Differential diagnosis of focal pancreatic masses is based on endoscopic ultrasound (EUS) guided ... more Differential diagnosis of focal pancreatic masses is based on endoscopic ultrasound (EUS) guided fine needle aspiration biopsy (EUS-FNA/FNB). Several imaging techniques (i.e. gray-scale, color Doppler, contrast-enhancement and elastography) are used for differential diagnosis. However, diagnosis remains highly operator dependent. To address this problem, machine learning algorithms (MLA) can generate an automatic computer-aided diagnosis (CAD) by analyzing a large number of clinical images in real-time. We aimed to develop a MLA to characterize focal pancreatic masses during the EUS procedure. The study included 65 patients with focal pancreatic masses, with 20 EUS images selected from each patient (grayscale, color Doppler, arterial and venous phase contrast-enhancement and elastography). Images were classified based on cytopathology exam as: chronic pseudotumoral pancreatitis (CPP), neuroendocrine tumor (PNET) and ductal adenocarcinoma (PDAC). The MLA is based on a deep learning m...
Serous effusion is a condition of excess accumulation of fluids in serous cavities due to differe... more Serous effusion is a condition of excess accumulation of fluids in serous cavities due to different underlying pathological conditions. The basis of cytopathological assessment of serous effusions is the identification of cells in the fluid based on their morphology and texture. This assessment is a physically and mentally laborious task, and it can also lead to variability among pathologists. In literature, only a small number of feature-based methods are conducted for automated serous cell classification. In this study, a transfer learning with pre-trained deep convolutional neural networks (ConvNets) is proposed to automatically identify 11 different categories of serous cells in effusion cytology. Unlike the methods which rely on the extraction of cellular features such as morphology and texture, this method is an appearance-based machine learning approach. We fine-tuned four pre-trained ConvNet architectures that are AlexNet, GoogleNet, ResNet and DenseNet on the serous cell dataset. To reduce the overfitting effect, we augmented the data by image rotation, translation, and mirroring. The proposed method was evaluated on both original and augmented sets of serous cells derived from a publicly available dataset. Among the four ConvNet models, ResNet and DenseNet obtained the highest accuracies of 93.44% and 92.90%. However, when two models were compared in terms of accuracy and model complexity, ResNet-TL was selected as the best network model. When compared to the results without data augmentation, data augmentation increased the accuracy rate approximately 10%. Results show that higher classification results were achieved than other traditional methods without requiring precise segmentation. Keywords Serous effusion • Cytopathological assessment • Cell classification • Convolutional neural networks • Transfer learning Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Journal of Gastrointestinal and Liver Diseases, 2021
Background and Aims: Mucosal healing (MH) is associated with a stable course of Crohn’s disease (... more Background and Aims: Mucosal healing (MH) is associated with a stable course of Crohn’s disease (CD) which can be assessed by confocal laser endomicroscopy (CLE). To minimize the operator’s errors and automate assessment of CLE images, we used a deep learning (DL) model for image analysis. We hypothesized that DL combined with convolutional neural networks (CNNs) and long short-term memory (LSTM) can distinguish between normal and inflamed colonic mucosa from CLE images. Methods: The study included 54 patients, 32 with known active CD, and 22 control patients (18 CD patients with MH and four normal mucosa patients with no history of inflammatory bowel diseases). We designed and trained a deep convolutional neural network to detect active CD using 6,205 endomicroscopy images classified as active CD inflammation (3,672 images) and control mucosal healing or no inflammation (2,533 images). CLE imaging was performed on four colorectal areas and the terminal ileum. Gold standard was repr...
Winning academic innovation requires a new team of interdisciplinary specialists and electronic r... more Winning academic innovation requires a new team of interdisciplinary specialists and electronic resources for information management. The industry model of innovation social network (ISN) can serve as an example of a proven culture of innovation. It contains a complex research and development network (RDN) led by a product specialist (PS) who coordinates technical, clinical, and business experts to ensure fitness for use of the newly developed products. Currently, the academic innovation social network (aISN) is inefficient and made of only a few of the industry functions such as the clinician/scientist inventors and their laboratory personnel or clinical team and the university technology transfer office (TTO). Moreover, the essential information and knowledge for innovation (i.e., the shortcomings of current clinical and research approaches) are not recorded, more emphasis being currently placed on what works rather than what needs to be improved. A new academic innovation program led by a director of medical innovation (DMI) was developed to accelerate academic innovation at the level of process and execution. As a result, a more complete aISN was created where more academic innovation projects are generated from real clinical needs and advanced toward licensing and venture capital investment. Furthermore, a new innovation electronic platform, the academic innovation management system (AIMS), was developed to connect academia with the external network of service providers and capital investors. The newly developed innovation team spans academic and industry environments and is better prepared to advance winning innovative projects toward commercialization and clinical use.
Total knee arthroplasty following valgus deformity is a challenging procedure due to the unique s... more Total knee arthroplasty following valgus deformity is a challenging procedure due to the unique set of problems that must be addressed. The aim of this study is to determine, with a finite element analysis, the load distribution for an inclined valgus prosthetic balanced knee and to compare these results with those of a prosthetic balanced knee with an uninclined interline. Computational simulations, using finite element analysis, focused on a comparision between load intensity and distribution for these situations. We studied valgus inclination at 3 and 8 degrees. We noticed that for an inclination of 3 degrees, the forces are distributed almost symmetrically on both condyles, similar to the distribution of forces in the uninclined interline case. The maximum contact pressure is greater, increasing from 15 MPa to 19.3 MPa (28%). At 8 degrees of inclination, the contact patch moved anterolateraly on the tibia, meaning that the tibial condyles will be unequally loaded. The maximum contact pressure increases to 25 MPa (66%). These greater forces could lead to polyethylene wear and collapse. Additional tibial resection could be a useful method for balancing in severe valgus knee, when valgus inlination does not exceed 3 degrees.
