Search

Background: Accurate and early detection of non-alcoholic fatty liver, which is a major cause of chronic diseases is very important and is vital to prevent the complications associated with this disease. Ultrasound of the liver is the most common and widely performed method of diagnosing fatty liver. However, due to the low quality of ultrasound images, the need for an automatic and intelligent classification method based on artificial intelligence methods to accurately detect the amount of liver fat is essential. This paper aims to develop an advanced machine learning model based on texture features to assess liver fat levels based on liver ultrasound images. Methods: In this analytic study, which is done from April to November 2020 in Tehran, ultrasound images of 55 obese people who have undergone laparoscopic surgery have been used and the histological result of a liver biopsy has been employed as a reference for liver fat. First, 88 texture-based features were extracted from the images using the Gray-Level Co-Occurrence Matrix (GLCM) method. In the next step, using the method of minimum redundancy and maximum correlation, the top features were selected from among 88 features and applied to the classifier input. Finally, using the three classifiers of linear discriminant analysis, support vector machine and AdaBoost, the images were classified into 4 groups based on the amount of liver fat. Results: The accuracy of the automatic liver fat prediction model from ultrasound images for AdaBoost classification was 92.72%. However, the accuracies obtained for support vector machine and linear discriminant analysis classification were 87.88% and 75.76%, respectively. Conclusion: The proposed approach based on texture features using the GLCM and the AdaBoost classification from ultrasound images automatically detects the amount of liver fat with high accuracy and can help physicians and radiologists in the final diagnosis.

Background: Surgery and accurate removal of the brain tumor in the operating room and after opening the scalp is one of the major challenges for neurosurgeons due to the removal of skull pressure and displacement and deformation of the brain tissue. This displacement of the brain changes the location of the tumor relative to the MR image taken preoperatively. Methods: This study, which is done from March to December 2019 in Tehran, is evaluated on the available database of RetroSpective Evaluation of Cerebral Tumors (RESECT) including pre-operative MR images, and intra-operative ultrasound from 22 patients with low-grade gliomas who underwent surgeries at St. Olavs University Hospital. This study is used for image registration of preoperative MR imaging and ultrasound imaging after resection of the skull to compensate for brain changes. By this method, we obtained a third image that resembles preoperative MR imaging but has the geometry of the brain shape changes. We used a combination of the two transformations named Affine and non-rigid Free Form Deformation (FFD) for hierarchically moving the pixels to compensate for global variations, and also nonlinear local and small variations. Also, by applying the mutual information function, we consider the entropy value as the criterion of similarity due to the non-similarity of the nature of the images. Also, Limited Broyden-Fletcher-Goldfarb-Shannon method is used for optimization. Results: The results of the proposed method were presented on the available database of RetroSpective Evaluation of Cerebral Tumors (RESECT) including images of 22 patients with glioma type 2 tumors and evaluated based on 15 landmarks per patient and also mutual information criteria. The mean target registration error for affine, FFD and the proposed method are 46.19, 42.85 and 38.01, respectively. It was shown that the proposed method achieved high accuracy by combining the two transformations of affine and FFD compared to the separate use of each of the two models. Conclusion: In image registration of preoperative MR and ultrasound images for compensation of brain shift, the combination of affine and FFD transformations had better results than the individual use of each of the transformations.

Background: Accurate and early detection of non-alcoholic fatty liver, which is a major cause of chronic diseases is very important and is vital to prevent the complications associated with this disease. Ultrasound of the liver is the most common and widely performed method of diagnosing fatty liver. However, due to the low quality of ultrasound images, the need for an automatic and intelligent classification method based on artificial intelligence methods to accurately detect the amount of liver fat is essential. This paper aims to develop an advanced machine learning model based on texture features to assess liver fat levels based on liver ultrasound images. Methods: In this analytic study, which is done from April to November 2020 in Tehran, ultrasound images of 55 obese people who have undergone laparoscopic surgery have been used and the histological result of a liver biopsy has been employed as a reference for liver fat. First, 88 texture-based features were extracted from the images using the Gray-Level Co-Occurrence Matrix (GLCM) method. In the next step, using the method of minimum redundancy and maximum correlation, the top features were selected from among 88 features and applied to the classifier input. Finally, using the three classifiers of linear discriminant analysis, support vector machine and AdaBoost, the images were classified into 4 groups based on the amount of liver fat. Results: The accuracy of the automatic liver fat prediction model from ultrasound images for AdaBoost classification was 92.72%. However, the accuracies obtained for support vector machine and linear discriminant analysis classification were 87.88% and 75.76%, respectively. Conclusion: The proposed approach based on texture features using the GLCM and the AdaBoost classification from ultrasound images automatically detects the amount of liver fat with high accuracy and can help physicians and radiologists in the final diagnosis. [ABSTRACT FROM AUTHOR]

Background: Surgery and accurate removal of the brain tumor in the operating room and after opening the scalp is one of the major challenges for neurosurgeons due to the removal of skull pressure and displacement and deformation of the brain tissue. This displacement of the brain changes the location of the tumor relative to the MR image taken preoperatively. Methods: This study, which is done from March to December 2019 in Tehran, is evaluated on the available database of RetroSpective Evaluation of Cerebral Tumors (RESECT) including pre-operative MR images, and intra-operative ultrasound from 22 patients with low-grade gliomas who underwent surgeries at St. Olavs University Hospital. This study is used for image registration of preoperative MR imaging and ultrasound imaging after resection of the skull to compensate for brain changes. By this method, we obtained a third image that resembles preoperative MR imaging but has the geometry of the brain shape changes. We used a combination of the two transformations named Affine and non-rigid Free Form Deformation (FFD) for hierarchically moving the pixels to compensate for global variations, and also nonlinear local and small variations. Also, by applying the mutual information function, we consider the entropy value as the criterion of similarity due to the non-similarity of the nature of the images. Also, Limited Broyden-Fletcher-Goldfarb-Shannon method is used for optimization. Results: The results of the proposed method were presented on the available database of RetroSpective Evaluation of Cerebral Tumors (RESECT) including images of 22 patients with glioma type 2 tumors and evaluated based on 15 landmarks per patient and also mutual information criteria. The mean target registration error for affine, FFD and the proposed method are 46.19, 42.85 and 38.01, respectively. It was shown that the proposed method achieved high accuracy by combining the two transformations of affine and FFD compared to the separate use of each of the two models. Conclusion: In image registration of preoperative MR and ultrasound images for compensation of brain shift, the combination of affine and FFD transformations had better results than the individual use of each of the transformations. [ABSTRACT FROM AUTHOR]

Background: In the telemedicine process, using digital techniques in disease diagnosis caused to have felt needs of archiving and storing patient information and high bandwidth in data transfer. Methods: This study aimed at introducing an efficient way of multi-stage compression of mammographic image data based LM algorithm and artificial neural networks. At First, data derived from mammographic images given to multi-layer neural network has achieved the possibility of forming with minimum damage and high degree of compaction in the first layer. Results: The compression process of the mammography images was implemented using images of 128 women aged 46.41±6.55 yrs with BMI 36.78 ±5.5 from three specialized clinics in Sabzevar. The analysis yielded a mean square error (MSE) of 4.24 with the highest difference ratio of 33.46 and compression ratio of 8: 1in the output of the algorithm. The system performance based on the accurate design of the software was acceptable therefore; it demonstrated high efficiency in practice. Conclusion: The diagnosis in the discovery stage is highly consistent with the diagnosis in real based on reliability of software output in the compression and release, and considering the fact of mammographic images are not completely degraded during compression; therefore, this system has the capacity to be implemented achieving mammography images in hospitals and justify its application.