Region Of Interest Segmentation Research Articles

Application of Multilevel Thresholding and CNN for the Diagnosis of Skin Cancer Utilizing a Multi-Agent Fuzzy Buzzard Algorithm

This paper presents a novel methodology for skin cancer diagnosis based on deep learning and metaheuristics. The method combines a multi-level optimal thresholding segmentation technique based on the histogram of the input images with a new improved metaheuristic algorithm, based on Multi-agent Fuzzy Buzzard Algorithm (MAFBUZO) for this purpose. The MAFBUZO combines local search agents in multi-agent systems with the global search ability of the BUZO algorithm, enabling an appropriate balance of exploitation and exploration steps during optimization. To show the algorithm efficiency, it has been validated with four benchmark functions and its comparison with some other bio-inspired methods shows that using the proposed MAFBUZO, with 16.3459–16, 9.6538, 2.7412, and7.3258e-5 error provides the minimum error values during the minimization which shows its effectiveness to use in our application. After the region of interest segmentation, the images have been injected into an optimum convolutional neural network based on MAFBUZO for feature extraction and the final diagnosis. The technique is evaluated against Dermquest and DermIS datasets, and the obtained results are compared against those provided by other published techniques. The results show that using the proposed method with 0.95, 0.88, 0.94, and 0.93 values for NPV, PPV, Accuracy, and Specificity, provides the best outcomes against the other comparative approaches. The final results show that this method improves the results of the existing techniques in terms of overcoming weaknesses such as being trapped in local optimal points or having raw convergence issues.

Contrast enhancement of region of interest of backlit image for surveillance systems based on multi-illumination fusion

The surveillance images under backlit conditions is a potential and challenging research problem in image processing and computer vision. Low contrast regions of interest of backlit images form an indiscernible part for the real-time surveillance and biometrics applications. However, current state-of-the-art techniques typically offer limited dark-region contrast enhancement, are susceptible to colour distortion. This research work uses region of interest segmentation and a transfer learning-based strategy to improve the contrast of the dark intensity regions in backlit image using multi-illumination mappings, motivated by its usefulness. To generate a diversified collection of illumination-maps and enhance the brightness and contrast of the backlit image, the proposed framework applies the varied set of log and gamma transforms. The experimental findings confirm the assertion that the proposed method provides better performance than state-of-the-art methods both objectively and subjectively.

Automatic bolus tracking in abdominal CT scans with convolutional neural networks.

Bolus tracking can optimize the time delay between contrast injection and diagnostic scan initiation in contrast-enhanced computed tomography (CT), yet the procedure is time-consuming and subject to inter- and intra-operator variances which affect the enhancement levels in diagnostic scans. The objective of the current study is to use artificial intelligence algorithms to fully automate the bolus tracking procedure in contrast-enhanced abdominal CT exams for improved standardization and diagnostic accuracy while providing a simplified imaging workflow. This retrospective study used abdominal CT exams collected under a dedicated Institutional Review Board (IRB). Input data consisted of CT topograms and images with high heterogeneity in terms of anatomy, sex, cancer pathologies, and imaging artifacts acquired with four different CT scanner models. Our method consisted of two sequential steps: (I) automatic locator scan positioning on topograms, and (II) automatic region-of-interest (ROI) positioning within the aorta on locator scans. The task of locator scan positioning is formulated as a regression problem, where the limited amount of annotated data is circumvented using transfer learning. The task of ROI positioning is formulated as a segmentation problem. Our locator scan positioning network offered improved positional consistency compared to a high degree of variance in manual slice positionings, verifying inter-operator variance as a significant source of error. When trained using expert-user ground-truth labels, the locator scan positioning network achieved a sub-centimeter error (9.76±6.78 mm) on a test dataset. The ROI segmentation network achieved a sub-millimeter absolute error (0.99±0.66 mm) on a test dataset. Locator scan positioning networks offer improved positional consistency compared to manual slice positionings and verified inter-operator variance as an important source of error. By significantly reducing operator-related decisions, this method opens opportunities to standardize and simplify the workflow of bolus tracking procedures for contrast-enhanced CT.

A Multiscale Attentional Unet Model for Automatic Segmentation in Medical Ultrasound Images.

