Burn wound classification model using spatial frequency-domain imaging and machine learning

Abstract. Accurate assessment of burn severity is critical for wound care and the course of treatment. Delays in classification translate to delays in burn management, increasing the risk of scarring and infection. To this end, numerous imaging techniques have been used to examine tissue properties to infer burn severity. Spatial frequency-domain imaging (SFDI) has also been used to characterize burns based on the relationships between histologic observations and changes in tissue properties. Recently, machine learning has been used to classify burns by combining optical features from multispectral or hyperspectral imaging. Rather than employ models of light propagation to deduce tissue optical properties, we investigated the feasibility of using SFDI reflectance data at multiple spatial frequencies, with a support vector machine (SVM) classifier, to predict severity in a porcine model of graded burns. Calibrated reflectance images were collected using SFDI at eight wavelengths (471 to 851 nm) and five spatial frequencies (0 to 0.2  mm  −  1). Three models were built from subsets of this initial dataset. The first subset included data taken at all wavelengths with the planar (0  mm  −  1) spatial frequency, the second comprised data at all wavelengths and spatial frequencies, and the third used all collected data at values relative to unburned tissue. These data subsets were used to train and test cubic SVM models, and compared against burn status 28 days after injury. Model accuracy was established through leave-one-out cross-validation testing. The model based on images obtained at all wavelengths and spatial frequencies predicted burn severity at 24 h with 92.5% accuracy. The model composed of all values relative to unburned skin was 94.4% accurate. By comparison, the model that employed only planar illumination was 88.8% accurate. This investigation suggests that the combination of SFDI with machine learning has potential for accurately predicting burn severity.

[1]  Bernard Choi,et al.  Quantitative long‐term measurements of burns in a rat model using Spatial Frequency Domain Imaging (SFDI) and Laser Speckle Imaging (LSI) , 2017, Lasers in surgery and medicine.

[2]  Gaël Varoquaux,et al.  Cross-validation failure: Small sample sizes lead to large error bars , 2017, NeuroImage.

[3]  Jeffrey W. Shupp,et al.  Critical Review of Burn Depth Assessment Techniques: Part I. Historical Review , 2009, Journal of burn care & research : official publication of the American Burn Association.

[4]  Bernard Choi,et al.  Utility of spatial frequency domain imaging (SFDI) and laser speckle imaging (LSI) to non-invasively diagnose burn depth in a porcine model. , 2015, Burns : journal of the International Society for Burn Injuries.

[5]  Juan Heredia Juesas,et al.  Non-invasive optical imaging techniques for burn-injured tissue detection for debridement surgery , 2016, 2016 38th Annual International Conference of the IEEE Engineering in Medicine and Biology Society (EMBC).

[6]  Asgeir Bjorgan,et al.  Can spectral–spatial image segmentation be used to discriminate experimental burn wounds? , 2016, Journal of biomedical optics.

[7]  Sylvain Gioux,et al.  Machine learning approach for rapid and accurate estimation of optical properties using spatial frequency domain imaging , 2018, Journal of biomedical optics.

[8]  C. White,et al.  Advances in surgical care: Management of severe burn injury , 2008, Critical care medicine.

[9]  William Stafford Noble,et al.  Support vector machine , 2013 .

[10]  Zheng Rong Yang,et al.  Biological applications of support vector machines , 2004, Briefings Bioinform..

[11]  Louise Poissant Part I , 1996, Leonardo.

[12]  Jeffrey E. Thatcher,et al.  Outlier detection and removal improves accuracy of machine learning approach to multispectral burn diagnostic imaging , 2015, Journal of biomedical optics.

[13]  Qing Zeng-Treitler,et al.  Predicting sample size required for classification performance , 2012, BMC Medical Informatics and Decision Making.

[14]  R. Matthiesen Bioinformatics Methods in Clinical Research , 2010, Methods in Molecular Biology.

[15]  Frederick P Rivara,et al.  What is the prevalence of hypertrophic scarring following burns? , 2003, Burns : journal of the International Society for Burn Injuries.

