henry Ford Hospital 58

59 Examples of Next Generation Advanced Image Analysis Content base image retrieval Multispectral Imaging Digital Staining J. Mark Tuthill, MD, Henry Ford Hospital 59

60 Histo-pattern Recognition through Machine Learning If histopathology patterns are in large part the basis for histopathologic diagnosis, can these patterns be analyzed and assessed by the computer? Machine learning in histopathology: Assign codes/diagnoses in a standard language to digitized pathology images Run thousands of such images through "untrained" computer and allow "machine learning" programs to assign binary morphologic attributes to specific diagnosis Use untested images on the educated computer and test for pattern recognition

61 Ulysses. J. Balis, MD Director of Clinical Informatics, Co-Director, Division of Informatics Department Pathology, University of Michigan Health System Presented at LabInfoTech Summit 2007

62

63

64

65 Computer aided quantification of biomarkers in histopathology specimens

66 Immunostaining provides specific molecular information Presence/absence of specific antigens is decisive Establish diagnosis Distinguish between competing diagnosis Highlight markers that predict response to various therapies Assess expression levels of antigens for prognostic/therapuetic uses Limitations of Visual Assessment Methods Qualitative and descriptive (0, 1+, 2+, 3+ staining) Subjective, depends on training and experience

67 Multispectral Image Analysis A multi-spectral image is one that captures image data at specific frequencies across the electromagnetic spectrum The wavelengths may be separated by filters Can allow extraction of information that the human eye cannot capture Originally developed for space-based imaging

68 Michael Feldman, MD, PhD, Assistant Professor Pathology University Pennsylvania, Presented at LabInfoTech Summit 2007

69 Michael Feldman, MD, PhD, Assistant Professor Pathology University Pennsylvania, Presented at LabInfoTech Summit 2007

70 Michael Feldman, MD, PhD, Assistant Professor Pathology University Pennsylvania, Presented at LabInfoTech Summit 2007

71 Hematoxylin, Cytokeratin, & HER2

72

73

74

75

76 Hematoxylin, cytokeratin, & HER2

77 Cell membrane boundries

78

79 Yellow dots = HER2+ cells

80 Ki-67

81 Composite

82 Ki-67 positive cells

83 1 st Round 2 nd Round 3 rd Round 4 th Round DAPI Cy3 a-actin Cy5 AR (-) Cy3 ER (-) Cy5 Her2 (+) Cy3 pan cytokeratin Patient tumor sample ER(-), AR(-), p53(+) red, Her2(+) white, keratin (green), actin (yellow), nucleus (blue) Cy3 pan cadherin Cy5-p53 (+) 5 Channel pseudo colored overlay

84 Flow on a slide

85 Digital Staining of Unstained Pathological Tissue Samples through Spectral Transmittance Classification Optical Review; The Optical Society of Japan, copublished with Springer-Verlag Volume 12, Number 1 Volume 12, Number 1 / January, Pinky A. Bautista, Tokiya Abe, Masahiro Yamaguchi, Yukako Yagi and Nagaaki Ohyama Abstract Histological structures of a pathological tissue sample convey information relevant to the diagnosis of the disease that might have afflicted the person. To reveal the morphology of these structures clearly, pathological tissues are stained. In this paper, a digital staining methodology for pathological tissue samples is introduced. Digital staining implies the application of digital processing techniques to transform the image of an unstained sample to its stained image counterpart. In the method, the transmittance spectra of the unstained and Hematoxylin and Eosin (H&E) stained multispectral images (16 bands) of specific tissue components are utilized. Two experiments were conducted to probe the possibility of the digital staining framework: the linear mapping of spectral transmittances, and the classification of spectral transmittances in conjunction with the linear mapping of specific transmittance data sets. The method classified the four tissue components, e.g. nucleus, cytoplasm, red blood cells, and the white region (region devoid of tissue structures), while the misclassifications between components with spectral transmittances that are closely similar were not completely rectified The Optical Society of Japan Key words Multispectral - digital staining - spectral transmittance - linear mapping - classification J. Mark Tuthill, MD, Henry Ford Hospital 85

86 Where are we now? Digital imaging in pathology has moved beyond photodocumentation Digital pathology will allow for new workflow and improved patient care Digital pathology will enable the development of tissue analyzers instruments for anatomic pathology Technology and system design barriers exist, but will ultimately be broken down Timeline for adoption: widespread use 3-5 years J. Mark Tuthill, MD, Henry Ford Hospital 86

87 The End! Questions? J. Mark Tuthill, MD Division Head, Pathology Informatics Department of Pathology & Laboratory Medicine Henry Ford Hospital Detroit, MI ASCT Webinar, January 2016