Ultrasonography has become an essential part of clinical diagnosis owing to its noninvasive, and real-time nature. To assist diagnosis, automatically segmenting a region of interest (ROI) in ultrasound images is becoming a vital part of computer-aided diagnosis (CAD). However, segmenting ROIs on medical images with relatively low contrast is a challenging task. To better achieve medical ROI segmentation, we propose an efficient module denoted as multiscale attentional convolution (MSAC), utilizing cascaded convolutions and a self-attention approach to concatenate features from various receptive field scales. Then, MSAC-Unet is constructed based on Unet, employing MSAC instead of the standard convolution in each encoder and decoder for segmentation. In this study, two representative types of ultrasound images, one of the thyroid nodules and the other of the brachial plexus nerves, were used to assess the effectiveness of the proposed approach. The best segmentation results from MSAC-Unet were achieved on two thyroid nodule datasets (TND-PUH3 and DDTI) and a brachial plexus nerve dataset (NSD) with Dice coefficients of 0.822, 0.792, and 0.746, respectively. The analysis of segmentation results shows that our MSAC-Unet greatly improves the segmentation accuracy with more reliable ROI edges and boundaries, decreasing the number of erroneously segmented ROIs in ultrasound images.

A modified method for CT radiomics region-of-interest segmentation in adrenal lipid-poor adenomas: a two-institution comparative study.

This study aimed to investigate the application of modified region-of-interest (ROI) segmentation method in unenhanced computed tomography in the radiomics model of adrenal lipid-poor adenoma, and to evaluate the diagnostic performance using an external medical institution data set and select the best ROI segmentation method. The imaging data of 135 lipid-poor adenomas and 102 non-adenomas in medical institution A and 30 lipid-poor adenomas and 43 non-adenomas in medical institution B were retrospectively analyzed, and all cases were pathologically or clinically confirmed. The data of Institution A builds the model, and the data of Institution B verifies the diagnostic performance of the model. Semi-automated ROI segmentation of tumors was performed using uAI software, using maximum area single-slice method (MAX) and full-volume method (ALL), as well as modified single-slice method (MAX_E) and full-volume method (ALL_E) to segment tumors, respectively. The inter-rater correlation coefficients (ICC) was performed to assess the stability of the radiomics features of the four ROI segmentation methods. The area under the curve (AUC) and at least 95% specificity pAUC (Partial AUC) were used as measures of the diagnostic performance of the model. A total of 104 unfiltered radiomics features were extracted using each of the four segmentation methods. In the ROC analysis of the radiomics model, the AUC value of the model constructed by MAX was 0.925, 0.919, and 0.898 on the training set, the internal validation set, and the external validation set, respectively, and the AUC value of MAX_E was 0.937, 0.931, and 0.906, respectively. The AUC value of ALL was 0.929, 0.929, and 0.918, and the AUC value of ALL_E was 0.942, 0.926, and 0.927, respectively. In all samples, the pAUCs of MAX, MAX_E, ALL, and ALL_E were 0.021, 0.025, 0.018, and 0.028, respectively. The diagnostic performance of the radiomics model constructed based on the full-volume method was better than that of the model based on the single-slice method. The model constructed using the ALL_E method had a stronger generalization ability and the highest AUC and pAUC value.

Non-Destructive Method for Estimating Seed Weights from Intact Peanut Pods Using Soft X-ray Imaging

In the U.S., peanut farmers receive premium prices for crops with high seed grades. One component of seed grade is the proportion of seed weight to that of pod hulls and other matter. Seed weight and size are also important traits for food processors. Current methods for evaluating peanut seed grade require the opening of the pod and are time-consuming and labor-intensive. In this study, a non-destructive and efficient method to determine peanut seed weights was investigated. X-ray images of a total of 513 peanut pods from three commercial cultivars, each representing three market types, were taken using a soft X-ray imaging system. The region of interest of each image, the seeds, was extracted two ways, manually and with a differential evolution segmentation algorithm. The comprehensive attenuation index (CAI) value was calculated from the segmented regions of interest. Lastly, linear regression models were established between peanut seed weights and the CAI. The results demonstrated that the X-ray imaging technology, coupled with the differential evolution segmentation algorithm, may be used to estimate seed weights efficiently from intact peanut pods.

AUTOMATIC SEGMENTATION AND BONE ARCHITECTURE ANALYSIS OF SUBCHONDRAL BONE IN THE MURINE PROXIMAL TIBIA: EVALUATION OF AN OPEN-SOURCE PIPELINE