[16]  Lorenzo Leonardi,et al.  Classification of burn injuries using near-infrared spectroscopy. , 2006, Journal of biomedical optics.

[17]  Phillip Blondeel,et al.  Assessment of burn depth and burn wound healing potential. , 2008, Burns : journal of the International Society for Burn Injuries.

[18]  Browne,et al.  Cross-Validation Methods. , 2000, Journal of mathematical psychology.

[19]  Bernard Choi,et al.  Spatial frequency domain imaging of burn wounds in a preclinical model of graded burn severity , 2013, Journal of biomedical optics.

[20]  Z. Janžekovič,et al.  A new concept in the early excision and immediate grafting of burns. , 1970, The Journal of trauma.

[21]  Pedro Larrañaga,et al.  Machine learning: an indispensable tool in bioinformatics. , 2010, Methods in molecular biology.

[22]  Anthony J. Durkin,et al.  Evaluating clinical observation versus Spatial Frequency Domain Imaging (SFDI), Laser Speckle Imaging (LSI) and thermal imaging for the assessment of burn depth. , 2019, Burns : journal of the International Society for Burn Injuries.

[23]  Jeffrey E. Thatcher,et al.  Burn-injured tissue detection for debridement surgery through the combination of non-invasive optical imaging techniques. , 2018, Biomedical optics express.

[24]  Corinna Cortes,et al.  Support-Vector Networks , 1995, Machine Learning.

[25]  Bernard Choi,et al.  Quantitative assessment of graded burn wounds in a porcine model using spatial frequency domain imaging (SFDI) and laser speckle imaging (LSI). , 2014, Biomedical optics express.

[26]  Charles A DiMarzio,et al.  Multispectral method for skin imaging: development and validation. , 2007, Applied optics.

[27]  Jelle J Goeman,et al.  Efficient approximate k‐fold and leave‐one‐out cross‐validation for ridge regression , 2013, Biometrical journal. Biometrische Zeitschrift.

[28]  Bruce J. Tromberg,et al.  Quantitative In Vivo Imaging of Tissue Absorption, Scattering, and Hemoglobin Concentration in Rat Cortex Using Spatially Modulated Structured Light , 2009 .

[29]  Yanyu Zhao,et al.  Deep learning model for ultrafast multifrequency optical property extractions for spatial frequency domain imaging. , 2018, Optics letters.

[30]  David J. Cuccia,et al.  Spatial frequency domain imaging (SFDI): a technology overview and validation of an LED-based clinic friendly device , 2012, Other Conferences.

[31]  Anthony J. Durkin,et al.  Modulated imaging: quantitative analysis and tomography of turbid media in the spatial-frequency domain. , 2005, Optics letters.

[32]  D. Heimbach,et al.  Burn depth: A review , 2005, World Journal of Surgery.

[33]  Mohammad Torabzadeh,et al.  High-speed spatial frequency domain imaging of rat cortex detects dynamic optical and physiological properties following cardiac arrest and resuscitation , 2017, Neurophotonics.

[34]  Anthony J. Durkin,et al.  Multispectral imaging of tissue absorption and scattering using spatial frequency domain imaging and a computed-tomography imaging spectrometer. , 2011, Journal of biomedical optics.

[35]  Mohammad Torabzadeh,et al.  Compressed single pixel imaging in the spatial frequency domain. , 2017, Journal of biomedical optics.

[36]  Ronald Ignotz,et al.  Hyperspectral Imaging for Burn Depth Assessment in an Animal Model , 2015, Plastic and reconstructive surgery. Global open.

[37]  Gordon T. Kennedy,et al.  A simple burn wound severity assessment classifier based on spatial frequency domain imaging (SFDI) and machine learning , 2019, BiOS.

[38]  Anthony J. Durkin,et al.  Quantitation and mapping of tissue optical properties using modulated imaging. , 2009, Journal of biomedical optics.

[39]  Peter Dziewulski,et al.  Pathophysiology and types of burns , 2004, BMJ : British Medical Journal.