A distance-based loss for smooth and continuous skin layer segmentation in optoacoustic images

Raster-scan optoacoustic mesoscopy (RSOM) is a powerful, non-invasive optical imaging technique for functional, anatomical, and molecular skin and tissue analysis. However, both the manual and the automated analysis of such images are challenging, because the RSOM images have very low contrast, poor signal to noise ratio, and systematic overlaps between the absorption spectra of melanin and hemoglobin. Nonetheless, the segmentation of the epidermis layer is a crucial step for many downstream medical and diagnostic tasks, such as vessel segmentation or monitoring of cancer progression. We propose a novel, shape-specific loss function that overcomes discontinuous segmentations and achieves smooth segmentation surfaces while preserving the same volumetric Dice and IoU. Further, we validate our epidermis segmentation through the sensitivity of vessel segmentation. We found a 20\(\%\) improvement in Dice for vessel segmentation tasks when the epidermis mask is provided as additional information to the vessel segmentation network.

[1]  H. Kittler,et al.  Diagnostic accuracy of dermoscopy. , 2002, The Lancet. Oncology.

[2]  Xiaodong Wu,et al.  Optimal Surface Segmentation in Volumetric Images-A Graph-Theoretic Approach , 2006, IEEE Transactions on Pattern Analysis and Machine Intelligence.

[3]  A. Ormerod,et al.  Systematic review of dermoscopy and digital dermoscopy/ artificial intelligence for the diagnosis of melanoma , 2009, The British journal of dermatology.

[4]  V. Ntziachristos,et al.  Molecular imaging by means of multispectral optoacoustic tomography (MSOT). , 2010, Chemical reviews.

[5]  M. Arif Gok,et al.  Determination of Surface Qualities on Inclined Surface Machining with Acoustic Sound Pressure , 2012 .

[6]  Gábor Székely,et al.  Tissue metabolism driven arterial tree generation , 2012, Medical Image Anal..

[7]  B. Cyganek.  Tensor Methods in Computer Vision , 2013 .

[8]  Vasilis Ntziachristos,et al.  Broadband mesoscopic optoacoustic tomography reveals skin layers. , 2014, Optics letters.

[9]  Thomas Brox,et al.  U-Net: Convolutional Networks for Biomedical Image Segmentation , 2015, MICCAI.

[10]  Gábor Székely,et al.  Joint 3-D vessel segmentation and centerline extraction using oblique Hough forests with steerable filters , 2015, Medical Image Anal..

[11]  Vasilis Ntziachristos,et al.  Pushing the Optical Imaging Limits of Cancer with Multi-Frequency-Band Raster-Scan Optoacoustic Mesoscopy (RSOM) , 2015, Neoplasia.

[12]  Costantino Grana,et al.  Skin Surface Reconstruction and 3D Vessels Segmentation in Speckle Variance Optical Coherence Tomography , 2016, VISIGRAPP.

[13]  Wojciech M. Kwiatek,et al.  Effect of Magnetite Composite on the Amount of Double Strand Breaks Induced with X-Rays , 2016 .

[14]  Ghassan Hamarneh,et al.  Topology Aware Fully Convolutional Networks for Histology Gland Segmentation , 2016, MICCAI.

[15]  Hariharan Ravishankar,et al.  Learning and Incorporating Shape Models for Semantic Segmentation , 2017, MICCAI.

[16]  Subhransu Maji,et al.  3D Shape Segmentation with Projective Convolutional Networks , 2016, 2017 IEEE Conference on Computer Vision and Pattern Recognition (CVPR).

[17]  S. M. Masudur Rahman Al-Arif,et al.  Shape-Aware Deep Convolutional Neural Network for Vertebrae Segmentation , 2017, MSKI@MICCAI.

[18]  V. Ntziachristos,et al.  Precision assessment of label-free psoriasis biomarkers with ultra-broadband optoacoustic mesoscopy , 2017, Nature Biomedical Engineering.

[19]  Qi Wu,et al.  Skin Lesion Classification in Dermoscopy Images Using Synergic Deep Learning , 2018, MICCAI.

[20]  Jin U. Kang,et al.  Towards a Fast and Safe LED-Based Photoacoustic Imaging Using Deep Convolutional Neural Network , 2018, MICCAI.

[21]  Jun Cheng,et al.  Three-dimensional graph-based skin layer segmentation in optical coherence tomography images for roughness estimation. , 2018, Biomedical optics express.

[22]  John Willian Branch,et al.  Automatic skin lesion segmentation on dermoscopic images by the means of superpixel merging , 2018, MICCAI.

[23]  Vasilis Ntziachristos,et al.  Optoacoustic mesoscopy for biomedicine , 2019, Nature Biomedical Engineering.

[24]  Bjoern H Menze,et al.  Shape-Aware Complementary-Task Learning for Multi-Organ Segmentation , 2019, MLMI@MICCAI.

[25]  Dimitris Samaras,et al.  Topology-Preserving Deep Image Segmentation , 2019, NeurIPS.

[26]  Vasilis Ntziachristos,et al.  Fully Automated Identification of Skin Morphology in Raster-Scan Optoacoustic Mesoscopy using Artificial Intelligence. , 2019, Medical physics.

[27]  Bjoern H. Menze,et al.  Transfer learning from synthetic data reduces need for labels to segment brain vasculature and neural pathways in 3D , 2019 .

[28]  Bjoern H. Menze,et al.  DeepVesselNet: Vessel Segmentation, Centerline Prediction, and Bifurcation Detection in 3-D Angiographic Volumes , 2018, Frontiers in Neuroscience.

[29]  Bjoern H Menze,et al.  Machine learning analysis of whole mouse brain vasculature , 2020, Nature Methods.

[30]  clDice - a Topology-Preserving Loss Function for Tubular Structure Segmentation , 2020, ArXiv.

[31]  Nassir Navab,et al.  Skin Surface Detection in 3D Optoacoustic Mesoscopy Based on Dynamic Programming , 2020, IEEE Transactions on Medical Imaging.