Summary of Hematology and Pathology Devices Panel Meeting

A summary of speeches and events.

Panel Members

NOTE: The list below is the same as that provided by the FDA with the addition of each members specialty.

Chair: Dorothy M. Adcock, MD
Medical and Laboratory Director
Esoterix Laboratory Englewood, Colorado
Specialty/Area of Expertise: coagulation

Secretary: Louise E. Magruder

FDA
Robert Becker, MD, PhD
CMO OIVD
Specialty: Previous R&D work included digital image analysis in Pathology.

VOTING MEMBER
Piotr Kulesza, MD, PhD
Assistant Professor of Pathology, Feinberg School of Medicine NW University Chicago, IL
Specialty: Cytophatology and Cancer Research

TEMPORARY VOTING MEMBERS
George G. Birdsong, MD
Director of Anatomic Pathology, Grady Health Systems Atlanta, GA
Specialty: Cytophatology and Anatomic Pathology (AP) Informatics

Diane D. Davey, MD
Professor of Pathology, University of Central Florida College of Medicine Orlando, FL
Specialty: Cytopathology and Hematology

David J. Foran, PhD
Professor of Pathology, Laboratory Medicine UMDNJ – Robert Wood Johnson Med. School Piscataway, NJ
Specialty: Image Analysis and Pattern Recognition

John R. Gilbertson, MD
Associate Chief of Pathology, Massachusetts General Hospital Boston, MA
Specialty: Pathology

Stephen M. Hewitt, MD, PhD
Clinical Investigator, PHS, CC Laboratory of Pathology National Cancer Institute, Bethesda, MD
Specialty: High throughput pathology and AP

John H. Sinard, MD, PhD
Professor of Pathology, Yale University School of Medicine New Haven, CN
Specialty: Pathology Informatics

Claudia Mello‐Thoms, PhD
Research Assistant Professor, Department of Biomedical Informatics UPMC Shadyside Hospital, Cancer Pavilion Pittsburgh, PA
Specialty: Image Perception and Informatics

Xiao‐Hua Andrew Zhou, PhD
Professor, Dept. of Biostatistics School of Public Health University of Washington Seattle, WA
Specialty: Statistical methods in diagnostic medicine

INDUSTRY REPRESENTATIVE
Dan Bracco Vice President, Clinical, Quality and Regulatory
Constitution Medical Inc. Westford, MA

CONSUMER REPRESENTATIVE
The consumer representative scheduled to be at this meeting was not able to attend and there was no time to select a replacement.

Day One Overview

Day One Introduction

Dr. Adcock gave some opening remarks about the purpose of the panel. Ms. Magruder made a compliance statement and reviewed some of the rules of order. Panel members were each asked to introduce themselves. There was a brief ceremony to recognize Ms. Magruder for her years of service.

NOTE: Given the new structure at FDA, Margaret Janicki will be replacing Louise Magruder. Ms. Magruder will focus on her work as a scientific reviewer.

Day One Invited and FDA Speakers

Historical Overview of FDA Regulation of Digital Pathology Imaging
Tremel Faison, MS, RAC, SCT (ASCP), Scientific Reviewer FDA/CDRH/OIVD/DIHD

Ms. Tremel provided an overview of current regulations governing microscopes and stated that based on 21 CFR 864.9, WSI is not exempt from 510(k) because it has a different fundamental scientific technology and if incorrect could result in serious health consequences. She differentiated WSI from current IVDs using digital imaging, which have specific and limited intended uses, and stated that those are not applicable to the WSI paradigm. In order to ensure safety and efficacy, FDA generally requires analytical and clinical studies to validate performance, knowledge of risks/benefits/limitations, standardization and postmarket studies.

Her presentation assumed the following intended use for digital pathology:

  • Primary diagnosis of surgical pathology microscope slides in lieu of a microscope
  • Not an adjunct
  • Broad application (not organ or disease specific).

Most subsequent speakers and panel members did not believe that digital pathology would be used instead of a microscope; instead they argued that the two would have to be used together.

Light Microscopy: The physical characteristics and measurements of optical imaging
Michael Descour, PhD, Associate Professor, College of Optical Sciences, University of Arizona

Dr. Descour provided a detailed technical overview of optical microscopes (OM) and aberrations commonly seen in light microscopy. Aberrations focused on taxonomy of aberrations (position dependent and independent), chromatic aberrations (related to wavelength) and distortion (field dependent). He discussed testing of microscope performance and provided a list of standards used for light microscopes.