Aim: In this paper we proposed different architectures of convolutional neural network (CNN) to c... more Aim: In this paper we proposed different architectures of convolutional neural network (CNN) to classify fatty liver disease in images using only pixels and diagnosis labels as input. We trained and validated our models using a dataset of 629 images consisting of 2 types of liver images, normal and liver steatosis. Material and methods: We assessed two pre-trained models of convolutional neural networks, Inception-v3 and VGG-16 using fine-tuning. Both models were pre-trained on ImageNet dataset to extract features from B-mode ultrasound liver images. The results obtained through these methods were compared for selecting the predictive model with the best performance metrics. We trained the two models using a dataset of 262 images of liver steatosis and 234 images of normal liver. We assessed the models using a dataset of 70 liver steatosis images and 63 normal liver images. Results. The proposed model that used Inception v3 obtained a 93.23% test accuracy with a sensitivity of 89.9%% and a precision of 96.6%, and areas under each receiver operating characteristic curves (ROC AUC) of 0.93. The other proposed model that used VGG-16, obtained a 90.77% test accuracy with a sensitivity of 88.9% and a precision of 92.85%, and areas under each receiver operating characteristic curves (ROC AUC) of 0.91. Conclusion. The deep learning algorithms that we proposed to detect steatosis and classify the images in normal and fatty liver images, yields an excellent test performance of over 90%. However, future larger studies are required in order to establish how these algorithms can be implemented in a clinical setting.
To evaluate the accuracy of a 3-dimensional (3D) navigation system using electromagnetically trac... more To evaluate the accuracy of a 3-dimensional (3D) navigation system using electromagnetically tracked tools to explore its potential in patients. Methods: The 3D navigation accuracy was quantified on a phantom and in a porcine model using the same setup and vascular interventional suite. A box-shaped phantom with 16 markers was scanned in 5 different positions using computed tomography (CT). The 3D navigation system registered each CT volume in the magnetic field. A tracked needle was pointed at the physical markers, and the spatial distances between the tracked needle positions and the markers were calculated. Contrast-enhanced CT images were acquired from 6 swine. The 3D navigation system registered each CT volume in the magnetic field. An electromagnetically tracked guidewire and catheter were visualized in the 3D image and navigated to 4 specified targets. At each target, the spatial distance between the tracked guidewire tip position and the actual position, verified by a CT control, was calculated. Results: The mean accuracy on the phantom was 1.2860.53 mm, and 90% of the measured distances were #1.90 mm. The mean accuracy in swine was 4.1861.76 mm, and 90% of the measured distances were #5.73 mm. Conclusion: This 3D navigation system demonstrates good ex vivo accuracy and is sufficiently accurate in vivo to explore its potential for improved endovascular navigation.
Needle insertion in biological tissue has attracted considerable attention due to its application... more Needle insertion in biological tissue has attracted considerable attention due to its application in minimally invasive procedures such as laparoscopy or transcutaneous biopsy. In this paper the force of the Veress needle insertion into the abdominal wall and the von Mises stress were studied, demonstrating the ability of finite element models to provide additional understanding of the processes taking place. Veress needle insertion force may cause complications during surgery, the most common being vascular lesions, thus affecting the precision and duration of surgery assisted by a portable abdominal insufflation device. This study was the first step in developing a force feedback for needle insertion into the abdominal wall assisted by a portable abdominal insufflation device. The CAD model of the prototype of a portable abdominal insufflation device was made. Then the prototype of a portable abdominal insufflation device was developed. For testing purposes an artificial silicone ...
Lung cancer is the leading cause of death among different type of cancers worldwide. To decrease ... more Lung cancer is the leading cause of death among different type of cancers worldwide. To decrease the recorded mortality rates, earlier diagnosis and timely treatment are essential. The two-degrees-of-freedom robotic and image guided navigation system, ENDORO is designed to find peripheral pulmonary nodules using an electromagnetic tracking equipment, a navigation software and a specially designed biopsy catheter. The aim of the present study was to assess the accuracy of the ENDORO system in a rigid lung phantom built from a set of computerized tomography scans of a patient. We describe the testing environment, necessary configurations and navigation results for main lung airways. The evaluation consists in identifying all navigable paths from the entry point to the lung periphery, counting the number of intersection crossing, the time required for navigation and the number of movements the robot performed. The results demonstrate appropriate navigation accuracy for catheters with a rigid tip. Future studies will test robotic navigation with a catheter with a bending tip.
To study the biomechanical changes that appear in different pathological cases and to establish t... more To study the biomechanical changes that appear in different pathological cases and to establish the efficiency of some types of osteothomies with different fixation systems and prostheses, a special five degrees of freedom simulator for the knee joint was developed. A model of the knee joint bones of a patient with total knee replacement (TKR) was reconstructed physically by rapid-prototyping
Medical Imaging 2007: Visualization and Image-Guided Procedures, Mar 8, 2007
PET (Positron Emission Tomography) scanning has become a dominant force in oncology care because ... more PET (Positron Emission Tomography) scanning has become a dominant force in oncology care because of its ability to identify regions of abnormal function. The current generation of PET scanners is focused on whole-body imaging, and does not address aspects that might be required by surgeons or other practitioners interested in the function of particular body parts. We are therefore developing and testing a new class of hand-operated molecular imaging scanners designed for use with physical examinations and intraoperative visualization. These devices integrate several technological advances, including (1) nanotechnology-based quantum photodetectors for high performance at low light levels, (2) continuous position tracking of the detectors so that they form a larger 'virtual detector', and (3) novel reconstruction algorithms that do not depend on a circular or ring geometry. The first incarnations of this device will be in the form of a glove with finger-mounted detectors or in a "sash" of detectors that can be draped over the patient. Potential applications include image-guided biopsy, surgical resection of tumors, assessment of inflammatory conditions, and early cancer detection. Our first prototype is in development now along with a clinical protocol for pilot testing.