µCT images are commonly analysed to assess changes in bone density and architecture in preclinical murine models. Several platforms provide automated analysis of bone architecture parameters from volumetric regions of interest (ROI). However, segmentation of the regions of subchondral bone to create the volumetric ROIs remains a manual and time-consuming task. This study aimed to develop and evaluate automated pipelines for trabecular bone architecture analysis of mouse proximal tibia subchondral bone.A segmented dataset involving 62 knees (healthy and arthritic) from 10-week male C57BL/6 mice were used to train a U-Net type architecture, with µCT scans (downsampled) input that output segmentation and bone volume density (BV/TV) of the subchondral trabecular bone. Segmentations were upsampled and used in tandem with the original scans (10µ) as input for architecture analysis along with the thresholded trabecular bone. The analysis considered the manually and U-Net segmented ROIs using two available pipelines: the ITKBoneMorphometry library and CTan (SKYSCAN). The analyses included: bone volume (BV), total volume (TV), BV/TV, trabecular number (TbN), trabecular thickness (TbTh), trabecular separation (TbSp), and bone surface density (BSBV).There was good agreement for bone measures between the manual and U-Net pipelines utilizing ITK (R=0.88-0.98) and CTan (R=0.91-0.98). ITK and CTan showed good agreement for BV, TV, BV/TV, TbTh and BSBV (R=0.9-0.98). However, a limited agreement was seen between TbN (R=0.73) and TbSb (R=0.59) due to methodological differences in how spacing is evaluated.This U-Net/ITK pipeline seamlessly automated both segmentation and quantification of the proximal tibia subchondral bone. This automated pipeline allows the analysis of large volumes of data, and its open-source nature may enable the standardization of stereologic analysis of trabecular bone across different research groups.

MP73-12 ARTIFICIAL INTELLIGENCE-BASED CANCER MAPPING TO AID IN PROSTATE CANCER THERAPY

You have accessJournal of UrologyCME1 Apr 2023MP73-12 ARTIFICIAL INTELLIGENCE-BASED CANCER MAPPING TO AID IN PROSTATE CANCER THERAPY Wayne Brisbane, Alan Priester, Sakina Mohammed Mota, Joshua Shubert, Jeremy Bong, James Sayre, Brittany Berry-Pusey, and Shyam Natarajan Wayne BrisbaneWayne Brisbane More articles by this author , Alan PriesterAlan Priester More articles by this author , Sakina Mohammed MotaSakina Mohammed Mota More articles by this author , Joshua ShubertJoshua Shubert More articles by this author , Jeremy BongJeremy Bong More articles by this author , James SayreJames Sayre More articles by this author , Brittany Berry-PuseyBrittany Berry-Pusey More articles by this author , and Shyam NatarajanShyam Natarajan More articles by this author View All Author Informationhttps://doi.org/10.1097/JU.0000000000003341.12AboutPDF ToolsAdd to favoritesDownload CitationsTrack CitationsPermissionsReprints ShareFacebookLinked InTwitterEmail Abstract INTRODUCTION AND OBJECTIVE: Focal therapy (FT) for prostate cancer is gaining prominence as an alternative to whole-gland treatment. FT efficacy relies on predicting disease margins, but their underestimation and patient-specific optimization remain largely unaddressed problems. An artificial intelligence (AI)-based cancer mapping and decision support tool was built using pre-biopsy MRI, targeted biopsy data, and PSA to aid urologists in identifying cancer margins. A reader study was conducted to compare this AI-based software against the standard of care (SOC) in determining the extent of clinically significant prostate cancer (csPCa). METHODS: Seven urologists and three radiologists from five institutions with 2 – 23 years of expertise each evaluated 50 prostatectomy cases (total of 1000 reads) Cases were prospectively eligible for FT, with GG 2-3 csPCa, ≥1 region of interest (ROI), and disease that appeared localized to a single hemisphere or the anterior gland. Each case included T2-weighted MRI, ROI segmentation, and pathology reports with conventional locations. Readers were asked to produce contours on each image that prioritized the inclusion of all csPCa, excluding non-csPCa tissue as a secondary objective. First, readers manually defined margins using all given data (SOC). Then, after≥4 weeks had passed, readers produced AI-assisted margins using custom software (iQuest, Avenda Health, CA) [Figure B]. Margins from each method [Figure C] were evaluated against WM pathology data [Figure D] as ground truth. Statistical tests were performed using generalized estimating equations. RESULTS: AI margins had superior sensitivity (97.4% vs. 38.2%, p<0.0001) to SOC margins in classifying csPCa [Figure A]. AI-assisted margins also had superior balanced accuracy, i.e. (specificity+sensitivity)/2, to SOC margins (84.7% vs. 67.2%, p<0.0001). On average, AI-assisted margins completely encapsulated csPCa in 72.8% of cases, compared to only 1.6% of cases with SOC methods (p<0.0001). Furthermore, the average time spent fell from 3.5 minutes (SOC) to 2.0 minutes (AI-assisted, p<0.0001). CONCLUSIONS: AI-assisted cancer mapping helps address the systematic underestimation of csPCa by SOC methods. This study establishes that AI-assisted margins greatly improve csPCa encapsulation, which could improve oncological efficacy for focal treatments. Source of Funding: Avenda Health © 2023 by American Urological Association Education and Research, Inc.FiguresReferencesRelatedDetails Volume 209Issue Supplement 4April 2023Page: e1039 Advertisement Copyright & Permissions© 2023 by American Urological Association Education and Research, Inc.MetricsAuthor Information Wayne Brisbane More articles by this author Alan Priester More articles by this author Sakina Mohammed Mota More articles by this author Joshua Shubert More articles by this author Jeremy Bong More articles by this author James Sayre More articles by this author Brittany Berry-Pusey More articles by this author Shyam Natarajan More articles by this author Expand All Advertisement PDF downloadLoading ...