Panel members asked several questions to understand different light sources, dynamic range of WSI compared to light microscopy, ability to manipulate the image and whether this is truly a benefit, and the depth of field.

Features Essential for Interpretation of Histology and Cytology Glass Slides
Ulysses Balis, MD, Associate Professor of Pathology and Director of the Division of Pathology Informatics, University of Michigan Health Systems

Dr. Balis discussed the limitations of optical systems, focusing on the lack of standardization. He noted that inherent limitations in the current optical systems include (1) diffraction limitation of light, (2) contrast, (3) overall calibration of microscopes, and (4) field of view capabilities. He reviewed histologic features essential for diagnosis, including pattern recognition/cytological feature classification at various magnifications, histologic structure and cell patterns. He further discussed the new risks and limitations presented by WSI. He felt that based on the lack of standardization and continued use of OM (optical microscopy) as a diagnostic tool, there is an inherent robustness in the human visual diagnostic interpretive process. This process, he noted, would carry over to WSI. For “inadequate” WSIs, a pathologist would defer (common practice). In concluding, he noted that adoption of WSI should be validated based on organ system/diagnostic entity basis, by use of kappa statistical approach (NOTE: more work is needed here due to inter-viewer variability), in concert with ROC (region of concern) analysis.

After the presentation, Dr. Davey noted that her concern is that with WSI the file view is so large that rare events are missed. Dr. Zhou asked several questions related to reader variability using WSI and OM, and use of a gold standard. There was no clear answer, however, Dr. Balis did note that WSI is lower in diagnostic detection. Dr. Gilbertson asked about color, which Dr. Balis responded is not a big concern. There was also a comment made that WSI are currently used on the board exams.

Digitization of Glass Microscope Slides: Hardware and software issues
Michael Becich, MD, PhD, Chairman and Professor of Biomedical Informatics and Pathology University of Pittsburgh School of Medicine

Dr. Becich discussed the benefits of WSI and the use of WSI, which has steadily increased since it’s first use in 2001, at Pittsburgh Medical School. He believes that the use of WSI will ultimately result in improved quality of care. He discussed several papers on the use of WSI, including one authored by Dr. Gilberston, and the lessons learned from these early “validation” studies. These early studies generally showed that pathologists reach the same diagnosis with WSI as they do with OM, despite artifacts, focus issues, etc. Based on these studies, he discussed study design considerations critical to WSI, including the physical characteristics of the hardware and software. He noted that WSI imaging does not replace the glass slide, pathologist should continue deferring in cases where more information is needed. Sample preparation for both WSI and light microscopy will be the same. In concluding he noted that (1) there are limitations to WSI, (2) light microscopy (due to practice variability) should not be used as the gold standard, and (3) that phantoms (non-human tissue) should be developed for use as reference standards.

Basics of Digital Imaging: Digital Image Capture and Display
Kevin L. Lorick, PhD, FDA/CDRH/OIVD/DIHD

Dr. Lorick discussed the technical aspects of digital imaging. For purposes of the discussion, he defined WSI as the creation of digital images from glass slides. He discussed limitations/variability related to the capture, storage and display of digital images. Concluding that there is a need for standardization.

Display of Digital Images: Hardware and software issues
Aldo Badano, PhD, Research Physicist FDA/CDRH/OSEL/DIAM

Dr. Badano gave a detailed overview on monitor characteristics. He noted that the imagining chain is as effective as its weakest link, which can be the display. He mainly discussed standards and guidelines in radiology, noting that a similar whole system is approached needed for WSI. He believes that the key parameters related to WSI are image size, reading environment and color. In concluding, he stated that standard characterization methods are needed to evaluate performance and facilitate FDA review.

Dr. Mello-Thoms asked how long we were away from standardization. Dr. Badano thought it would be at least 1-2 years.

Preclinical Bench Testing of Whole Slide Imaging Systems
Anant Agrawal, MSE Optical Engineer FDA/CDRH/OSEL/DP
Max Robinowitz, MD, Senior Medical Officer/Pathologist FDA/CDRH/OIVD/DIHD

NOTE: This presentation was moved to the afternoon, after the public speaker hearings.

Mr. Agrawal and Dr. Robinowitz presented two presentations on this topic. Their presentations focused on (1) physical testing of imaging performance characteristics and (2) pathologist-based assessment of image quality, respectively. The basic thesis was that some kind of ‘phantom’ (synthetic specimen) should be created to provide an objective measure of image quality and a standard of comparison between OM and WSI.