Introduction: Numerous anti-angiogenic agents are currently developed to limit tumor growth and m... more Introduction: Numerous anti-angiogenic agents are currently developed to limit tumor growth and metastasis. While these drugs offer hope for cancer patients, their transient effect on tumor vasculature is difficult to assess in clinical settings. Confocal laser endomicroscopy (CLE) is a novel endoscopic imaging technology that enables histological examination of the gastrointestinal mucosa. The aim of the present study was to evaluate the feasibility of using CLE to image the vascular network in fresh biopsies of human colorectal tissue. For this purpose we have imaged normal and malignant biopsy tissue samples and compared the vascular network parameters obtained with CLE with established histopathology techniques. Materials and Methods: Fresh non-fixed biopsy samples of both normal and malignant colorectal mucosa were stained with fluorescently labeled anti-CD31 antibodies and imaged by CLE using a dedicated endomicroscopy system. Corresponding biopsy samples underwent immunohistochemical staining for CD31, assessing the microvessel density (MVD) and vascular areas for comparison with CLE data, which were measured offline using specific software. Results: The vessels were imaged by CLE in both normal and tumor samples. The average diameter of normal vessels was 8.560.9 mm whereas in tumor samples it was 13.560.7 mm (p = 0.0049). Vascular density was 188.7624.9 vessels/mm 2 in the normal tissue vs. 242.4616.1 vessels/mm 2 in the colorectal cancer samples (p = 0.1201). In the immunohistochemistry samples, the MVD was 211.2642.9/mm 2 and 351.3639.6/mm 2 for normal and malignant mucosa, respectively. The vascular area was 2.960.5% of total tissue area for the normal mucosa and 8.562.1% for primary colorectal cancer tissue. Conclusion: Selective imaging of blood vessels with CLE is feasible in normal and tumor colorectal tissue by using fluorescently labeled antibodies targeted against an endothelial marker. The method could be translated into the clinical setting for monitoring of anti-angiogenic therapy.
no. 3SEE/30.06.2014. The funders had no significantly different between normal and cancer samples... more no. 3SEE/30.06.2014. The funders had no significantly different between normal and cancer samples. Next, a two-layer feed forward neural network was used to train and automatically diagnose the malignant samples, based on the seven parameters tested. The neural network operations were cross-entropy with the results: training:
W249 "ultrasonography" or "endoscopic ultrasound" and "tomography x-ray computed" (MeSH) and "MR ... more W249 "ultrasonography" or "endoscopic ultrasound" and "tomography x-ray computed" (MeSH) and "MR image fusion" were searched. We also searched for cases in which image fusion was replaced with hybrid imaging or an electromagnetic navigation device. Furthermore, the terms "real-time virtual sonography," "volume navigation," and "virtual navigator" were added. Finally, the reference lists of the retrieved articles were hand-searched for further references. All articles involving offline image fusion were excluded. Technical Solutions To fuse medical imaging information obtained from different modalities at different times, a spatial coregistration is mandatory to ensure that the pixels from the various datasets represent approximately the same volume. There are several methods to achieve this goal depending on the imaging modalities and the accuracy needed. For a correct coregistration, a two-step technique is performed automatically by a computer: image registration and data reslicing. The process is virtually instantaneous. The registration process requires the computation of a transformation matrix composed of translations and rotations that define the differences in spatial location between two datasets [3]. One method to coregister two datasets is to define a series of standard registration points, which can be either external (fiducials, placed on the patient) or internal (common anatomic structures) [4]. The coordinates of each marker are measured and the
Feasibility Study of Tridimensional Co-Registration of Endoscopic Ultrasound and Dynamic Spiral C... more Feasibility Study of Tridimensional Co-Registration of Endoscopic Ultrasound and Dynamic Spiral Computer Tomography Procedures for Real-Time Evaluation of Tumor Angiogenesis Lucian G. Gruionu, Adrian Saftoiu, Alexandru L. Iordache, Ana Maria Ioncica, Daniela Burtea, Daniela Dumitrescu Department of Engineering, University of Craiova, Craiova, Romania; Medinsys, Craiova, Romania; Research Center of Gastroenterology and Hepatology, University of Medicine and Pharmacy Craiova, Craiova, Romania; Department of Radiology and Imaging, University of Medicine and Pharmacy Craiova, Craiova, Romania Background: Endoscopic ultrasound (EUS) and computer tomography (CT) are considered procedures of choice for the diagnosis and staging of both digestive cancers (esophago-gastric and pancreatico-biliary), but also lung cancer. CT has the advantage of a large observation field with increased accuracy for the definition of N and M stage, while EUS is performing better for the targeted assessment of T and N stage, including EUS-guided fine needle aspiration procedures that allow tissue confirmation of malignancy. A hybrid imaging procedure with co-registration of both EUS and CT during the same examination would be highly desirable for improved TNM staging, but also better description of anatomical structures, increased diagnostic confidence and shorter learning curve for linear EUS procedures. Patients and method: The aim of this feasibility study was to test a new hybrid system of real-time EUS displayed simultaneously with the corresponding dynamic CT section, reconstructed virtually based on a previously stored 3D volume data set. The images were co-registered based on electromagnetical (EM) tracking of the EUS transducer position, using a wired magnetic positioning sensor embedded and fixed into the sheet of a usual EUSFNA needle inserted and locked into the biopsy channel of the EUS scope. The initial