340 Machine Learning Segmentation of Amyloid Load in Ligamentum Flavum Specimens From Spinal Stenosis Patients. (April 2022)

• Record Type: Journal Article
• Authors: Wang, Andy Y.; Sharma, Vaishnavi; Saini, Harleen; Tingen, Joseph N.; Flores, Alexandra; Liu, Diang; Safain, Mina G.; Kryzanski, James; McPhail, Ellen D.; Arkun, Knarik; Riesenburger, Ron I.
• Is Part Of: Journal of clinical and translational science. Volume 6(2022)Supplement 1
• Page Start: 63
• Page End: 63
• Publication Date: 2022-04
• Subjects: Clinical medicine -- Research -- Periodicals; Medicine, Experimental -- Periodicals; Human experimentation in medicine -- Periodicals; 616.027
• DOI: 10.1017/cts.2022.193
• Languages: English
• ISSNs: 2059-8661

Abstract

Abstract : OBJECTIVES/GOALS: Wild-type transthyretin amyloid (ATTRwt) deposits have been found to deposit in the ligamentum flavum (LF) of spinal stenosis patients prior to systemic and cardiac amyloidosis, and is implicated in LF hypertrophy. Currently, no precise method of quantifying amyloid deposits exists. Here, we present our machine learning quantification method. METHODS/STUDY POPULATION: Images of ligamentum flavum specimens stained with Congo red are obtained from spinal stenosis patients undergoing laminectomies and confirmed to be positive for ATTRwt. Amyloid deposits in these specimens are classified and quantified by TWS through training the algorithm via user-directed annotations on images of LF. TWS can also be automated through exposure to a set of training images with user- directed annotations, and then application to a set of new images without additional annotations. Additional methods of color thresholding and manual segmentation are also used on these images for comparison to TWS. RESULTS/ANTICIPATED RESULTS: We develop the use of TWS in images of LF and demonstrate its potential for automated quantification. TWS is strongly correlated with manual segmentation in the training set of images with user-directed annotations (R = 0.98; p = 0.0033) as well as in the application set of images where TWS was automated (R = 0.94; p = 0.016). Color thresholding was weakly correlated with manual segmentation in the training set of images (R = 0.78; p = 0.12) and in the application set of images (R = 0.65; p = 0.23). DISCUSSION/SIGNIFICANCE: Our machine learning method correlates with the gold standard comparator of manual segmentation and outperforms color thresholding. This novel machine learning quantification method is a precise, objective, accessible, high throughput, and powerful tool that will hopefully pave the way towards future research and clinical applications.