There was much discussion by the panel following Dr. Robinowitz’s presentation. They generally disagree with the TMA and counting approach. Dr. Robinowitz reiterated that FDA is looking for objectivity and that they are looking to the panel for answers.

In response to Mr. Agrawal’s presentation, Dr. Zhou also noted that a good image does not translate to diagnostic accuracy.

Day One Public Speakers

George Netto (Johns Hopkins), a colleague of Juan Rosai (Centro Diagnostico Italiano), read a letter from Dr. Rosai to the FDA panel praising digital pathology, including mention of Aperio’s Second Slide pilot in which he diagnosed cases referred to him by correspondents in Peru, Argentina, and Brazil.

Mahtab Fatemi, Longwell & Associates, read a statement regarding the need to use DICOM standards to evaluate WSI and that FDA should adopt the DICOM standard as a Class II Special Control FDA Guidance.

Dirk Soenksen (in his capacity as President of the Digital Pathology Association) gave a presentation on the applications and adoption of Digital Pathology.

Ole Eichhorn presented “Measuring Accuracy in Digital Pathology.”

Bob Dunstan (Biogen-Idec) talked about the importance of digital pathology for clinical studies.

Gene Cartright (CEO of Omnyx) gave some thoughts on digital pathology from an Omnyx perspective.

Lasko Fanand (University of Budapest and affiliated with 3D Histech) talked about studies they’ve conducted on use of WSI.

Panel members were invited to ask questions of any of the public speakers.

Day One Questions

Background: The FDA is asking the panel for advice and recommendations for the scientific and technical characterization of the physical characteristics of images produced by the conventional light microscope and digital whole slide imaging systems (WSI) ‐according to the requirements to maintain the present level of accuracy and precision expected or required for safe and effective routine surgical pathology.

Each of the questions about the optimal bench testing methods and studies needed for the objective analytical evaluation of the physical characteristics of OM and WSI may involve one or more of the following parts of the OM and WSI modalities:

  • The optical physical features (magnification, numerical aperture, working distance, xyz‐limiting resolution, wavelength band, field of view, depth of focus, stroke, focus resolution, ocular tilt range, optical axis offset, light source color spectrum, etc.)
  • The image acquisition process (the color and spatial resolution, sampling, bit‐depth, and sensitivity, as well as field of view, frame rate, shutter speed, quantum efficiency, gain control, output format, etc.)
  • Image processing (compression, tile stitching, smoothing, sharpening, etc.)
  • The display system (display type, size, image resolution, dynamic range, color calibration etc.)

FDA Question 1: What US or global standards, guideline, industry operating practices are available to guide the characterization of OM and WSI?

Panel Response: The panel was not aware of any. There was some philosophical discussion regarding whether OM is equivalent to WSI.

One of Dr. G’s key concerns (which he reiterated throughout the panel meeting) was surrounding things that could be missed and the pathologist not knowing. Detectability by the pathologist matters.

FDA Question 2: Are there objective methods for using physical phantoms that would be useful in the design, development and validation of WSI in comparison to the OM without the need for human reader interpretations, for example grids?

Panel Response: The panel agreed that it would be difficult to compare OM to WSI since OM is not standardized. The panel did not specify any objective methods. They stated that systems need some measure to verify characteristics and performance for maintenance and development. They agreed that phantoms could be used for verification but do not meet the need for validation/diagnostic process. Phantoms must recognize pitfalls and can be used as part of validation but do not replace slides.

FDA Question 3: Are there biological specimens (non‐human or human) that could be used to objectively challenge the performance of the human observer in a more controlled way compared to a clinical study? For example, invasion of cancer through basement membrane, fine chromatin details necessary for lineage of hematopoietic cells in tumors and inflammation exudates requiring at least 40X optical objective and Z‐dimension focusing, cytoplasmic fine details, etc.

Panel Response: The panel agreed that the human observer must be trained and familiar with WSI. Studies should include different pathologists, a spectrum of tissue types and spectrum of disease within one tissue type. They also noted that the emphasis should not be cancer but should also consider infectious disease. Overall challenging the human observers was not perceived as being useful.

FDA Question 4: What features of image quality (such as resolution, contrast, compression, etc.) are critical requirements and must be achieved for safe and effective use in routine surgical pathology?

