Comparison of Normalization Methods to Combine Batches of High-Dimensional Multiplexed Images

• Activity Number: 121 - In the Pipeline: Statistical Advances to Preserve Biological Signal in High-Throughput, Single-Cell Imaging and Sequencing Methods
• Type: Topic-Contributed
• Date/Time: Monday, August 9, 2021 : 1:30 PM to 3:20 PM
• Sponsor: Section on Statistics in Imaging
• Author(s): Coleman R Harris* and Qi Liu and Eliot McKinley and Joseph Roland and Ken Lau and Robert Coffey and Simon N Vandekar
• Companies: Vanderbilt University Medical Center and Vanderbilt University Medical Center and Vanderbilt University Medical Center and Vanderbilt University Medical Center and Vanderbilt University Medical Center and Vanderbilt University Medical Center and Vanderbilt University
• Keywords: multiplexed; imaging; normalization; harmonization; biological signal; batch effect

Abstract

High-dimensional multiplexed imaging methods can help quantify the heterogeneity of cell populations in healthy and tumorous tissues, offering insight into tumor progression and improved treatment strategies. However, implicit biases exist in the imaging pipeline - images are distorted by optical effects, slide and batch effects, and instrument variability. Normalization of this data is compounded by the number of markers and natural tissue variability within each image, introducing systematic differences that impacts inference. In this work, we introduce an image normalization pipeline to reduce systematic variability in multiplexed images. We build on existing methods, namely the ComBat algorithm and other normalization techniques like histogram matching and registration to remove slide and image effects, reducing variability in multiplexed images. We then compare these methods by analyzing multiplexed immunofluorescence (MxIF) images to quantify overall reduction in the variability of the data, and further compare each method's ability to retain biological signal by evaluating clustering accuracy for specific biological pathways.