English

Globally, Coronary Heart Disease (CHD) is one of the main causes of death. Early detection of CHD can improve patient outcomes and reduce mortality rates. We propose a novel framework for predicting the presence of CHD using a combination of machine learning and image processing techniques. The framework comprises various phases, including analyzing the data, feature selection using ReliefF, 3D CNN-based segmentation, feature extraction by means of transfer learning, feature fusion as well as classification, and Adagrad optimization. The first step of the proposed framework involves analyzing the data to identify patterns and correlations that may be indicative of CHD. Next, ReliefF feature selection is applied to decide on the most relevant features from the sample images. The 3D CNN-based segmentation technique is then used to segment the optic disc and macula, which are important regions for CHD diagnosis. Feature extraction using transfer learning is performed to extract features from the segmented regions of interest. The extracted features are then fused using a feature fusion technique, and a classifier is trained to predict the presence of CHD. Finally, Adagrad optimization is used to optimize the performance of the classifier. Our framework is evaluated on a dataset of sample images collected from patients with and without CHD. The results show that the anticipated framework accomplishes elevated accuracy in predicting the presence of CHD. For the purpose of predicting as well as categorizing the patient with heart disease, we applied various ML classifier algorithms. By using Random Forest, Multilayer Perception, and Gradient Boosted Tree, the suggested model's intensity was quite exciting. Also, it was capable of forecasting symptoms associated with cardiovascular disease in either a particular user with a reasonable degree of accuracy compared to the previously employed classifiers like SVM, etc.

Automated Prediction of Early Recurrence in Advanced Sinonasal Squamous Cell Carcinoma With Deep Learning and Multi-parametric MRI-based Radiomics Nomogram

Preoperative prediction of the recurrence risk in patients with advanced sinonasal squamous cell carcinoma (SNSCC) is critical for individualized treatment. To evaluate the predictive ability of radiomics signature (RS) based on deep learning and multiparametric MRI for the risk of 2-year recurrence in advanced SNSCC. Preoperative MRI datasets were retrospectively collected from 265 SNSCC patients (145 recurrences) who underwent preoperative MRI, including T2-weighted (T2W), contrast-enhanced T1-weighted (T1c) sequences and diffusion-weighted (DW). All patients were divided into 165 training cohort and 70 test cohort. A deep learning segmentation model based on VB-Net was used to segment regions of interest (ROIs) for preoperative MRI and radiomics features were extracted from automatically segmented ROIs. Least absolute shrinkage and selection operator (LASSO) and logistic regression (LR) were applied for feature selection and radiomics score construction. Combined with meaningful clinicopathological predictors, a nomogram was developed and its performance was evaluated. In addition, X-title software was used to divide patients into high-risk or low-risk early relapse (ER) subgroups. Recurrence-free survival probability (RFS) was assessed for each subgroup. The radiomics score, T stage, histological grade and Ki-67 predictors were independent predictors. The segmentation models of T2WI, T1c, and apparent diffusion coefficient (ADC) sequences achieved Dice coefficients of 0.720, 0.727, and 0.756, respectively, in the test cohort. RS-T2, RS-T1c and RS-ADC were derived from single-parameter MRI. RS-Combined (combined with T2WI, T1c, and ADC features) was derived from multiparametric MRI and reached area under curve (AUC) and accuracy of 0.854 (0.749-0.927) and 74.3% (0.624-0.840), respectively, in the test cohort. The calibration curve and decision curve analysis (DCA) illustrate its value in clinical practice. Kaplan-Meier analysis showed that the 2-year RFS rate for low-risk patients was significantly greater than that for high-risk patients in both the training and testing cohorts (p < 0.001). Automated nomograms based on multi-sequence MRI help to predict ER in SNSCC patients preoperatively.

