Abstract 2332: Enhanced analysis of CO-Detection by indEXing (CODEX) imaging through a novel image processing technique utilizing two-dimensional Fourier transform for stitching artifact detection

Abstract

Abstract Objective: The objective of this study is to address the challenge of stitching artifacts in CODEX imaging, which hinders the accurate spatial quantification of positive cell locations—a critical step for reliable spatial analysis in computational pathology. Methods: Our algorithm applies 2D-FT and inverse transformations to process pathological images. We tested the algorithm on 16 different histology images, each containing 5 or more regions. Within these regions, we evaluated 29 protein markers, culminating in the analysis of over 2800 images. Results: The algorithm demonstrated a recall rate of over 85 percent, indicating a high likelihood of accurately identifying true positive regions with stitching artifacts. However, the precision varied, suggesting further investigation is needed to understand the variability and improve consistency. Conclusion: The proposed method significantly improves CODEX imaging reliability by detecting and quantifying stitching artifacts. The method's high recall rates are promising, but the variable precision indicates the necessity for ongoing refinement to ensure consistent diagnostic accuracy across diverse and extensive image sets. Keywords: Computational pathology, Fourier transform, Image stitching, CODEX, Spatial analysis, Diagnostic precision, Recall. Citation Format: Afrooz Jahedi, Aliya Khan, Kasthuri Kannan, Krishna Bhat. Enhanced analysis of CO-Detection by indEXing (CODEX) imaging through a novel image processing technique utilizing two-dimensional Fourier transform for stitching artifact detection [abstract]. In: Proceedings of the American Association for Cancer Research Annual Meeting 2024; Part 1 (Regular Abstracts); 2024 Apr 5-10; San Diego, CA. Philadelphia (PA): AACR; Cancer Res 2024;84(6_Suppl):Abstract nr 2332.