24 July 2015 Meeting Minutes
A teleconference was held on 24 July 2015 to review the survey answers provided in response to the Histopathology Visualization Project Questionnaire (see below). A request was made for additional demonstrations of visual displays and software tools being used by members of the team. The following are notes taken during a discussion of each question of the survey.
A1) Common themes include organ weight analysis, cross study comparisons, aggregating data from multiple studies. Data from multiple studies have been aggregated together by sponsors to answer/address health authority questions. Examples include historical control or background lesion incidence. These data have been presented in tabular formats, but not in graphical displays. Desire expressed for producing graphics of historical data.
A2) Common themes include evaluation of dose-related trends, pattern recognition/trend detection, quick visual overviews of findings, cross-study comparisons, correlation to biomarker data, communication of findings in meetings. Graphical displays could benefit non-pathologists who are less familiar with pathology data.
B1) A list of various software tools is avaialbel, a request was made for team members to share case examples of graphical displays produced with the software tool used by their organization.
B2) Histopath data from both single and multiple studies are being analyzed.
B3) There are various software challenges. Challenges with user interface, functionality and user familiarity with the software tools and options. Problem in part due to software engineers creating software that are at odds with actual user desires and lack of intuitive usability. Instructions, user interface and design can be complex.
C1) FDA requires that software displays be in compliance with Americans with Disabilities Act (see Section 508) and require that the graphics/visual displays can be interpreted by color blind individuals. This requires grey-scale for visual cues. Color-coded cues need to be convertible to grey-scale. Histograms and heat maps were mentioned. Software from INDS can produce a heat map. PointCross produces a heat map for clinical lab data.
C2) Common challenges include terminology differences, qualitative differences between labs and personel reviewing studies. Histopathology data are qualitative in nature, not numeric.
C3) Numerical data are key. Need to ask specific questions. There was a lot of interest with correlating histopath data with exposure (TK) data, creating graphs of histopath incidence data versus TK data (x-y plots). Histopath and exposure data across studies could be combined or aggregated. Large animal path data could be graphed versus individual animal exposures.