Panel Response: After some discussion, Dr. Becker (FDA) clarified the question stating that the questions is asking about what features are necessary to evaluate use in routine surgical pathology. The panel agreed that a compression protocol must be created. Compression may vary based on tissue and specimen type. In addition to the features listed in the questions, the following should be considered: dynamic range, luminance (which may be more important than color, human factors/ease of navigation (keeping eyes on the monitor), imaging entire slide, fatigues.

FDA Question 5: Discuss the benefits and risks of one z‐plane of focus for routine surgical pathology.

Panel Response: There was little discussion regarding this question. It was noted that the key limitation occurs when the slide is out of focus. For surgical pathology no real risks exist when using one z-plane.

Day Two Overview

Day Two Introduction

Day two began much like day one, with Dr. Adcock giving an introduction, and Louise Magruder giving a compliance statement.

Day Two FDA Speakers

Overview and Recap of Previous Day

Tremel Faison mainly reviewed her own slides from Day 1 focusing on regulation of microscopes and why WSI differs. She also repeated the topics discussed the previous day. No detailed summaries were provided.

History and Lesson from FDA Regulation of Digital Radiology
Kyle Myers, PhD, Division Director FDA/CDRH/OSEL/DIAM

Dr. Myers presented an overview of FDA’s regulation of full-field digital mammography; the only digital radiography device requiring a PMA. Other digital radiography did not present new questions of safety and effectiveness, and therefore, required only 510(k)s.

Data requirements for the PMA included laboratory measurements, preclinical images (phantoms) and clinical data to evaluate diagnostic accuracy. Phantoms were used for premarket analysis and post-market QA. Paired image interpretations failed to demonstrate SE because of reader variability. Therefore, clinical trials are designed to look at both reader variability (multiple readers) and ranges of cases (multiple cases); MRMC (multiple-reader multiple case).

Now, 10 years later, FDA is moving towards down-classifying these devices to Class II with Special Controls.

Statistical Considerations in the Evaluation of Digital Pathology Devices
Shanti Gomatam, PhD, Mathematical Statistician FDA/CDRH/OSB/DBS

Dr. Gomatam provided an overview of the key statistical considerations and challenges with evaluating WSI. She noted at the beginning of her presentation that bias (accuracy) and variance (precision) are key. The presentation was very clear and concise and provided the panel with various options to consider for a clinical trial. She seemed open to the panels advice regarding what options would be best.

Options discussed were (1) the retrospective v prospective studies, (2) comparison of WSI to OM (where OM is the “control”)/other reference standards, (3) paired designs (with a washout period), (4) reader selection, (5) sample selection, (6) measurement selection, (7) accuracy, and (8) precision studies.

NOTE: Other industry members seemed pleased with Dr. Gomatam’s presentation. They felt that she presented a “black box” approach, although the final requirement may be closer to a “gray box” approach.

Evaluation of Reader Performance and Variability
Brandon Gallas, PhD, Mathematician FDA/CDRH/OSEL/DIAM

Dr. Gallas provided a more detailed overview of the challenges with reader performance. He reviewed (1) different measurement scales for data, (2) evaluation of clinical findings – sensitivity/specificity, (3) ROC curve (area under the ROC curve) from sensitivity/specificity to evaluate diagnostic performance (note: his own research focuses on ROC), (5) sources of variability – reader and cases (MRMC variance analysis), (5) study designs, (6) general truth and (7) biological variability/ROIs – reducing variability by studying one ROI.

He seemed partial to using the ROC curve; seemed to think that it is a good way to assess ambiguity between different modalities (ie reader and cases).

Regarding study design, he recommended a pilot study to determine appropriate numbers for a pivotal trial. Using mammography as an example, he noted that more than 5 readers and 44 cancers are required for a pivotal trial.

Several references are provided at the end of his presentation.

Clinical Trial Design for Digital Pathology Devices
Tan Nguyen, MD, PhD, RAC, Medical Officer/Pathologist FDA/CDRH/OIVD/DCTD

Dr. Nguyen was the last to present from FDA. He discussed both the quality of images and requirements for a clinical trial. With regard to quality of images he focused on image acquisition and image display, noting that perhaps the individual workstations may have to be controlled. He presented an additional slide (not included in the packet), which discussed the need for a QA protocol to verify the test system (cleanliness, focus, color, etc).
Regarding the clinical trial design, he seemed open to both prospective and retrospective clinical trials. He commented on the need for having a variety of cases, reducing observer variability, “reference” diagnosis and evaluating diagnosis agreement. He noted that there may be cases where there is a disagreement in opinion, however no “error” has occurred. The trial design needs to account for these situations.