COMPARATIVE STUDY OF HEURISTIC-BASED SUPPORT VECTOR MACHINE AND NEURAL NETWORK FOR THERMOGRAM BREAST CANCER DETECTION WITH ENTROPY FEATURES

Thermography is a noncontact, noninvasive imaging technology that is commonly utilized in the medical profession. As early identification of cancer is critical, the computer-assisted method can enhance the diagnosis rate, curing, and survival of cancer patients. Early diagnosis is one of the major essential steps in decreasing the health and socioeconomic consequences of this condition, given the high cost of therapy and the large prevalence of afflicted people. Mammography is currently the majorly utilized procedure for detecting breast cancer. Yet, owing to the low contrast that occurs from a thick breast, mammography is not advised for young women, and alternate methods must be investigated. This work plans to develop a comparative evaluation of two well-performing heuristic-based expert systems for detecting thermogram breast cancer. The thermogram images are taken from the standard DMR dataset. Then, the given images are transferred to the pre-processing stage. Here, the input thermogram images are accomplished by contrast enhancement and mean filtering. Then the Gradient Vector Flow Snakes (GVFS) model is adopted for breast segmentation, and Optimized Fuzzy [Formula: see text]-Means Clustering (OFCM) is developed for abnormality segmentation. From the segmented region of interest, the entropy-based features are acquired. In the classification phase, the “Heuristic-based Support Vector Machine” (HSVM) and “Heuristic-based Neural Network” (HNN) are introduced, which diagnose the breast cancer-affected images. The modifications on SVM and NN are extended by the Oppositional Improvement-based Tunicate Swarm Algorithm (OI-TSA). Furthermore, the suggested models are compared to the traditional SVM and NN classifiers, as well as other classifiers, to validate their competitive performance. From the results, the better accuracy and precision of the designed OI-TSA–HNN model are found to be 96% and 98.4%, respectively. Therefore, the findings confirm that the offered approach shows effectiveness in thermogram breast cancer detection.

What Is the Linear Accuracy of Regional Voxel-Based Registration for Orthognathic Surgery Landmarks?

While regional voxel-based registration (R-VBR) has been shown to have excellent reproducibility and angular accuracy, there are limited data on the linear accuracy of R-VBR for common orthognathic surgery landmarks, or on whether angular accuracy correlates with linear accuracy. The purpose of this study was to estimate the linear accuracy of R-VBR for several skeletal landmarks commonly used in orthognathic surgical planning, and to measure the correlation between angular and linear discrepancies. This is a retrospective cross-sectional study of consecutive patients treated at a single center with nonsegmental LeFort I and bilateral sagittal split osteotomy surgery from January 2019 to November 2020. Cone beam computed tomography at the preoperative (T0) and immediate postoperative (T1) stages were analyzed to measure the postoperative positional changes of 11 orthognathic landmarks in 4 regions of interest (ROI) using R-VBR performed twice by two examiners. Pairwise correlation analysis and canonical correlation analysis were performed for the angular discrepancies (primary predictor variable) and the linear discrepancies (primary outcome variable) to measure the correlation between the two. In cone beam computed tomography analysis of 28 eligible subjects (16 males, 12 females; mean age 18.9years, range 15 to 25), the mean absolute (MA) angular discrepancies ranged from 0.15° to 0.55°, while the corresponding MA linear discrepancies ranged from 0.05 to 0.41mm. There was a strong correlation between angular and linear discrepancies that was statistically significant (P=.001 to .04, Spearman's rank correlation coefficient 0.38 to 0.87). For nonsegmental LeFort I osteotomies and bilateral sagittal split osteotomy, R-VBR has excellent linear accuracy within a single voxel size (0.3mm) for commonly used orthognathic landmarks in the maxillary and distal mandibular ROI. The MA linear discrepancy for the proximal mandibular segment ROI was greater than a single voxel size, with a maximum of 0.41mm.

Introducing a time-efficient workflow for processing IMU data to identify sport-specific movement patterns