calibration (positioning) of the needle-scope assembly relative to the 3D coordinate system was based on several external markers previously fixed, closed to the anatomic region of interest, and also evident on the CT scans. Results: The system was tested initially on a specially designed EUS phantom filled with de-aerated water and silicon inclusions simulating malignant masses, showing small errors (maximum 3 mm) during co-registration of EUS and CT images. Furthermore, the same system was also tested in 6 patients with digestive and lung cancers with good results in decreasing the time of tumor localization and identification as compared with classical EUS procedures. Also, based on the EUS system capabilities, contrast-enhanced power Doppler EUS was also visualized simultaneously with dynamic spiral CT data, allowing an excellent estimation of angiogenesis inside the tumors. Conclusion: Based on the EM tracking of the EUS transducer position and co-registration software with 3D dynamic CT reconstructions, a hybrid system of real-time EUS-CT co-registration was developed. The system should be further tested in larger clinical studies, to describe better the clinical impact of increased diagnostic confidence by direct comparisons between the same lesions based on different imaging modalities, but also to shorten the difficult learning curve of linear EUS. Mo1512 Identification of the Line Demarcating Gastric Cancer From Normal Mucosa by Magnification Endoscopy With NBI Toshihisa Takeuchi, Yuichi Kojima, Yukiko Yoda, Satoshi Tokioka, Eiji Umegaki, Kazuhide Higuchi 2nd Dep of Internal Medicine, Osaka Medical Collage, Takatsuki, Japan [Introduction] Because endoscopic submucosal dissection (ESD) in patients with gastric cancer has become widespread, it is important to accurately identify the demarcation line of gastric cancer. Until now, it has been shown that in qualitative diagnosis of gastric cancer, observation of the surface microstructure and microvascular pattern by magnified endoscopy combined with narrow-band imaging (NBI) enables differentiation of benign and malignant lesions and histological determination of the depressed-type gastric cancer. However, there is no consensus on the identification of a demarcation line of gastric cancer by magnification endoscopy combined with NBI. [Objective] To elucidate the usefulness of magnified endoscopy combined with NBI for identifying a demarcation line of gastric cancer. [Subjects and Method] We included 572 lesions from patients with early gastric cancer who received ESD from 2002 to 2009. We introduced magnified endoscopy combined with NBI from 2006, after which we identified a demarcation line of gastric cancer prior to ESD in 264 lesions and compared the pathological findings with resection samples. We examined the incidence of positive lateral resection margins (i.e., the percentage of inconsistency in the identification of a demarcation line) among the resection samples as well as the factors before and after NBI. [Results] (1) The incidence of positive lateral…
American Journal of Physiology-heart and Circulatory Physiology, Jun 1, 2005
Arteriolar arcades provide alternate pathways for blood flow after obstruction of arteries or art... more Arteriolar arcades provide alternate pathways for blood flow after obstruction of arteries or arterioles such as occurs in stroke and coronary and peripheral vascular disease. When obstruction is prolonged, remaining vessels adjust their diameters chronically in response to altered hemodynamic and metabolic conditions. Here, the effectiveness of arcades in maintaining perfusion both immediately following obstruction and after structural adaptation was examined. Morphometric data from a vascular casting of the pig triceps brachii muscle and published data were used to develop a computational model for the hemodynamics and structural adaptation of the arcade network between two feed artery branches, FA1 and FA2. The predicted total flow to capillaries (Q TA) in the region initially supplied by FA2 decreased to 26% of the normal value immediately after FA2 obstruction but was restored to 78% of the normal value after adaptation. After obstruction of 1-10 randomly selected arcade segments, Q TA was on average 18% higher in the arcade network than in a corresponding two-tree network without arcades. Structural adaptation increased Q TA by an additional 16% in the arcade network but had almost no effect in the two-tree network. These results indicate that arcades can partially maintain blood flow after vascular blockage and that this effect is substantially enhanced by structural adaptation.
At present, deep learning becomes an important tool in medical image analysis, with good performa... more At present, deep learning becomes an important tool in medical image analysis, with good performance in diagnosing, pattern detection, and segmentation. Ultrasound imaging offers an easy and rapid method to detect and diagnose thyroid disorders. With the help of a computer-aided diagnosis (CAD) system based on deep learning, we have the possibility of real-time and non-invasive diagnosing of thyroidal US images. This paper proposed a study based on deep learning with transfer learning for differentiating the thyroidal ultrasound images using image pixels and diagnosis labels as inputs. We trained, assessed, and compared two pre-trained models (VGG-19 and Inception v3) using a dataset of ultrasound images consisting of 2 types of thyroid ultrasound images: autoimmune and normal. The training dataset consisted of 615 thyroid ultrasound images, from which 415 images were diagnosed as autoimmune, and 200 images as normal. The models were assessed using a dataset of 120 images, from which 80 images were diagnosed as autoimmune, and 40 images diagnosed as normal. The two deep learning models obtained very good results, as follows: the pre-trained VGG-19 model obtained 98.60% for the overall test accuracy with an overall specificity of 98.94% and overall sensitivity of 97.97%, while the Inception v3 model obtained 96.4% for the overall test accuracy with an overall specificity of 95.58% and overall sensitivity of 95.58.
