RESUMEN
This paper proposes a comprehensive method for estimating thrombus formation factors in the left atrial appendage (LAA). First, using 3D CT (Computer Tomography) image data as input, classification of thrombus presence/absence is learned using 3D ResNet. Besides, 3D Grad-CAM is applied to the prediction results to visualize regions of interest in thrombus formation. Second, features are extracted based on the visualization of regions of interest. Using the extracted features and numerical data obtained from the hospital as input, a regression analysis is performed to predict the presence/absence of thrombus using LightGBM. Visualization of regions of interest using 3D ResNet and 3D Grad-CAM shows that the right inferior pulmonary vein and the LAA were particularly correlated with thrombus formation. Estimation of important factors for thrombus formation using LightGBM shows that the LAA ostium area has the greatest influence on thrombus formation.Clinical Relevance-This paper shows the factors that contribute to thrombus formation in the LAA from the viewpoint of three-dimensional structure. In addition, the features considered important in thrombus formation were identified by comparing a variety of features.
Asunto(s)
Apéndice Atrial , Fibrilación Atrial , Cardiopatías , Trombosis , Humanos , Apéndice Atrial/diagnóstico por imagen , Ecocardiografía Transesofágica/métodos , Cardiopatías/diagnóstico por imagen , Trombosis/diagnóstico por imagen , Tomografía Computarizada por Rayos X , Aprendizaje AutomáticoRESUMEN
Laser thermal therapy is one of the treatments for malignant tumors. We developed a thermal endoscope using an ultra-compact thermo-sensor and established a new laparoscopic laser thermal therapy system to heat cancer tissue at an appropriate temperature, focusing on the fact that thermographic cameras are capable of two-dimensional temperature mapping. Hepatocellular carcinoma (N1S1) cells were implanted into the livers of Sprague-Dawley rats (n = 13) to create orthotopic hepatocellular carcinoma. Six of the rats underwent laparoscopic laser thermotherapy (70 °C, 5 min) using the newly developed system, and the others underwent laparoscopic insertion only. Lesion volume measurement and histological evaluation were performed in all of the rats. The laparoscopic laser thermal therapy system provided stable temperature control. When a temperature of 70 °C was used for the set temperature, the temperature of the target cancer was maintained within the range of 68-72 °C for 93.2% of the irradiation time (5 min). The median volume of the tumors that were thermally treated was significantly smaller than that of the untreated tumors. The newly developed laparoscopic laser thermal therapy system was capable of maintaining the temperature of the tumor surface at any desired temperature and was proven to be effective in treatment of the rat hepatocellular carcinoma model.
Asunto(s)
Carcinoma Hepatocelular , Laparoscopía , Terapia por Láser , Neoplasias Hepáticas , Ratas , Animales , Carcinoma Hepatocelular/cirugía , Temperatura , Ratas Sprague-Dawley , Terapia por Láser/métodos , Neoplasias Hepáticas/cirugíaRESUMEN
A screening system for early-stage oral cancer and precancerous lesions should be established because it is difficult to detect them even for specialists and they are often detected too late. In this paper, we propose a method for automatically classifying fluorescence images acquired by ALA-PDD (Photodynamic Diagnosis using 5-Aminolevulinic Acid) into three classes: Normal, Low-Risk, High-Risk. We augment a small image dataset by training GAN (Generative adversarial networks) with Differentiable Augmentation, and then train CNN (Convolutional Neural Network) for the classification by the augmented dataset. Experimental results show good classification results, which suggest that the combination of ALA-PDD and CNN classification is a promising method for oral cancer screening. Clinical Relevance- The method proposed in this paper has a potential to be used as a screening method for early-stage oral cancer and precancerous lesions, that is non-invasive, accurate, easy to use, and does not require specialization.
