Automated segmentation of dermal fillers in OCT images of mice using convolutional neural networks

Abstract

We present a system for automatic determination of the intradermal volume of hydrogels based on optical coherence tomography (OCT) and deep learning. Volumetric image data was acquired using a custom-built OCT prototype that employs an akinetic swept laser at ∼1310 nm with a bandwidth of 87 nm, providing an axial resolution of ∼6.5 µm in tissue. Three-dimensional data sets of a 10 mm × 10 mm skin patch comprising the intradermal filler and the surrounding tissue were acquired. A convolutional neural network using a u-net-like architecture was trained from slices of 100 OCT volume data sets where the dermal filler volume was manually annotated. Using six-fold cross-validation, a mean accuracy of 0.9938 and a Jaccard similarity coefficient of 0.879 were achieved.

Comparison of optical coherence tomography and high frequency ultrasound imaging in mice for the assessment of skin morphology and intradermal volumes

“…Optical coherence tomography data was acquired using a custom-built swept-source OCT system based on an akinetic light source (Insight Photonic Solutions, Inc.) operating at a central wavelength of 1310 nm and providing a spectral bandwidth of 87 nm. Details of the experimental setup and data processing are described elsewhere 31 . For assessment of intradermal volumes, volumetric data sets of a skin area of 10 × 10 mm², each comprising 512 × 1500 × 1536 voxels, were recorded.…”

“…In order to obtain three-dimensional representations of the applied deposits and to measure their respective intradermal volumes, OCT and HFUS data sets were first used to develop a machine-learning-based algorithm for automatic segmentation 31 . Based on this algorithm, filler borders within the generated cross-sectional images were automatically segmented and the volume was calculated, taking into account the scan range in both lateral dimensions and the refractive indices of the media.…”

Cutaneous optical coherence tomography for longitudinal volumetric assessment of intradermal volumes in a mouse model

“…Dermal filler volumes were determined from OCT volume data sets by automatic segmentation of the filler area within the cross-sectional images and accounting for the scan range in both lateral dimensions. A machine-learning based algorithm for segmenting the deposits has been developed in order to give a three-dimensional representation of the volumes 41 . All automatic filler segmentations were visually examined by an experienced operator and manually corrected in those cases where the automatic algorithm failed to correctly capture the filler area.…”

Deep learning segmentation for optical coherence tomography measurements of the lower tear meniscus

“…3. Except for the layer dimensions, the network is very similar to our previously published segmentation network [23]. In short, the main differences to Ronneberger's U-Net are a reduction in the amount of feature channels, a batch normalization layer before every rectified linear unit (ReLU) activation layer, application of image padding for every convolution and a sigmoid activation function instead of a softmax function for the last network layer.…”

“…Although known for a long time [14], machine learning has been increasingly used only in the past decade to solve image classification [15] and segmentation [16] tasks, mainly driven by advances in the parallelization capabilities of graphics processing units (GPUs) [17]. Using OCT data, segmentation applications can range from detection of retinal layer boundaries [18,19], macular edema [20] and macular fluid [21] to corneal layers [22] and intradermal volumes [23].…”