14 August 2015 Meeting Minutes

From PHUSE Wiki
Jump to: navigation, search

A teleconference and webex was held on 14 August 2015. Karen Porter from PointCross displayed a demonstration heatmap of clinical laboratory data (see attachement below) as an possibile graphical display for histopathology data. The color green indicates correlation to data whereas red indicates lack of a correlation. A heat map requires numeric data (such as fold-change versus control), such as change from control. It may be possible to map select clinical laboratory data agains histopathology data, such as specific organ findings. For example, liver enzymes, liver findings or organ weights could be mapped against one another. The x-axis could be dose level for a select organ and the y-axis fold-changes against control for lab test or organ weight data. A heat map could be used to aggregate data from multiple studies, or for displaying and presenting complex data to an audience. A heat map would not be efficient for primary/initial (first tier) data analysis but could be used for secondary (second tier) analysis. Following presentation of the heat map, the group discussed an initial draft summary of responses to the histopathology visualization questionnaire (see attached below). Comments regarding A1: a major use of graphical displays is to present complex data to project teams and groups. Visualization tools are not the primary means for data analysis, but could be of value for aggregating data such as historical control data. Comments for A2: the majority of responses indicated that visualization/graphical displays are not the primary means for data analysis. Problem could be that current tools are not effective. Because of the subjective nature of pathology, statistical analysis of data is challenging, especially for comparing data across studies. The group proposed the following conclusion: visual displays of histopathology data constitute a second tier of data interpretation and analysis. Hypothesis testing should be focused. Value is seen with investigating correlations across different domains and for aggregating data across multiple studies.

PointCross ToxVision Sample Heatmap - Clinical Lab Data demo

First draft of histopath visualization survey summary - bullet points