A Sparse Deep Learning Approach for Automatic Segmentation of Human Vasculature in Multispectral Optoacoustic Tomography

Abstract

Multispectral Optoacoustic Tomography (MSOT) resolves oxy-(HbO2) and deoxy-hemoglobin (Hb) to perform vascular imaging. MSOT suffers from gradual signal attenuation with depth due to light-tissue interactions: an effect that hinders the precise manual segmentation of vessels. Furthermore, vascular assessment requires functional tests, which last several minutes and result in recording thousands of images. Here, we introduce a deep learning approach with a sparse UNET (S-UNET) for automatic vascular segmentation in MSOT images to avoid the rigorous and time-consuming manual segmentation. We evaluated the S-UNET on a test-set of 33 images, achieving a median DICE score of 0.88. Apart from high segmentation performance, our method based its decision on two wavelengths with physical meaning for the task-at-hand: 850 nm (peak absorption of oxy-hemoglobin) and 810 nm (isosbestic point of oxy-and deoxy-hemoglobin). Thus, our approach achieves precise data-driven vascular segmentation for automated vascular assessment and may boost MSOT further towards its clinical translation.

Clinical noninvasive imaging and spectroscopic tools for dermatological applications: Review of recent progress

“…Furthermore, in quantifying the features of vascular modeling seen in skin vascular diseases, the vasculature network in the images have to be manually segmented which is tedious and time-consuming, especially in deeper depths and large functional data sets. Chlis et al reported a specialized deep learning method, the sparse UNET (S-UNET) which is the leading deep network in image segmentation, to automatically segment vascular networks in MSOT images to address this challenge [48]. This method was able to segment both arteries and veins from unconstructed MSOT acquisitions precisely, even in signal attenuated depths from images of two wavelengths: 850 nm (maximum absorption of HbO 2 ) and 810 nm (isosbestic point of HbO 2 and Hb).…”

Laser speckle contrast imaging and machine learning in application to physiological fluids flow rate recognition

“…The blood vessel visualization using multispectral optoacoustic tomography is considered in Ref. [11]. As an alternative to manual segmentation of the obtained images, the authors propose the use of "sparse-UNET".…”