Introduction&#x0D; The use of inertial measurement unit (IMU) has become popular in sports assessment. New IMU devices may make the monitoring process easier; however, their validity and reliability should be established prior to widespread use. IMU devices use a combination of gyroscopic and accelerometer data which allow the derivation of velocity and position vectors by integrating the data over time. Because the process of time integration suffers from time varying biases and noise, the resulting velocity and position vectors are prone to drift after a few seconds. This must be accounted for when processing data from IMUs.&#x0D; Aim&#x0D; Motivated by the variety of approaches to IMU-based human motion tracking, the aim of this paper is to deliver a report of the author’s experience in processing and handling acceleration data from a wearable IMU sensor recorded during resistance training and present a workflow to identify specific movement patterns across different sports.&#x0D; Methods&#x0D; Given acceleration data from a wearable sensor during sports practice, the workflow to derive velocity and position measures of specific movement patterns is divided into the following seven steps: 1) Rough cropping of region of interest (ROI). 2) Application of low pass filter to remove jittering upon visual inspection. Depending on ROI length, a detrend filter should be applied on the integrated position and maybe on the velocity data to correct for drift. 3) Visual analysis of characteristics of at least one movement pattern (more if the pattern shows a high inter-repetition variability) to identify key events (e.g. maximal velocity). 4) Automatically find and count key events along ROI. 5) Reassess characteristics of movement pattern to determine other relevant events. 6) Next, segmentation of ROI based on selected events and integration of individual sections to avoid drift. The aim is to integrate the smallest pieces possible. 7) Finally, check to make sure that segmentation worked correctly (e.g. correct number of repetitions, resulting values in a possible range).&#x0D; Results&#x0D; Acceleration data was captured with an Apple Watch 7 (Apple Inc. California) using the SensorLog app streaming to a customized node.js server application. For the processing and visualization of the data, the programming language Python with usage of the Pandas and SciPy libraries were utilized. The velocity and position data were determined by finding the integral of the acceleration and velocity respectively. Using the previous mentioned workflow 306 repetitions of the back squat executed by 11 recreational athletes (w: 5/m: 6, age: 22-37, weight:58-90kg) were successful segmented.&#x0D; Discussion&#x0D; The technology underlying commercial IMU sensors are often not communicated transparently. Thus, it is important to properly study the task of calibration the IMU and calculating the vertical component before using it for sport science measurements. Furthermore, it is rarely the case that the movement pattern remains the same over each training session. Therefore, characteristics of the movement pattern must be studied thoroughly to create a robust identification criterium.&#x0D; By applying the presented workflow researcher have a structured, easy to apply and time efficient approach to analyze recorded acceleration data on different sport-specific movement patterns.

Prediction of microvascular invasion in hepatocellular carcinoma based on preoperative Gd-EOB-DTPA-enhanced MRI: Comparison of predictive performance among 2D, 2D-expansion and 3D deep learning models.

To evaluate and compare the predictive performance of different deep learning models using gadolinium ethoxybenzyl diethylenetriamine pentaacetic acid (Gd-EOB-DTPA)-enhanced MRI in predicting microvascular invasion (MVI) in hepatocellular carcinoma. The data of 233 patients with pathologically confirmed hepatocellular carcinoma (HCC) treated at our hospital from June 2016 to June 2021 were retrospectively analyzed. Three deep learning models were constructed based on three different delineate methods of the region of interest (ROI) using the Darwin Scientific Research Platform (Beijing Yizhun Intelligent Technology Co., Ltd., China). Manual segmentation of ROI was performed on the T1-weighted axial Hepatobiliary phase images. According to the ratio of 7:3, the samples were divided into a training set (N=163) and a validation set (N=70). The receiver operating characteristic (ROC) curve was used to evaluate the predictive performance of three models, and their sensitivity, specificity and accuracy were assessed. Among 233 HCC patients, 109 were pathologically MVI positive, including 91 men and 18 women, with an average age of 58.20 ± 10.17 years; 124 patients were MVI negative, including 93 men and 31 women, with an average age of 58.26 ± 10.20 years. Among three deep learning models, 2D-expansion-DL model and 3D-DL model showed relatively good performance, the AUC value were 0.70 (P=0.003) (95% CI 0.57-0.82) and 0.72 (P<0.001) (95% CI 0.60-0.84), respectively. In the 2D-expansion-DL model, the accuracy, sensitivity and specificity were 0.7143, 0.739 and 0.688. In the 3D-DL model, the accuracy, sensitivity and specificity were 0.6714, 0.800 and 0.575, respectively. Compared with the 3D-DL model (based on 3D-ResNet), the 2D-DL model is smaller in scale and runs faster. The frames per second (FPS) for the 2D-DL model is 244.7566, which is much larger than that of the 3D-DL model (73.3374). The deep learning model based on Gd-EOB-DTPA-enhanced MRI could preoperatively evaluate MVI in HCC. Considering that the predictive performance of 2D-expansion-DL model was almost the same as the 3D-DL model and the former was relatively easy to implement, we prefer the 2D-expansion-DL model in practical research.

Automatic ROI recognition and parameters selection for digital image correlation in measuring structures with complex shapes

For the digital image correlation (DIC) method, the measurement of specimens with complex shapes may encounter difficulties due to the time-consuming recognition of region of interest (ROI), and the indeterminate parameter selection caused by the non-uniform deformation. This paper proposes an automatic DIC for the measurement of structures with complex shapes. An automatic ROI segmentation is developed by combining a convolutional neural network and image morphology, so the boundary of the specimen can be acquired accurately and efficiently. In dealing with the non-uniform deformation, a strain-related automatic selection of DIC parameters is developed, in which the sampling intervals and the subset sizes at different areas can be automatically determined. Both results of the simulated experiment and real experiment show that, by combing the two approaches with segmentation-aided DIC, the proposed automatic DIC can characterize the complex deformation including the boundary of the structures effectively.

