Visual parameter optimisation for biomedical image processing.

Aj Pretorius,Y Zhou,Ra Ruddle

doi:10.1186/1471-2105-16-s11-s9

Aj Pretorius, Y Zhou + Show 1 more

Open Access

https://doi.org/10.1186/1471-2105-16-s11-s9

Copy DOI

Abstract

BackgroundBiomedical image processing methods require users to optimise input parameters to ensure high-quality output. This presents two challenges. First, it is difficult to optimise multiple input parameters for multiple input images. Second, it is difficult to achieve an understanding of underlying algorithms, in particular, relationships between input and output.ResultsWe present a visualisation method that transforms users' ability to understand algorithm behaviour by integrating input and output, and by supporting exploration of their relationships. We discuss its application to a colour deconvolution technique for stained histology images and show how it enabled a domain expert to identify suitable parameter values for the deconvolution of two types of images, and metrics to quantify deconvolution performance. It also enabled a breakthrough in understanding by invalidating an underlying assumption about the algorithm.ConclusionsThe visualisation method presented here provides analysis capability for multiple inputs and outputs in biomedical image processing that is not supported by previous analysis software. The analysis supported by our method is not feasible with conventional trial-and-error approaches.

Highlights

Biomedical image processing methods require users to optimise input parameters to ensure highquality output
In further sections we describe a novel visualisation method to address the challenges and discuss a case study where our approach was used
Challenges We describe two unaddressed challenges, identified by analysing the above case study, a review of related work, and discussions with domain experts (Broad Institute, Leeds, and TU Darmstadt)

Summary

Introduction

Biomedical image processing methods require users to optimise input parameters to ensure highquality output. Biomedical image processing is fundamental to many biological research methods [1] These algorithms take parameter values and images as input, and produce annotated images and quantitative measures as output. Because they are sensitive to parameter values, imaging artefacts, and factors like tissue type, it is difficult to find robust parameter values that ensure high-quality output. Users want to review image-based output for up to five images We obtained these numbers by consulting domain experts and by observing users.

Methods

Results

Conclusion