RESUMO
Angiogenesis has an essential role in the de novo evolution of choroidal melanoma as well as choroidal nevus transformation into melanoma. Differentiating early-stage melanoma from nevus is of high clinical importance; thus, imaging techniques that provide objective information regarding tumor microvasculature structures could aid accurate early detection. Herein, we investigated the feasibility of quantitative high-definition microvessel imaging (qHDMI) for differentiation of choroidal tumors in humans. This new ultrasound-based technique encompasses a series of morphological filtering and vessel enhancement techniques, enabling the visualization of tumor microvessels as small as 150 microns and extracting vessel morphological features as new tumor biomarkers. Distributional differences between the malignant melanomas and benign nevi were tested on 37 patients with choroidal tumors using a non-parametric Wilcoxon rank-sum test, and statistical significance was declared for biomarkers with p-values < 0.05. The ocular oncology diagnosis was choroidal melanoma (malignant) in 21 and choroidal nevus (benign) in 15 patients. The mean thickness of benign and malignant masses was 1.70 ± 0.40 mm and 3.81 ± 2.63 mm, respectively. Six HDMI biomarkers, including number of vessel segments (p = 0.003), number of branch points (p = 0.003), vessel density (p = 0.03), maximum tortuosity (p = 0.001), microvessel fractal dimension (p = 0.002), and maximum diameter (p = 0.003) exhibited significant distributional differences between the two groups. Contrast-free HDMI provided noninvasive imaging and quantification of microvessels of choroidal tumors. The results of this pilot study indicate the potential use of qHDMI as a complementary tool for characterization of small ocular tumors and early detection of choroidal melanoma.
RESUMO
Deep learning is a powerful tool that became practical in 2008, harnessing the power of Graphic Processing Unites, and has developed rapidly in image, video, and natural language processing. There are ongoing developments in the application of deep learning to medical data for a variety of tasks across multiple imaging modalities. The reliability and repeatability of deep learning techniques are of utmost importance if deep learning can be considered a tool for assisting experts, including physicians, radiologists, and sonographers. Owing to the high costs of labeling data, deep learning models are often evaluated against one expert, and it is unknown if any errors fall within a clinically acceptable range. Ultrasound is a commonly used imaging modality for breast cancer screening processes and for visually estimating risk using the Breast Imaging Reporting and Data System score. This process is highly dependent on the skills and experience of the sonographers and radiologists, thereby leading to interobserver variability and interpretation. For these reasons, we propose an interobserver reliability study comparing the performance of a current top-performing deep learning segmentation model against three experts who manually segmented suspicious breast lesions in clinical ultrasound (US) images. We pretrained the model using a US thyroid segmentation dataset with 455 patients and 50,993 images, and trained the model using a US breast segmentation dataset with 733 patients and 29,884 images. We found a mean Fleiss kappa value of 0.78 for the performance of three experts in breast mass segmentation compared to a mean Fleiss kappa value of 0.79 for the performance of experts and the optimized deep learning model.
Assuntos
Aprendizado Profundo , Mama/diagnóstico por imagem , Feminino , Humanos , Processamento de Imagem Assistida por Computador , Variações Dependentes do Observador , Reprodutibilidade dos Testes , UltrassonografiaRESUMO
Medical segmentation is an important but challenging task with applications in standardized report generation, remote medicine and reducing medical exam costs by assisting experts. In this paper, we exploit time sequence information using a novel spatio-temporal recurrent deep learning network to automatically segment the thyroid gland in ultrasound cineclips. We train a DeepLabv3+ based convolutional LSTM model in four stages to perform semantic segmentation by exploiting spatial context from ultrasound cineclips. The backbone DeepLabv3+ model is replicated six times and the output layers are replaced with convolutional LSTM layers in an atrous spatial pyramid pooling configuration. Our proposed model achieves mean intersection over union scores of 0.427 for cysts, 0.533 for nodules and 0.739 for thyroid. We demonstrate the potential application of convolutional LSTM models for thyroid ultrasound segmentation.