Current Medical Imaging - Volume 16, Issue 3, 2020

Background: Measuring cornea thickness is an essential parameter for patients undergoing refractive Laser-Assisted in SItu Keratomileusis (LASIK) surgeries. Discussion: This paper describes about the various available imaging and non-imaging methods for identifying cornea thickness and explores the most optimal method for measuring it. Along with the thickness measurement, layer segmentation in the cornea is also an essential parameter for diagnosing and treating eye-related disease and problems. The evaluation supports surgical planning and estimation of the corneal health. After surgery, the thickness estimation and layer segmentation are also necessary for identifying the layer surface disorders. Conclusion: Hence the paper reviews the available image processing techniques for processing the corneal image for thickness measurement and layer segmentation.

Purpose: The study aims to investigate the specificity, sensitivity, and accuracy of nuclear medicine, CT scan, and ultrasonography to diagnose the disorders related to the thyroid gland. Methodology: The study is based on the retrospective approach of recruiting 52 patients suffering from thyroid disorders. The demographic details of each patient have been recorded. Moreover, the results of previously conducted nuclear medicine scan, CT scan, and ultrasound have also been assessed. The findings of all the tests have been compared to evaluate and compare their sensitivity, specificity, and accuracy. Results: A total of 52 patients were recruited for the study among which 41 were female and 11 were males. The findings of SPECT and MRI were compared, which revealed that MRI possessed 38.8% sensitivity and 22.22% specificity. As compared to the findings of CT scan, increased specificity (71.42%) and sensitivity (70.96%) have been identified in MRI. Conclusion: There is an increase in the sensitivity and specificity of MRI outcomes as compared to the nuclear medicine and CT scan.

Background: Medical imaging plays an important role in the diagnosis of thyroid diseases. In the field of machine learning, multiple dimensional deep learning algorithms are widely used in image classification and recognition, and have achieved great success. Objective: The method based on multiple dimensional deep learning is employed for the auxiliary diagnosis of thyroid diseases based on SPECT images. The performances of different deep learning models are evaluated and compared. Methods: Thyroid SPECT images are collected with three types, they are hyperthyroidism, normal and hypothyroidism. In the pre-processing, the region of interest of thyroid is segmented and the amount of data sample is expanded. Four CNN models, including CNN, Inception, VGG16 and RNN, are used to evaluate deep learning methods. Results: Deep learning based methods have good classification performance, the accuracy is 92.9%-96.2%, AUC is 97.8%-99.6%. VGG16 model has the best performance, the accuracy is 96.2% and AUC is 99.6%. Especially, the VGG16 model with a changing learning rate works best. Conclusion: The standard CNN, Inception, VGG16, and RNN four deep learning models are efficient for the classification of thyroid diseases with SPECT images. The accuracy of the assisted diagnostic method based on deep learning is higher than that of other methods reported in the literature.

Background: Image evaluation of scar tissue plays a significant role in the diagnosis of cardiovascular diseases. Segmentation of the scar tissue is the first step towards evaluating the morphology of the scar tissue. Then, with the use of CT images, the deep learning approach can be applied to identify possible scar tissue in the left ventricular endocardial wall. Objectives: To develop an automated method for detecting the endocardial scar tissue in the left ventricular using Deep learning approach. Methods: Pixel values of the endocardial wall for each image in the sequence were extracted. Morphological operations, including defining regions of the endocardial wall of the LV where scar tissue could predominate, were performed. Convolutional Neural Networks (CNN) is a deep learning application, which allowed choosing appropriate features from delayed enhancement cardiac CT images to distinguish between endocardial scar and healthy tissues of the LV by applying pixel value-based concepts. Results: We achieved 89.23% accuracy, 91.11% sensitivity, and 87.75% specificity in the detection of endocardial scars using the CNN-based method. Conclusion: Our findings reveal that the CNN-based method yielded robust accuracies in LV endocardial scar detection, which is currently the most extensively used pixel-based method of deep learning. This study provides a new direction for the assessment of scar tissue in imaging modalities and provides a potential avenue for clinical adaptations of these algorithms. Additionally this methodology, in comparison with those in the literature, provides specific advantages in its translational ability to clinical use.