NOTE: Dr. Nguyen approached me after the meeting showing interest in the DICOM Committee and saying that I should contact Maria Chen to see who at FDA could get involved with the Committee.

Comments from Panel regarding Clinical Trial Design

Dr. Gilbertson noted that deferral is NEVER a failure (this was echoed by other panel members). Key is to consider sample prep. Additionally two questions must be considered (1) Can pathologist evaluate good image? Then, (2) can WSI give you a good image?

Dr. Hewitt noted that ancillary studies must be considered. If the pathologist is uncertain, the next actions must noted on the CRF (ie IHC, new slide, new image, etc.)

Dr. Zhu liked the MRMC framework.

Dr. Sinard noted that use of an expert panel does not equal truth (general agreement here from panel). Sometimes the correct answer is that the image is not good. If comparing WSI to OM, then the spread of diagnosis must be considered. WSI should provide the same spread.

Enriched samples with low-prevalence, challenges pathologist but not WSI.

Day Two Public Speakers

Andy Evans, MD, Pathologist at University Health Network, Toronto, Canada
Dr. Evans provided an overview of his network’s use of WSI and due diligence activities (ie validation, malpractice insurance, etc) they conducted prior to using WSI. As part of the due diligence, they also consulted with Health Canada (Therapeutic Products Programs). HC noted that telepathology does not involve a “medical device” because there is no patient contact.

Panel Member comments:
After being questioned regarding the use of technicians, Dr. Evans noted that the technicians are crucial along with the quality of the histology. They conduct QA every morning to address the quality of image and connectivity.

Dr. Davey noted that adjunctive use should be separated form primary use.

Dr. Wallenstein, College of American Pathologists (CAP)
Dr. Wallenstein spoke on behalf of CAP. He noted that CAP is agnostic. Ultimately they want an equal or better device. CAP believes that the validation protocol should focus on intra-operator variability and not accuracy of the diagnoses.

Panel Member comments:
Dr Mello-Thomas has issues with intra-operator variability stating that other factors like sleep, coffee, could affect the results. She also thought that it would increase the n. CAP commented that since the study would be randomized, the effect should be the same for OM and WSI. Other panel members (Dr. G and Dr. K) seemed to agree with the CAP approach.

Dr. Monroe, CMO, BioImagene (NOTE: no handouts were available for this presentation)
Dr. Monroe presented both his position and Dr. Singh’s, CEO of BioImagene who was not able to speak at the meeting.

Dr. Monroe is a pathologist, specializing in AP, clinical pathology and cytophatology. He is a practicing pathologist, focusing on rural areas in California. He discussed the benefits of WSI as it relates to working in rural area. Also mentioned that he believes WSI should be considered a pre-Amendment device based on a use in 1967.

Dr. Singh was formerly with Siemens and further commented on the benefits of WSI. Also noted that additional use cases, such as case retrievals should not have the same regulatory requirements.

Aurora Interactive, Montreal, Canada
Dr. Yo, hematologist at the University of Torronto and P. LeFavre, CEO of Aurora, spoke on behalf of Aurora Interactive. His talk focused on the benefits of WSI (similar to those already mentioned).

Mr. LeFavre discussed technological characteristics in great length. He discussed compression and conversion. Regarding compressions he noted that lossy and lossless compression should be allowed depending on the use of the image (visual v analytical). Conversion, however, he stated should not be allowed. He also discussed the different parts of a WSI system (transmission v image conversion) noting that each part should be evaluated separately and that transmission should not be regulated by FDA.

Day Two Questions

FDA Question 1: FDA clearance or approval requires that digital mammography devices for screening and diagnostic intended uses must use only non‐lossy compression. Many current digital WSI prototype devices employ some type of digital lossy compression. Please discuss the safety and effectiveness of lossy compression applied to diagnostic digital WSI and the trade‐off between image quality and decreasing the digital file size.

Panel Response: The panel did not discuss this question in great detail. They noted that there is a need to address issues of dynamic range and monitors first. Also offered other alternatives such as non-lossy followed by decompression.

FDA Question 2: Discuss the benefits and risks of allowing WSI for all organ systems based on the evidence from performance characteristics in a single or a limited number of organ systems?

Panel Response: The panel generally agreed that it is unrealistic to do an organ/disease based approach; rather features should be evaluated. They proposed using a fit-for-purpose model. Dr. Zhou was the only panel member who felt that studying the disease is better because it allows for easier evaluation of accuracy.