Dual-Channel Capsule Generative Adversarial Network Optimized with Golden Eagle Optimization for Pediatric Bone Age Assessment from Hand X-Ray Image

Bone age assessment (BAA) is mainly utilized for detecting the growth of pediatrics because a large number of bone diseases occur at young age. Several algorithms related to BAAs were used for detecting the maturity of bones, but it does not provide sufficient accuracy, and also increased the error rate. To deal with these problems, the dual-channel capsule generative adversarial network optimized with Golden eagle optimization (GEO) is proposed in this paper for pediatric BAA from hand X-ray image (DCCGAN-GEO-BAA-HX-ray). Initially, the input hand X-ray imageries are collected from the dataset of Radiological Society of North America (RSNA) pediatric bone age (BA). Then, region of interest (RoI) of input hand X-ray imageries is segmented based on Tsallis entropy-based multilevel 3D Otsu thresholding (TE-3D-Otsu). Here, TE-3D-Otsu method segments the RoI region of wrist, thumb, middle finger, little finger, which enhance the classification accuracy. Moreover, the segmented RoI is given to DCCGAN that predicts the BAA. Generally, the DCCGAN does not reveal any adoption of optimization methods to scale the optimum parameters to ensure accurate classification. Therefore, GEO is used for optimizing the weight parameters of DCCGAN. The proposed DCCGAN-GEO-BAA-HX-ray method is executed in MATLAB and its performance is examined under performance metrics such as accuracy, precision, sensitivity, F-scores, specificity, concordance correlation coefficient (CCC) and computational time. Finally, the proposed DCCGAN-GEO-BAA-HX-ray approach attains 14.68%, 7.142%, 9.23% and 4.65% higher accuracy, 38.18%, 12.02%, 11.56% and 7.59% lower computational time is compared with existing FRCNN-AF-SFO-BAA-HX-ray, DCNN-W-CTO-BAA-HX-ray, CNN-MLP-BAA-HX-ray and CNN-BAA-HX-ray methods.

Open-source pipeline for automatic segmentation and microstructural analysis of murine knee subchondral bone.

A segmented dataset of μCT scans from 62 knees (healthy and arthritic) from 10-week male C57BL/6 mice was used to train a U-Net type architecture to automate segmentation of the subchondral trabecular bone. These segmentations were used in tandem with the original scans as input for microstructural analysis along with thresholded trabecular bone. Manually and U-Net segmented ROIs were fed into two available pipelines for microstructural analysis: the ITKBoneMorphometry library and CTan (SKYSCAN). Outcome parameters were compared between pipelines, including: bone volume (BV), total volume (TV), BV/TV, trabecular number (TbN), trabecular thickness (TbTh), trabecular separation (TbSp), and bone surface density (BSBV). There was good agreement for all bone measures comparing the manual and U-Net pipelines utilizing ITK (R = 0.88-0.98) and CTAn (R = 0.91-0.98). ITK and CTAn showed good agreement for BV, TV, BV/TV, TbTh and BSBV (R = 0.9-0.98). However, limited agreement was seen between TbN (R = 0.73) and TbSb (R = 0.59) due to methodological differences in how spacing is evaluated. Microstructural parameters generated from manual and automatic segmentations showed high correlation across all measures. Using the CTAn pipeline yielded strong R2 values (0.83-0.96) and very strong agreement based on ICC (0.90-0.98). The ITK pipeline yielded similarly high R2 values (0.91-0.96, except for TbN (0.77)), and ICC values (0.88-0.98). The automated segmentations yield lower average values for BV, TV and BV/TV (ranging from 14 % to 6.3 %), but differences were not found to be influenced by the mean ROI values. This integrated pipeline seamlessly automated both segmentation and quantification of the proximal tibia subchondral bone microstructure. This automated pipeline allows the analysis of large volumes of data, and its open-source nature may enable the standardization of microstructural analysis of trabecular bone across different research groups.

Automatic treatment outcome prediction with DeepInteg based on multimodal radiological images in rectal cancer