Background: Left ventricular diastolic dysfunction (LVDD) is a common abnormality among patients in T2DM. Aims: We aimed to evaluate the feasibility of coronary computed tomography angiography (CCTA) for the assessment of LVDD in type 2 diabetes mellitus (T2DM) patients. Methods: 80 consecutive T2DM patients who were referred for a clinically dual-source CCTA examination to evaluate suspected coronary artery disease and also underwent 2D echocardiography within 7 days of CCTA inclusion and exclusion criteria, were performed. Correlation between CCTA and echocardiography was tested through linear regression and Bland-Altman analysis. Results: In total, 60 T2DM patients were included for the analysis. Pearson correlation showed good correlation for E (r = 0.28; P = 0.028), E/A (r = 0.69; P < 0.01); E (r = -0.06; P = 0.776), E/A (r = 0.54; P = 0.003) and E (r = 0.64; P < 0.01), E/A (r = 0.83; P < 0.01) in three groups, respectively. Overall, diagnostic accuracy for assessment in CCTA of diastolic dysfunction was 79.76% (95% CI: 68%-91%), 71.43% (95% CI: 58%-85%) and 87.50 (95% CI: 79%-96%) in three groups. Conclusion: The presented study proved that CCTA showed good correlations in the estimation of LV filling pressures compared with echocardiography in T2DM patients. Accordingly, retrospectively ECG-gated CCTA may provide valuable information on the evaluation of LVDD in T2DM patients.

Background: There are a few studies about the evaluation of maxillary first premolars internal structure with micro-computed tomography (micro-CT). The aim of this study was to assess morphological features of the pulp chamber in maxillary first premolar teeth using micro- CT. Methods: Extracted 15 maxillary first premolar teeth were selected from the patients who were in different age groups. The distance between the pulp orifices, the diameter of the pulp and the width of the pulp chamber floor were measured on the micro-CT images with the slice thickness of 13.6 μm. The number of root canal orifices and the presence of isthmus were evaluated. Results: The mean diameter of orifices was 0.73 mm on the buccal side while it was 0.61 mm on palatinal side. The mean distance between pulp orifices was 2.84 mm. The mean angle between pulp orifices was -21.53°. The mean height of pulp orifices on the buccal side was 4.32 mm while the mean height of pulp orifices on the palatinal side was 3.56 mm. The most observed shape of root canal orifices was flattened ribbon. No isthmus was found in specimens. Conclusion: Minor anatomical structures can be evaluated in more detail with micro-CT. The observation of the pulp cavity was analyzed using micro-CT.

Background: Automatic approach to vertebrae segmentation from computed tomography (CT) images is very important in clinical applications. As the intricate appearance and variable architecture of vertebrae across the population, cognate constructions in close vicinity, pathology, and the interconnection between vertebrae and ribs, it is a challenge to propose a 3D automatic vertebrae CT image segmentation method. Objective: The purpose of this study was to propose an automatic multi-vertebrae segmentation method for spinal CT images. Methods: Firstly, CLAHE-Threshold-Expansion was preprocessed to improve image quality and reduce input voxel points. Then, 3D coarse segmentation fully convolutional network and cascaded finely segmentation convolutional neural network were used to complete multi-vertebrae segmentation and classification. Results: The results of this paper were compared with the other methods on the same datasets. Experimental results demonstrated that the Dice similarity coefficient (DSC) in this paper is 94.84%, higher than the V-net and 3D U-net. Conclusion: Method of this paper has certain advantages in automatically and accurately segmenting vertebrae regions of CT images. Due to the easy acquisition of spine CT images. It was proven to be more conducive to clinical application of treatment that uses our segmentation model to obtain vertebrae regions, combining with the subsequent 3D reconstruction and printing work.

