Saak Transform-Based Machine Learning for Light-Sheet Imaging of Cardiac Trabeculation

Abstract

Objective: Recent advances in light-sheet fluorescence microscopy (LSFM) enable 3dimensional (3-D) imaging of cardiac architecture and mechanics in toto. However, segmentation of the cardiac trabecular network to quantify cardiac injury remains a challenge. Methods:We hereby employed "subspace approximation with augmented kernels (Saak) transform" for accurate and efficient quantification of the light-sheet image stacks following chemotherapy-treatment. We established a machine learning framework with augmented kernels based on the Karhunen-Loeve Transform (KLT) to preserve linearity and reversibility of rectification. Results:The Saak transform-based machine learning enhances computational efficiency and obviates iterative optimization of cost function needed for neural networks, minimizing the number of training datasets for segmentation in our scenario. The integration of forward and inverse Saak transforms can also serve as a light-weight module to filter adversarial perturbations and reconstruct estimated images, salvaging robustness of existing classification methods. The accuracy and robustness of the Saak transform are evident following the tests of dice similarity coefficients and various adversary perturbation algorithms, respectively. The addition of edge detection further allows for quantifying the surface area to volume ratio (SVR) of the myocardium in response to chemotherapy-induced cardiac remodeling. Conclusion:The combination of Saak transform, random forest, and edge detection augments segmentation efficiency by 20-fold as compared to manual processing.Significance: This new methodology establishes a robust framework for post light-sheet imaging processing, and creating a data-driven machine learning for automated quantification of cardiac ultra-structure.

Virtual Reality for Interactive Medical Analysis

“…We desired to segment multiple structures in a rapid time frame, which led to our use of manual segmentation. We intend to move forward with an efficient and robust image segmentation method for the retention of detail in the future (Ding et al, 2021).…”

Tissue clearing and imaging methods for cardiovascular development

“…Advances in light sheet imaging of the cardiac system and its analysis are summarized in the review by Ding et al (2018). In the last 2 years segmentation analysis made progress in applying Deep learning image analysis on LSCM-generated images (Ding et al, 2021). This technique was developed and applied to obtain information about volumetric changes and cardiac function of zebrafish (Akerberg et al, 2019) and cardiac architecture and mechanics (Ding et al, 2021).…”

“…In the last 2 years segmentation analysis made progress in applying Deep learning image analysis on LSCM-generated images (Ding et al, 2021). This technique was developed and applied to obtain information about volumetric changes and cardiac function of zebrafish (Akerberg et al, 2019) and cardiac architecture and mechanics (Ding et al, 2021). Deep learning was also successfully used on segmentation of brain microvasculature and analysis of vessels density in various brain areas (Todorov et al, 2020) (Veith and B Baker, 2020).…”

Deep Neural Network-Aided Histopathological Analysis of Myocardial Injury

“…These computational models are based on algorithms that can extract features from clinical data (12). Compared to traditional machine learning methods that rely on expert knowledge to transform raw image data into features (e.g., texture, statistics, and wavelet transform coefficients) (13,14), deep neural networks (DNN) can achieve better accuracy without defining features explicitly. In the field of cardiovascular diseases, deep learning has been widely implemented for image classification and segmentation in multiple modalities, including echocardiography, coronary artery calcium scoring, coronary computed tomography angiography, single-photon emission computed tomography, positron emission tomography, magnetic resonance imaging, and optical coherence tomography (8,(15)(16)(17).…”