An Automated Method for Classifying Liver Lesions in Contrast-Enhanced Ultrasound Imaging Based on Deep Learning Algorithms.

BACKGROUND: Contrast-enhanced ultrasound (CEUS) is an important imaging modality in the diagnosis of liver tumors. By using contrast agent, a more detailed image is obtained. Time-intensity curves (TIC) can be extracted using a specialized software, and then the signal can be analyzed for further investigations. METHODS: The purpose of the study was to build an automated method for extracting TICs and classifying liver lesions in CEUS liver investigations. The cohort contained 50 anonymized video investigations from 49 patients. Besides the CEUS investigations, clinical data from the patients were provided. A method comprising three modules was proposed. The first module, a lesion segmentation deep learning (DL) model, handled the prediction of masks frame-by-frame (region of interest). The second module performed dilation on the mask, and after applying colormap to the image, it extracted the TIC and the parameters from the TIC (area under the curve, time to peak, mean transit time, and maximum intensity). The third module, a feed-forward neural network, predicted the final diagnosis. It was trained on the TIC parameters extracted by the second model, together with other data: gender, age, hepatitis history, and cirrhosis history. RESULTS: For the feed-forward classifier, five classes were chosen: hepatocarcinoma, metastasis, other malignant lesions, hemangioma, and other benign lesions. Being a multiclass classifier, appropriate performance metrics were observed: categorical accuracy, F1 micro, F1 macro, and Matthews correlation coefficient. The results showed that due to class imbalance, in some cases, the classifier was not able to predict with high accuracy a specific lesion from the minority classes. However, on the majority classes, the classifier can predict the lesion type with high accuracy. CONCLUSIONS: The main goal of the study was to develop an automated method of classifying liver lesions in CEUS video investigations. Being modular, the system can be a useful tool for gastroenterologists or medical students: either as a second opinion system or a tool to automatically extract TICs.

Similarity Analysis for Medical Images Using Color and Texture Histogramss.

Medical databases usually contain a significant volume of images, therefore search engines based on low-level features frequently used to retrieve similar images are necessary for a fast operation. Color, texture, and shape are the most common features used to characterize an image, however extracting the proper features for image retrievals in a similar manner with the human cognition remains a constant challenge. These algorithms work by sorting the images based on a similarity index that defines how different two or more images are, and histograms are one of the most employed methods for image comparison. In this paper, we have extended the concept of image database to the set of frames acquired following wireless capsule endoscopy (from a unique patient). Then, we have used color and texture histograms to identify very similar images (considered duplicates) and removed one of them for each pair of two successive frames. The volume reduction represented an average of 20% from the initial data set, only by removing frames with very similar informational content.

Deep Learning Algorithms in the Automatic Segmentation of Liver Lesions in Ultrasound Investigations.

BACKGROUND: The ultrasound is one of the most used medical imaging investigations worldwide. It is non-invasive and effective in assessing liver tumors or other types of parenchymal changes. METHODS: The aim of the study was to build a deep learning model for image segmentation in ultrasound video investigations. The dataset used in the study was provided by the University of Medicine and Pharmacy Craiova, Romania and contained 50 video examinations from 49 patients. The mean age of the patients in the cohort was 69.57. Regarding presence of a subjacent liver disease, 36.73% had liver cirrhosis and 16.32% had chronic viral hepatitis (5 patients: chronic hepatitis C and 3 patients: chronic hepatitis B). Frames were extracted and cropped from each examination and an expert gastroenterologist labelled the lesions in each frame. After labelling, the labels were exported as binary images. A deep learning segmentation model (U-Net) was trained with focal Tversky loss as a loss function. Two models were obtained with two different sets of parameters for the loss function. The performance metrics observed were intersection over union and recall and precision. RESULTS: Analyzing the intersection over union metric, the first segmentation model obtained performed better compared to the second model: 0.8392 (model 1) vs. 0.7990 (model 2). The inference time for both models was between 32.15 milliseconds and 77.59 milliseconds. CONCLUSIONS: Two segmentation models were obtained in the study. The models performed similarly during training and validation. However, one model was trained to focus on hard-to-predict labels. The proposed segmentation models can represent a first step in automatically extracting time-intensity curves from CEUS examinations.

Clinical impact of wireless capsule endoscopy for small bowel investigation (Review).