FDA Question 3: Regarding clinical validation of WSI, please comment on the following components of a clinical study:

Question 3.1: Reference standard for diagnostic truth

  • Composition of panel
  • Number of panel members
  • Rules for agreement

Panel Response: There was disagreement among the panel members regarding the benefit of an expert panel given that the variability in diagnosis is unknown. Dr. Mello‐Thoms also reiterated her concern with only looking at intra-operator variability because of variables such as sleep, coffee, etc. All agreed, however, that diagnosis and the quality of images are important components of any study.

Washout (WO) period was also discussed. Specifically it was mentioned that a long WO (such as 5 years) may not be ideal because the user experience will have changed over that time.

Question 3.2: Type of clinical studies – with respect to

  • Prospective/retrospective
  • Enrichment
  • Types of specimen samples to be included
  • Total area of the whole slide or demarcated areas to be included?
  • Length of washout required
  • Selection mechanism

Panel Response:
Prospective/retrospective: The panel was open to both approaches. A retrospective study may make more sense if we assume that the study will not be organ/disease based. A retrospective study, however, must be representative of day to day activities/samples and samples could be recut. For a prospective study slides would have to be cut to specifications. From a statistical perspective (Dr. Zhou), a cross-over design and prospective study would be preferred.

Enrichment will be needed for “rare” cases.

Types of specimen samples to be included – already discussed in Q 3.1.

Total area of the whole slide or demarcated areas to be included?
For diagnostic accuracy, the whole case and whole slide must be presented.

Length of washout required: may not need to be long depending on specimen/rarity of specimen. Changing the orientation of the slides on WSI may be one way to create ‘confusion.’

NOTE: There was a short break, after which it was noted that further comments could be submitted by *anyone* to Maria Chan (maria.chan@fda.hhs.gov).

Question 3.3: Minimization of reader variability

Would the multi‐reader, multi‐case (MRMC) paradigm frequently used to account for reader variability in digital mammography be applicable in accounting for reader variability in the validation of diagnostic pathology?

Panel Response: Discussed in Q3.1; panel generally agreed a “fully crossed” study design was best.

Question 3.4: Selection of study pathologists, number of study participants

Panel Response: This should be based on statistical considerations; introduce as much variability as possible.

Question 3.5: Performance assessment

  • Should the study pathologist be instructed to find the most clinically significant lesions or all deviations from normal?
  • On what measurements would acceptable level of performance be required (primary measurements)?
  • What additional measurements should be included in the evaluation (secondary)?
  • What hypotheses are appropriate on primary and secondary measurements?

Panel Response: The panel agreed that replicating what is on the surgical pathology report is important. They identified several factors that must be considered:

  • deferral
  • bad images – deferral because of bad image versus diagnosis despite bad image. Both of these scenarios need to be reported.
  • final diagnosis must still be considered even with deferral.

FDA Question 4: Discuss factors that would be important for inclusion in precision studies.

Panel Response: Diagnostic precision was addressed by the intra-operator variability discussed in Q3.1.

For the device precision, the panel agreed that the following should be considered for a precision study.

  • tissue should not be scanned
  • focus
  • phantoms, which include range of specimen (discussed earlier), could be used
  • slide types
  • technical staff scanning the slides

FDA Question 5: Discuss what training should be required for study pathologist evaluating WSI and for pathologists in the post-market use of WSI.

Panel Response: For study pathologists evaluating WSI, some comfort and familiarity must be there before the study starts. This can be done by showing pathologists a series of easy slides to familiarize them with the new technique. Methods used can include self-training but must also address technical staff training/preanalytical aspect. There was consensus that no certification should be required. Similar training should be applied post-market.

FDA Question 6: Are there safety and effectiveness issues in the validation of WSI that cannot be addressed by premarketing studies? Is there a need for a post‐marketing study to gather additional safety and effectiveness data? If so, how should such a study be conducted?

Panel Response: There was consensus that there is no implicit need for the manufacturer to conduct a post-market study unless a specific question is brought forth in the clinical trials. There would be a need for continued evaluation of WSI, however, this would be done through peer literature.

Closing

Maria Chan (FDA IVD Director) thanked the panel for their great participation and input. Dr. Becker gave a closing summation.

Other Interesting Notes

Radiology Device panel meeting to be held in November to discuss reclassifying full-field digital mammography devices to Class II with Special Controls.