Differential diagnosis of focal pancreatic masses is based on endoscopic ultrasound (EUS) guided ... more Differential diagnosis of focal pancreatic masses is based on endoscopic ultrasound (EUS) guided fine needle aspiration biopsy (EUS-FNA/FNB). Several imaging techniques (i.e. gray-scale, color Doppler, contrast-enhancement and elastography) are used for differential diagnosis. However, diagnosis remains highly operator dependent. To address this problem, machine learning algorithms (MLA) can generate an automatic computer-aided diagnosis (CAD) by analyzing a large number of clinical images in real-time. We aimed to develop a MLA to characterize focal pancreatic masses during the EUS procedure. The study included 65 patients with focal pancreatic masses, with 20 EUS images selected from each patient (grayscale, color Doppler, arterial and venous phase contrast-enhancement and elastography). Images were classified based on cytopathology exam as: chronic pseudotumoral pancreatitis (CPP), neuroendocrine tumor (PNET) and ductal adenocarcinoma (PDAC). The MLA is based on a deep learning m...
Serous effusion is a condition of excess accumulation of fluids in serous cavities due to differe... more Serous effusion is a condition of excess accumulation of fluids in serous cavities due to different underlying pathological conditions. The basis of cytopathological assessment of serous effusions is the identification of cells in the fluid based on their morphology and texture. This assessment is a physically and mentally laborious task, and it can also lead to variability among pathologists. In literature, only a small number of feature-based methods are conducted for automated serous cell classification. In this study, a transfer learning with pre-trained deep convolutional neural networks (ConvNets) is proposed to automatically identify 11 different categories of serous cells in effusion cytology. Unlike the methods which rely on the extraction of cellular features such as morphology and texture, this method is an appearance-based machine learning approach. We fine-tuned four pre-trained ConvNet architectures that are AlexNet, GoogleNet, ResNet and DenseNet on the serous cell dataset. To reduce the overfitting effect, we augmented the data by image rotation, translation, and mirroring. The proposed method was evaluated on both original and augmented sets of serous cells derived from a publicly available dataset. Among the four ConvNet models, ResNet and DenseNet obtained the highest accuracies of 93.44% and 92.90%. However, when two models were compared in terms of accuracy and model complexity, ResNet-TL was selected as the best network model. When compared to the results without data augmentation, data augmentation increased the accuracy rate approximately 10%. Results show that higher classification results were achieved than other traditional methods without requiring precise segmentation. Keywords Serous effusion • Cytopathological assessment • Cell classification • Convolutional neural networks • Transfer learning Publisher's note Springer Nature remains neutral with regard to jurisdictional claims in published maps and institutional affiliations.
Journal of Gastrointestinal and Liver Diseases, 2021
Background and Aims: Mucosal healing (MH) is associated with a stable course of Crohn’s disease (... more Background and Aims: Mucosal healing (MH) is associated with a stable course of Crohn’s disease (CD) which can be assessed by confocal laser endomicroscopy (CLE). To minimize the operator’s errors and automate assessment of CLE images, we used a deep learning (DL) model for image analysis. We hypothesized that DL combined with convolutional neural networks (CNNs) and long short-term memory (LSTM) can distinguish between normal and inflamed colonic mucosa from CLE images. Methods: The study included 54 patients, 32 with known active CD, and 22 control patients (18 CD patients with MH and four normal mucosa patients with no history of inflammatory bowel diseases). We designed and trained a deep convolutional neural network to detect active CD using 6,205 endomicroscopy images classified as active CD inflammation (3,672 images) and control mucosal healing or no inflammation (2,533 images). CLE imaging was performed on four colorectal areas and the terminal ileum. Gold standard was repr...
Winning academic innovation requires a new team of interdisciplinary specialists and electronic r... more Winning academic innovation requires a new team of interdisciplinary specialists and electronic resources for information management. The industry model of innovation social network (ISN) can serve as an example of a proven culture of innovation. It contains a complex research and development network (RDN) led by a product specialist (PS) who coordinates technical, clinical, and business experts to ensure fitness for use of the newly developed products. Currently, the academic innovation social network (aISN) is inefficient and made of only a few of the industry functions such as the clinician/scientist inventors and their laboratory personnel or clinical team and the university technology transfer office (TTO). Moreover, the essential information and knowledge for innovation (i.e., the shortcomings of current clinical and research approaches) are not recorded, more emphasis being currently placed on what works rather than what needs to be improved. A new academic innovation program led by a director of medical innovation (DMI) was developed to accelerate academic innovation at the level of process and execution. As a result, a more complete aISN was created where more academic innovation projects are generated from real clinical needs and advanced toward licensing and venture capital investment. Furthermore, a new innovation electronic platform, the academic innovation management system (AIMS), was developed to connect academia with the external network of service providers and capital investors. The newly developed innovation team spans academic and industry environments and is better prepared to advance winning innovative projects toward commercialization and clinical use.
Total knee arthroplasty following valgus deformity is a challenging procedure due to the unique s... more Total knee arthroplasty following valgus deformity is a challenging procedure due to the unique set of problems that must be addressed. The aim of this study is to determine, with a finite element analysis, the load distribution for an inclined valgus prosthetic balanced knee and to compare these results with those of a prosthetic balanced knee with an uninclined interline. Computational simulations, using finite element analysis, focused on a comparision between load intensity and distribution for these situations. We studied valgus inclination at 3 and 8 degrees. We noticed that for an inclination of 3 degrees, the forces are distributed almost symmetrically on both condyles, similar to the distribution of forces in the uninclined interline case. The maximum contact pressure is greater, increasing from 15 MPa to 19.3 MPa (28%). At 8 degrees of inclination, the contact patch moved anterolateraly on the tibia, meaning that the tibial condyles will be unequally loaded. The maximum contact pressure increases to 25 MPa (66%). These greater forces could lead to polyethylene wear and collapse. Additional tibial resection could be a useful method for balancing in severe valgus knee, when valgus inlination does not exceed 3 degrees.