Asunto(s)
Neoplasias de la Boca , Lesiones Precancerosas , Detección Precoz del Cáncer , Humanos , Neoplasias de la Boca/diagnóstico , Redes Neurales de la Computación , Lesiones Precancerosas/diagnósticoRESUMEN
PURPOSE: To develop and assess the accuracy of a mixed reality (MR) needle guidance application on smartglasses. MATERIALS AND METHODS: An MR needle guidance application on HoloLens2, without pre-procedural CT image reconstruction or import by manually matching the spatial and MR coordinate systems, was developed. First, the accuracy of the target locations in the image overlay at 63 points arranged on a 45 × 35 × 21 cm box and needle angles from 0° to 80°, placed using the MR application, was verified. The needle placement errors from 12 different entry points in a phantom by seven operators (four physicians and three non-physicians) were compared using a linear mixed model between the MR guidance and conventional methods using protractors. RESULTS: The average errors of the target locations and needle angles placed using the MR application were 5.9 ± 2.6 mm and 2.3 ± 1.7°, respectively. The average needle insertion error using the MR guidance was slightly smaller compared to that using the conventional method (8.4 ± 4.0 mm vs. 9.6 ± 5.1 mm, p = 0.091), particularly in the out-of-plane approach (9.6 ± 3.5 mm vs. 12.3 ± 4.6 mm, p = 0.003). The procedural time was longer with MR guidance than with the conventional method (412 ± 134 s vs. 219 ± 66 s, p < 0.001). CONCLUSION: MR needle guidance without pre-procedural CT image import is feasible when matching coordinate systems, and the accuracy of needle insertion is slightly better than that of the conventional method.
Asunto(s)
Realidad Aumentada , Gafas Inteligentes , Humanos , Agujas , Fantasmas de Imagen , Tomografía Computarizada por Rayos X/métodosRESUMEN
PURPOSE: Augmented reality (AR) technology improves the learning process in interventional radiology. This study hypothesized that using AR to train for central venous access is superior to using ultrasound alone. METHODS: This study used an AR central venous catheterization phantom with an internal jugular vein (IJV) and subclavian vein (SCV) made of resin body and soft tubing. Ten radiologists attempted to punctuate, using needle placement simulation, under three conditions (ultrasound-, augmented reality-, and ultrasound and AR-guided methods; US-only, AR-only, and US+AR, respectively) using a smart-glass device (HoloLens, Microsoft, Redmond, WA, USA). Subjective (anatomical understanding and self-confidence for procedure) and objective evaluations (optimized needle position and time) were recorded for each condition. RESULTS: The subjective IJV evaluation showed no difference among the guiding methods (p = 0.26 and p = 0.07 for anatomical understanding and self-confidence for procedure, respectively). Conversely, there were significant improvements in subjective and objective evaluations for SCV using the AR-only and US+AR methods (p < 0.05) and US+AR method (p < 0.05), respectively. The AR-only method reduced the time required to fix the needle position to puncture the SCV (p < 0.05), but its objective evaluation did not improve compared with the US-only method (p = 0.20). CONCLUSION: Adding the AR-guided method to the US-guided method improved subjective and objective evaluations in the SVC procedure. The AR technology-assisted training may be more beneficial for use in difficult procedures. Though the AR-only method saved time, no time saving is expected with AR+US method.
Asunto(s)
Cateterismo Venoso Central/métodos , Venas Yugulares/diagnóstico por imagen , Fantasmas de Imagen , Punciones/métodos , Ultrasonografía Intervencional/métodos , Realidad Aumentada , HumanosRESUMEN
This paper proposes an automatic method for classifying Aortic valvular stenosis (AS) using ECG (Electrocardiogram) images by the deep learning whose training ECG images are annotated by the diagnoses given by the medical doctor who observes the echocardiograms. Besides, it explores the relationship between the trained deep learning network and its determinations, using the Grad-CAM.In this study, one-beat ECG images for 12-leads and 4-leads are generated from ECG's and train CNN's (Convolutional neural network). By applying the Grad-CAM to the trained CNN's, feature areas are detected in the early time range of the one-beat ECG image. Also, by limiting the time range of the ECG image to that of the feature area, the CNN for the 4-lead achieves the best classification performance, which is close to expert medical doctors' diagnoses.Clinical Relevance-This paper achieves as high AS classification performance as medical doctors' diagnoses based on echocardiograms by proposing an automatic method for detecting AS only using ECG.