Wireless capsule endoscopy is currently considered the gold standard in the investigation of the small bowel. It is both practical for physicians and easily accepted by patients. Prior to its development, two types of imaging investigations of the small bowel were available: radiologic and endoscopic. The first category is less invasive and comfortable for patients; it presents the ensemble of the small bowel, but it may imply radiation exposure. Images are constructed based on signals emitted by various equipment and require special interpretation. Endoscopic techniques provide real-time colored images acquired by miniature cameras from inside the small bowel, require interpretation only from a medical point of view, may allow the possibility to perform biopsies, but the investigation only covers a part of the small bowel and are more difficult to accept by patients. Wireless capsule endoscopy is the current solution that overcomes a part of the previous drawbacks: it covers the entire small bowel, it provides real-time images acquired by cameras, it is painless for patients, and it represents an abundant source of information for physicians. Yet, it lacks motion control and the possibility to perform biopsies or administer drugs. However, significant effort has been oriented in these directions by technical and medical teams, and more advanced capsules will surely be available in the following years.

Ecological Momentary Assessment of Masseter Muscle Activity in Patients with Bruxism.

According to the International Bruxism Consensus, bruxism refers to the activity of the masticatory muscles reflecting contraction disorders, regardless of whether it is during sleep (SB) or an awake (AB) state. The objective of the present study was to evaluate the activity of the masseter muscle by surface electromyographic (sEMG) recordings. This study was performed on 20 participants with self-reported "possible bruxism" (study group) and 20 participants with no self-reported bruxism (control group); all participants underwent an evaluation of the masseter muscle activity using the dia-BRUXO device, which provides numerical parameters regarding sEMG (the total duration and the type of bruxism specific events, the effort made by the masticatory muscles during the recording period, and the personal bruxism index of each participant). Participants from the study group presented more clenching events during AB, three times more frequent than the control group (p = 0.002, Mann−Whitney U test); for SB, the frequency of clenching and grinding events was comparable within the study group, being more frequent than for the control group; the mean value of the effort index was higher for AB (1.177%) than SB (0.470%) and the same for the duration index, with a mean value of 2.788% for AB and 1.054% for SB. All participants from the control group presented reduced values for all acquired parameters. Overall, the personal bruxism index in AB was approximately four times higher for the study group (2.251%) compared to the control group (0.585%) (p < 0.005, Mann−Whitney U test). Similar values were obtained for SB. All participants with "possible bruxism" from the study group presented a higher activity of the masseter muscle, which is specific for bruxism, thus being defined as "definite bruxism".

A computer-aided diagnostic system for intestinal polyps identified by wireless capsule endoscopy.

Small bowel polyps present in images acquired by wireless capsule endoscopy are more difficult to detect using computer-aided diagnostic (CAD) systems. We aimed to identify the optimum morphological characteristics that best describe a polyp and convert them into feature vectors used for automatic detection of polyps present in images acquired by wireless capsule endoscopy (WCE). We prospectively included 54 patients with clinical indications for WCE. Initially, physicians analyzed all images acquired, identifying the frames that contained small bowel polyps. Subsequently, all images were analyzed using an automated computer-aided diagnostic system designed and implemented to convert physical characteristics into vectors of numeric values. The data set was completed with texture and color information, and then analyzed by a feed forward back propagation artificial neural network (ANN) trained to identify the presence of polyps in WCE frames. Overall, the neural network had 93.75% sensitivity, 91.38% specificity, 85.71% positive predictive value (PPV) and 96.36% negative predictive value (NPV). In comparison, physicians' diagnosis indicated 94.79% sensitivity, 93.68% specificity, 89.22% PPV and 97.02% NPV, thus showing that ANN diagnosis was similar to that of human interpretation. Computer-aided diagnostic of small bowel polyps, based on morphological features detection methods, emulation and neural networks classification, seems efficient, fast and reliable for physicians.

Interrelations between hepatic stellate cells and immune system cells in patients with hepatocellular carcinoma.