Aim: In this paper we proposed different architectures of convolutional neural network (CNN) to c... more Aim: In this paper we proposed different architectures of convolutional neural network (CNN) to classify fatty liver disease in images using only pixels and diagnosis labels as input. We trained and validated our models using a dataset of 629 images consisting of 2 types of liver images, normal and liver steatosis. Material and methods: We assessed two pre-trained models of convolutional neural networks, Inception-v3 and VGG-16 using fine-tuning. Both models were pre-trained on ImageNet dataset to extract features from B-mode ultrasound liver images. The results obtained through these methods were compared for selecting the predictive model with the best performance metrics. We trained the two models using a dataset of 262 images of liver steatosis and 234 images of normal liver. We assessed the models using a dataset of 70 liver steatosis images and 63 normal liver images. Results. The proposed model that used Inception v3 obtained a 93.23% test accuracy with a sensitivity of 89.9%% and a precision of 96.6%, and areas under each receiver operating characteristic curves (ROC AUC) of 0.93. The other proposed model that used VGG-16, obtained a 90.77% test accuracy with a sensitivity of 88.9% and a precision of 92.85%, and areas under each receiver operating characteristic curves (ROC AUC) of 0.91. Conclusion. The deep learning algorithms that we proposed to detect steatosis and classify the images in normal and fatty liver images, yields an excellent test performance of over 90%. However, future larger studies are required in order to establish how these algorithms can be implemented in a clinical setting.
To evaluate the accuracy of a 3-dimensional (3D) navigation system using electromagnetically trac... more To evaluate the accuracy of a 3-dimensional (3D) navigation system using electromagnetically tracked tools to explore its potential in patients. Methods: The 3D navigation accuracy was quantified on a phantom and in a porcine model using the same setup and vascular interventional suite. A box-shaped phantom with 16 markers was scanned in 5 different positions using computed tomography (CT). The 3D navigation system registered each CT volume in the magnetic field. A tracked needle was pointed at the physical markers, and the spatial distances between the tracked needle positions and the markers were calculated. Contrast-enhanced CT images were acquired from 6 swine. The 3D navigation system registered each CT volume in the magnetic field. An electromagnetically tracked guidewire and catheter were visualized in the 3D image and navigated to 4 specified targets. At each target, the spatial distance between the tracked guidewire tip position and the actual position, verified by a CT control, was calculated. Results: The mean accuracy on the phantom was 1.2860.53 mm, and 90% of the measured distances were #1.90 mm. The mean accuracy in swine was 4.1861.76 mm, and 90% of the measured distances were #5.73 mm. Conclusion: This 3D navigation system demonstrates good ex vivo accuracy and is sufficiently accurate in vivo to explore its potential for improved endovascular navigation.
Needle insertion in biological tissue has attracted considerable attention due to its application... more Needle insertion in biological tissue has attracted considerable attention due to its application in minimally invasive procedures such as laparoscopy or transcutaneous biopsy. In this paper the force of the Veress needle insertion into the abdominal wall and the von Mises stress were studied, demonstrating the ability of finite element models to provide additional understanding of the processes taking place. Veress needle insertion force may cause complications during surgery, the most common being vascular lesions, thus affecting the precision and duration of surgery assisted by a portable abdominal insufflation device. This study was the first step in developing a force feedback for needle insertion into the abdominal wall assisted by a portable abdominal insufflation device. The CAD model of the prototype of a portable abdominal insufflation device was made. Then the prototype of a portable abdominal insufflation device was developed. For testing purposes an artificial silicone ...
Lung cancer is the leading cause of death among different type of cancers worldwide. To decrease ... more Lung cancer is the leading cause of death among different type of cancers worldwide. To decrease the recorded mortality rates, earlier diagnosis and timely treatment are essential. The two-degrees-of-freedom robotic and image guided navigation system, ENDORO is designed to find peripheral pulmonary nodules using an electromagnetic tracking equipment, a navigation software and a specially designed biopsy catheter. The aim of the present study was to assess the accuracy of the ENDORO system in a rigid lung phantom built from a set of computerized tomography scans of a patient. We describe the testing environment, necessary configurations and navigation results for main lung airways. The evaluation consists in identifying all navigable paths from the entry point to the lung periphery, counting the number of intersection crossing, the time required for navigation and the number of movements the robot performed. The results demonstrate appropriate navigation accuracy for catheters with a rigid tip. Future studies will test robotic navigation with a catheter with a bending tip.
To study the biomechanical changes that appear in different pathological cases and to establish t... more To study the biomechanical changes that appear in different pathological cases and to establish the efficiency of some types of osteothomies with different fixation systems and prostheses, a special five degrees of freedom simulator for the knee joint was developed. A model of the knee joint bones of a patient with total knee replacement (TKR) was reconstructed physically by rapid-prototyping
Medical Imaging 2007: Visualization and Image-Guided Procedures, Mar 8, 2007
PET (Positron Emission Tomography) scanning has become a dominant force in oncology care because ... more PET (Positron Emission Tomography) scanning has become a dominant force in oncology care because of its ability to identify regions of abnormal function. The current generation of PET scanners is focused on whole-body imaging, and does not address aspects that might be required by surgeons or other practitioners interested in the function of particular body parts. We are therefore developing and testing a new class of hand-operated molecular imaging scanners designed for use with physical examinations and intraoperative visualization. These devices integrate several technological advances, including (1) nanotechnology-based quantum photodetectors for high performance at low light levels, (2) continuous position tracking of the detectors so that they form a larger 'virtual detector', and (3) novel reconstruction algorithms that do not depend on a circular or ring geometry. The first incarnations of this device will be in the form of a glove with finger-mounted detectors or in a "sash" of detectors that can be draped over the patient. Potential applications include image-guided biopsy, surgical resection of tumors, assessment of inflammatory conditions, and early cancer detection. Our first prototype is in development now along with a clinical protocol for pilot testing.