Background: Basal-like carcinoma is one of the breast subtypes that lacks expression of the estrogen receptor (ER), progesterone receptor (PR) and human epidermal growth factor receptor 2 (HER2). It has a poor prognosis and aggressive behavior. It is in a heterogeneous group with various other types of cancer, including metaplastic carcinoma, carcinomas with medullary features, medullary carcinoma, adenoid cystic carcinoma, secretory carcinoma, and invasive carcinoma arising in the setting of BRCA1 mutations. Imaging features of basal-like cancers have not been uniform, and there are no studies with imaging comparisons between basal-like carcinomas. Objectives: To compare imaging features of basal-like carcinomas and to understand their characteristics. Methods: By using our radiologic database, we retrospectively searched 37 cases of metaplastic carcinoma and 44 cases of invasive carcinoma with medullary features (ICMF). Two radiologists reviewed images according to ACR BI-RADS lexicon. Results: The higher Ki-67 and absence of calcifications were statistically significant in ICMF than in metaplastic carcinoma. Metaplastic carcinoma demonstrated oval shape and parallel orientation more frequently. ICMF showed more irregular shape and angular margin on ultrasound, irregular or spiculated margin on breast MRI. ICMF showed more delayed washout pattern of enhancement than metaplastic carcinoma. Intratumoral T2, a very high signal was noted more in metaplastic carcinoma. Conclusion: Our study presents variable imaging features observed between basal-like carcinomas. Although it is not sufficient to predict clinical progress, aggressiveness or prognosis of basal-like carcinomas, the results of this study will be helpful in understanding and diagnosing various basallike carcinomas.

Background: Data mining algorithms are extensively used to classify the data, in which prediction of disease using minimal computation time plays a vital role. Objectives: The aim of this paper is to develop the classification model from reduced features and instances. Methods: In this paper we proposed four search algorithms for feature selection the first algorithm is Random Global Optimal (RGO) search algorithm for searching the continuous, global optimal subset of features from the random population. The second is Global and Local Optimal (GLO) search algorithm for searching the global and local optimal subset of features from population. The third one is Random Local Optimal (RLO) search algorithm for generating random, local optimal subset of features from the random population. Finally the Random Global and Optimal (RGLO) search algorithm for searching the continuous, global and local optimal subset of features from the random population. RGLO search algorithm combines the properties of first three stated algorithm. The subsets of features generated from the proposed four search algorithms are evaluated using the consistency based subset evaluation measure. Instance based learning algorithm is applied to the resulting feature dataset to reduce the instances that are redundant or irrelevant for classification. The model developed using naïve Bayesian classifier from the reduced features and instances is validated with the tenfold cross validation. Results: Classification accuracy based on RGLO search algorithm using naïve Bayesian classifier is 94.82% for Breast, 97.4% for DLBCL, 98.83% for SRBCT and 98.89% for Leukemia datasets. Conclusion: The RGLO search based reduced features results in the high prediction rate with less computational time when compared with the complete dataset and other proposed subset generation algorithm.

Background: This study is carried out targeting the problem of slow response time and performance degradation of imaging system caused by large data of medical ultrasonic imaging. In view of the advantages of CS, it is applied to medical ultrasonic imaging to solve the above problems. Objectives: Under the condition of satisfying the speed of ultrasound imaging, the quality of imaging can be further improved to provide the basis for accurate medical diagnosis. Methods: According to CS theory and the characteristics of the array ultrasonic imaging system, block compressed sensing ultrasonic imaging algorithm is proposed based on wavelet sparse representation. Results: Three kinds of observation matrices have been designed on the basis of the proposed algorithm, which can be selected to reduce the number of the linear array channels and the complexity of the ultrasonic imaging system to some extent. Conclusion: The corresponding simulation program is designed, and the result shows that this algorithm can greatly reduce the total data amount required by imaging and the number of data channels required for linear array transducer to receive data. The imaging effect has been greatly improved compared with that of the spatial frequency domain sparse algorithm.