OBJECTIVE: Our aim was to identify potential correlations between activated hepatic stellate cells (HSCs) and immune system's cells in patients with viral C hepatocellular carcinoma, by quantifying the percentage of activated HSCs, T-lymphocytes, natural killer cells and B-lymphocytes, in three distinct regions: tumor, transition area and the vicinity tissue (2-5 mm). PATIENTS AND METHODS: We prospectively included 20 samples prelevated at necropsy from patients with HCC and C viral infection. We assessed the percentage of alpha-smooth muscle actin (α-SMA), CD45RO, NK1 and CD20 expression using immunohistochemistry and a semi-quantitative scoring method. RESULTS: We found an inverse correlation between the number of α-SMA-positive HSCs and the number of NK1-positive cells in tumor (p=0.0007), in the transition area/tumor capsule (p=0.024) and in the vicinity tissue (p=0.038). Regarding T-lymphocytes, we have also identified an inverse correlation with the number of α-SMA-positive HSCs in tumor (p=0.0036), in the transition area/tumor capsule (p=0.034) and in the vicinity tissue (p=0.047). We found no correlation between the number of activated HSCs and the number of CD20-positive cells in all three examined areas. CONCLUSIONS: The analysis of HSCs activity within specified areas of tumoral liver tissue may lead to new perspectives in early diagnosis of relapses and in the development of future neoadjuvant therapies.

Role of intrahepatic innervation in regulating the activity of liver cells.

Liver innervation comprises sympathetic, parasympathetic and peptidergic nerve fibers, organized as either afferent or efferent nerves with different origins and roles. Their anatomy and physiology have been studied in the past 30 years, with different results published over time. Hepatocytes are the main cell population of the liver, making up almost 80% of the total liver volume. The interaction between hepatocytes and nerve fibers is accomplished through a wealth of neurotransmitters and signaling pathways. In this short review, we have taken the task of condensing the most important data related to how the nervous system interacts with the liver and especially with the hepatocyte population, how it influences their metabolism and functions, and how different receptors and transmitters are involved in this complex process.

Cirrhosis and Chronic Viral Hepatitis as Risk Factors for Hepatocellular Carcinoma: Romanian Single-clinic Experience.

INTRODUCTION: Hepatocellular carcinoma (HCC) represents the fifth most common cancer worldwide, while being the third leading cause of death by cancer. The primary risk factor for HCC seems to be liver cirrhosis. A large majority of these patients have a history of viral hepatitis. MATERIALS AND METHODS: We selected a study lot consisting of 244 patients diagnosed with HCC, admitted between 2006 and October 2009 in the Emergency County Hospital of Craiova, Romania along with an age and gender matched control group, consisting of patients with no history of HCC or other malignancies. We interviewed all subjects regarding their alcohol consumption and background environment. All subjects underwent hepatitis B surface antigen (Hbs Ag) and anti-HCV antibodies (Anti-HCV Atb) serological determinations. RESULTS: The study group consisted of 148 males and 96 females. Liver cirrhosis (LC) was present in 84% of the study lot, 10% associated viral B hepatitis (HBV) and 6% viral C hepatitis (HCV), with no signs of LC. We found LC to be an important risk factor for HCC (RR 6.53, CI 95% 3.18-13.38). The RR and 95% CI of HCC were 4.51 (2.48-8.21) for HbsAg positivity. We noticed a strong correlation (Chi-square test, p<0.001) between the rural environment and the association with LC. HVB was also more present in patients coming from rural areas (p< 0.01). Alcohol intake was present in 89% of the whole lot, being more correlated with the presence of LC as well as with HbsAg positivity (RR 9.165, CI 95% 4.43-18.92). CONCLUSION: Cirrhosis proved to be the primary risk factors for HCC. We underline the fact that HCC was found to be directly associated with viral hepatitis, without evident LC. Further studies are needed in order to establish if intensified HCC screening, especially in rural areas, is required in patients with newly diagnosed viral hepatitis. The increased prevalence of HBV infections might encourage HBV vaccinations as an efficient tool to prevent HCC.

Psychosocial stress and liver disease status.

"Psychosocial stress" is an increasingly common concept in the challenging and highly-demanding modern society of today. Organic response to stress implicates two major components of the stress system, namely the hypothalamic-pituitary-adrenal axis and the sympathetic nervous system. Stress is anamnestically reported by patients during the course of disease, usually accompanied by a decline in their overall health status. As the mechanisms involving glucocorticoids and catecholamines have been deciphered, and their actions on immune cell function deeper understood, it has become clear that stress has an impact on hepatic inflammatory response. An increasing number of articles have approached the link between psychosocial stress and the negative evolution of hepatic diseases. This article reviews a number of studies on both human populations and animal models performed in recent years, all linking stress, mainly of psychosocial nature, and the evolution of three important liver-related pathological entities: viral hepatitis, cirrhosis and hepatocellular carcinoma.