Introduction: Numerous anti-angiogenic agents are currently developed to limit tumor growth and m... more Introduction: Numerous anti-angiogenic agents are currently developed to limit tumor growth and metastasis. While these drugs offer hope for cancer patients, their transient effect on tumor vasculature is difficult to assess in clinical settings. Confocal laser endomicroscopy (CLE) is a novel endoscopic imaging technology that enables histological examination of the gastrointestinal mucosa. The aim of the present study was to evaluate the feasibility of using CLE to image the vascular network in fresh biopsies of human colorectal tissue. For this purpose we have imaged normal and malignant biopsy tissue samples and compared the vascular network parameters obtained with CLE with established histopathology techniques. Materials and Methods: Fresh non-fixed biopsy samples of both normal and malignant colorectal mucosa were stained with fluorescently labeled anti-CD31 antibodies and imaged by CLE using a dedicated endomicroscopy system. Corresponding biopsy samples underwent immunohistochemical staining for CD31, assessing the microvessel density (MVD) and vascular areas for comparison with CLE data, which were measured offline using specific software. Results: The vessels were imaged by CLE in both normal and tumor samples. The average diameter of normal vessels was 8.560.9 mm whereas in tumor samples it was 13.560.7 mm (p = 0.0049). Vascular density was 188.7624.9 vessels/mm 2 in the normal tissue vs. 242.4616.1 vessels/mm 2 in the colorectal cancer samples (p = 0.1201). In the immunohistochemistry samples, the MVD was 211.2642.9/mm 2 and 351.3639.6/mm 2 for normal and malignant mucosa, respectively. The vascular area was 2.960.5% of total tissue area for the normal mucosa and 8.562.1% for primary colorectal cancer tissue. Conclusion: Selective imaging of blood vessels with CLE is feasible in normal and tumor colorectal tissue by using fluorescently labeled antibodies targeted against an endothelial marker. The method could be translated into the clinical setting for monitoring of anti-angiogenic therapy.
no. 3SEE/30.06.2014. The funders had no significantly different between normal and cancer samples... more no. 3SEE/30.06.2014. The funders had no significantly different between normal and cancer samples. Next, a two-layer feed forward neural network was used to train and automatically diagnose the malignant samples, based on the seven parameters tested. The neural network operations were cross-entropy with the results: training:
W249 "ultrasonography" or "endoscopic ultrasound" and "tomography x-ray computed" (MeSH) and "MR ... more W249 "ultrasonography" or "endoscopic ultrasound" and "tomography x-ray computed" (MeSH) and "MR image fusion" were searched. We also searched for cases in which image fusion was replaced with hybrid imaging or an electromagnetic navigation device. Furthermore, the terms "real-time virtual sonography," "volume navigation," and "virtual navigator" were added. Finally, the reference lists of the retrieved articles were hand-searched for further references. All articles involving offline image fusion were excluded. Technical Solutions To fuse medical imaging information obtained from different modalities at different times, a spatial coregistration is mandatory to ensure that the pixels from the various datasets represent approximately the same volume. There are several methods to achieve this goal depending on the imaging modalities and the accuracy needed. For a correct coregistration, a two-step technique is performed automatically by a computer: image registration and data reslicing. The process is virtually instantaneous. The registration process requires the computation of a transformation matrix composed of translations and rotations that define the differences in spatial location between two datasets [3]. One method to coregister two datasets is to define a series of standard registration points, which can be either external (fiducials, placed on the patient) or internal (common anatomic structures) [4]. The coordinates of each marker are measured and the
Feasibility Study of Tridimensional Co-Registration of Endoscopic Ultrasound and Dynamic Spiral C... more Feasibility Study of Tridimensional Co-Registration of Endoscopic Ultrasound and Dynamic Spiral Computer Tomography Procedures for Real-Time Evaluation of Tumor Angiogenesis Lucian G. Gruionu, Adrian Saftoiu, Alexandru L. Iordache, Ana Maria Ioncica, Daniela Burtea, Daniela Dumitrescu Department of Engineering, University of Craiova, Craiova, Romania; Medinsys, Craiova, Romania; Research Center of Gastroenterology and Hepatology, University of Medicine and Pharmacy Craiova, Craiova, Romania; Department of Radiology and Imaging, University of Medicine and Pharmacy Craiova, Craiova, Romania Background: Endoscopic ultrasound (EUS) and computer tomography (CT) are considered procedures of choice for the diagnosis and staging of both digestive cancers (esophago-gastric and pancreatico-biliary), but also lung cancer. CT has the advantage of a large observation field with increased accuracy for the definition of N and M stage, while EUS is performing better for the targeted assessment of T and N stage, including EUS-guided fine needle aspiration procedures that allow tissue confirmation of malignancy. A hybrid imaging procedure with co-registration of both EUS and CT during the same examination would be highly desirable for improved TNM staging, but also better description of anatomical structures, increased diagnostic confidence and shorter learning curve for linear EUS procedures. Patients and method: The aim of this feasibility study was to test a new hybrid system of real-time EUS displayed simultaneously with the corresponding dynamic CT section, reconstructed virtually based on a previously stored 3D volume data set. The images were co-registered based on electromagnetical (EM) tracking of the EUS transducer position, using a wired magnetic positioning sensor embedded and fixed into the sheet of a usual EUSFNA needle inserted and locked into the biopsy channel of the EUS scope. The initial calibration (positioning) of