Neoadjuvant systemic treatment before surgery is a prevalent regimen in the patients with advanced-stage or high-risk tumor, which has shaped the treatment strategies and cancer survival in the past decades. However, some patients present with poor response to the neoadjuvant treatment. Therefore, it is of great significance to develop tools to help distinguish the patients that could achieve pathological complete response before surgery to avoid inappropriate treatment. Here, this study demonstrated a multi-task deep learning tool called DeepInteg. In the DeepInteg framework, the segmentation module was constructed based on the CE-Net with a context extractor to achieve end-to-end delineation of region of interest (ROI) from radiological images, then the features of segmented Magnetic Resonance Imaging (MRI) and Computed Tomography (CT) images of each case were fused and input to the classification module based on a convolution neural network for treatment outcome prediction. The dataset with 1700 MRI and CT slices collected from the prospectively randomized clinical trial (NCT01211210) on systemic treatment for rectal cancer was used to develop and systematically optimize DeepInteg. As a result, DeepInteg achieved automatic segmentation of tumoral ROI with Dices of 0.766 and 0.719 and mIoUs of 0.788 and 0.756 in CT and MRI images, respectively. In addition, DeepInteg achieved AUC of 0.833, accuracy of 0.826 and specificity of 0.856 in the prediction for pathological complete response after treatment, which showed better performance compared with the model based on CT or MRI alone. This study provide a robust framework to develop disease-specific tools for automatic delineation of ROI and clinical outcome prediction. The well-trained DeepInteg could be readily applied in clinic to predict pathological complete response after neoadjuvant therapy in rectal cancer patients.

Distinguishing common renal cell carcinomas from benign renal tumors based on machine learning: comparing various CT imaging phases, slices, tumor sizes, and ROI segmentation strategies.

To determine whether a CT-based machine learning (ML) can differentiate benign renal tumors from renal cell carcinomas (RCCs) and improve radiologists' diagnostic performance, and evaluate the impact of variable CT imaging phases, slices, tumor sizes, and region of interest (ROI) segmentation strategies. Patients with pathologically proven RCCs and benign renal tumors from our institution between 2008 and 2020 were included as the training dataset for ML model development and internal validation (including 418 RCCs and 78 benign tumors), and patients from two independent institutions and a public database (TCIA) were included as the external dataset for individual testing (including 262 RCCs and 47 benign tumors). Features were extracted from three-phase CT images. CatBoost was used for feature selection and ML model establishment. The area under the receiver operating characteristic curve (AUC) was used to assess the performance of the ML model. The ML model based on 3D images performed better than that based on 2D images, with the highest AUC of 0.81 and accuracy (ACC) of 0.86. All three radiologists achieved better performance by referring to the classifier's decision, with accuracies increasing from 0.82 to 0.87, 0.82 to 0.88, and 0.76 to 0.87. The ML model achieved higher negative predictive values (NPV, 0.82-0.99), and the radiologists achieved higher positive predictive values (PPV, 0.91-0.95). A ML classifier based on whole-tumor three-phase CT images can be a useful and promising tool for differentiating RCCs from benign renal tumors. The ML model also perfectly complements radiologist interpretations. • A machine learning classifier based on CT images could be a reliable way to differentiate RCCs from benign renal tumors. • The machine learning model perfectly complemented the radiologists' interpretations. • Subtle variances in ROI delineation had little effect on the performance of the ML classifier.

A radiographic, deep transfer learning framework, adapted to estimate lung opacities from chest x-rays

Chest radiographs (CXRs) are the most widely available radiographic imaging modality used to detect respiratory diseases that result in lung opacities. CXR reports often use non-standardized language that result in subjective, qualitative, and non-reproducible opacity estimates. Our goal was to develop a robust deep transfer learning framework and adapt it to estimate the degree of lung opacity from CXRs. Following CXR data selection based on exclusion criteria, segmentation schemes were used for ROI (Region Of Interest) extraction, and all combinations of segmentation, data balancing, and classification methods were tested to pick the top performing models. Multifold cross validation was used to determine the best model from the initial selected top models, based on appropriate performance metrics, as well as a novel Macro-Averaged Heatmap Concordance Score (MA HCS). Performance of the best model is compared against that of expert physician annotators, and heatmaps were produced. Finally, model performance sensitivity analysis across patient populations of interest was performed. The proposed framework was adapted to the specific use case of estimation of degree of CXR lung opacity using ordinal multiclass classification. Acquired between March 24, 2020, and May 22, 2020, 38,365 prospectively annotated CXRs from 17,418 patients were used. We tested three neural network architectures (ResNet-50, VGG-16, and ChexNet), three segmentation schemes (no segmentation, lung segmentation, and lateral segmentation based on spine detection), and three data balancing strategies (undersampling, double-stage sampling, and synthetic minority oversampling) using 38,079 CXR images for training, and validation with 286 images as the out-of-the-box dataset that underwent expert radiologist adjudication. Based on the results of these experiments, the ResNet-50 model with undersampling and no ROI segmentation is recommended for lung opacity classification, based on optimal values for the MAE metric and HCS (Heatmap Concordance Score). The degree of agreement between the opacity scores predicted by this model with respect to the two sets of radiologist scores (OR or Original Reader and OOBTR or Out Of Box Reader) in terms of performance metrics is superior to the inter-radiologist opacity score agreement.