the needle-scope assembly relative to the 3D coordinate system was based on several external markers previously fixed, closed to the anatomic region of interest, and also evident on the CT scans. Results: The system was tested initially on a specially designed EUS phantom filled with de-aerated water and silicon inclusions simulating malignant masses, showing small errors (maximum 3 mm) during co-registration of EUS and CT images. Furthermore, the same system was also tested in 6 patients with digestive and lung cancers with good results in decreasing the time of tumor localization and identification as compared with classical EUS procedures. Also, based on the EUS system capabilities, contrast-enhanced power Doppler EUS was also visualized simultaneously with dynamic spiral CT data, allowing an excellent estimation of angiogenesis inside the tumors. Conclusion: Based on the EM tracking of the EUS transducer position and co-registration software with 3D dynamic CT reconstructions, a hybrid system of real-time EUS-CT co-registration was developed. The system should be further tested in larger clinical studies, to describe better the clinical impact of increased diagnostic confidence by direct comparisons between the same lesions based on different imaging modalities, but also to shorten the difficult learning curve of linear EUS. Mo1512 Identification of the Line Demarcating Gastric Cancer From Normal Mucosa by Magnification Endoscopy With NBI Toshihisa Takeuchi, Yuichi Kojima, Yukiko Yoda, Satoshi Tokioka, Eiji Umegaki, Kazuhide Higuchi 2nd Dep of Internal Medicine, Osaka Medical Collage, Takatsuki, Japan [Introduction] Because endoscopic submucosal dissection (ESD) in patients with gastric cancer has become widespread, it is important to accurately identify the demarcation line of gastric cancer. Until now, it has been shown that in qualitative diagnosis of gastric cancer, observation of the surface microstructure and microvascular pattern by magnified endoscopy combined with narrow-band imaging (NBI) enables differentiation of benign and malignant lesions and histological determination of the depressed-type gastric cancer. However, there is no consensus on the identification of a demarcation line of gastric cancer by magnification endoscopy combined with NBI. [Objective] To elucidate the usefulness of magnified endoscopy combined with NBI for identifying a demarcation line of gastric cancer. [Subjects and Method] We included 572 lesions from patients with early gastric cancer who received ESD from 2002 to 2009. We introduced magnified endoscopy combined with NBI from 2006, after which we identified a demarcation line of gastric cancer prior to ESD in 264 lesions and compared the pathological findings with resection samples. We examined the incidence of positive lateral resection margins (i.e., the percentage of inconsistency in the identification of a demarcation line) among the resection samples as well as the factors before and after NBI. [Results] (1) The incidence of positive lateral…
American Journal of Physiology-heart and Circulatory Physiology, Jun 1, 2005
Arteriolar arcades provide alternate pathways for blood flow after obstruction of arteries or art... more Arteriolar arcades provide alternate pathways for blood flow after obstruction of arteries or arterioles such as occurs in stroke and coronary and peripheral vascular disease. When obstruction is prolonged, remaining vessels adjust their diameters chronically in response to altered hemodynamic and metabolic conditions. Here, the effectiveness of arcades in maintaining perfusion both immediately following obstruction and after structural adaptation was examined. Morphometric data from a vascular casting of the pig triceps brachii muscle and published data were used to develop a computational model for the hemodynamics and structural adaptation of the arcade network between two feed artery branches, FA1 and FA2. The predicted total flow to capillaries (Q TA) in the region initially supplied by FA2 decreased to 26% of the normal value immediately after FA2 obstruction but was restored to 78% of the normal value after adaptation. After obstruction of 1-10 randomly selected arcade segments, Q TA was on average 18% higher in the arcade network than in a corresponding two-tree network without arcades. Structural adaptation increased Q TA by an additional 16% in the arcade network but had almost no effect in the two-tree network. These results indicate that arcades can partially maintain blood flow after vascular blockage and that this effect is substantially enhanced by structural adaptation.
At present, deep learning becomes an important tool in medical image analysis, with good performa... more At present, deep learning becomes an important tool in medical image analysis, with good performance in diagnosing, pattern detection, and segmentation. Ultrasound imaging offers an easy and rapid method to detect and diagnose thyroid disorders. With the help of a computer-aided diagnosis (CAD) system based on deep learning, we have the possibility of real-time and non-invasive diagnosing of thyroidal US images. This paper proposed a study based on deep learning with transfer learning for differentiating the thyroidal ultrasound images using image pixels and diagnosis labels as inputs. We trained, assessed, and compared two pre-trained models (VGG-19 and Inception v3) using a dataset of ultrasound images consisting of 2 types of thyroid ultrasound images: autoimmune and normal. The training dataset consisted of 615 thyroid ultrasound images, from which 415 images were diagnosed as autoimmune, and 200 images as normal. The models were assessed using a dataset of 120 images, from which 80 images were diagnosed as autoimmune, and 40 images diagnosed as normal. The two deep learning models obtained very good results, as follows: the pre-trained VGG-19 model obtained 98.60% for the overall test accuracy with an overall specificity of 98.94% and overall sensitivity of 97.97%, while the Inception v3 model obtained 96.4% for the overall test accuracy with an overall specificity of 95.58% and overall sensitivity of 95.58.
Uploads
Papers by Lucian Gruionu