Improving automatic segmentation of liver tumor images using a deep learning model.

Liver tumors are one of the most aggressive malignancies in the human body. Computer-aided technology and liver interventional surgery are effective in the prediction, identification and management of liver neoplasms. One of the important processes is to accurately grasp the morphological structure of the liver and liver blood vessels. However, accurate identification and segmentation of hepatic blood vessels in CT images poses a formidable challenge. Manually locating and segmenting liver vessels in CT images is time-consuming and impractical. There is an imperative clinical requirement for a precise and effective algorithm to segment liver vessels. In response to this demand, the current paper advocates a liver vessel segmentation approach that employs an enhanced 3D fully convolutional neural network V-Net. The network model improves the basic network structure according to the characteristics of liver vessels. First, a pyramidal convolution block is introduced between the encoder and decoder of the network to improve the network localization ability. Then, multi-resolution deep supervision is introduced in the network, resulting in more robust segmentation. Finally, by fusing feature maps of different resolutions, the overall segmentation result is predicted. Evaluation experiments on public datasets demonstrate that our improved scheme can increase the segmentation ability of existing network models for liver vessels. Compared with the existing work, the experimental outcomes demonstrate that the technique presented in this manuscript has attained superior performance on the Dice Coefficient index, which can promote the treatment of liver tumors.

Modified DeeplabV3+ with multi-level context attention mechanism for colonoscopy polyp segmentation.

The development of automated methods for analyzing medical images of colon cancer is one of the main research fields. A colonoscopy is a medical treatment that enables a doctor to look for any abnormalities like polyps, cancer, or inflammatory tissue inside the colon and rectum. It falls under the category of gastrointestinal illnesses, and it claims the lives of almost two million people worldwide. Video endoscopy is an advanced medical imaging approach to diagnose gastrointestinal disorders such as inflammatory bowel, ulcerative colitis, esophagitis, and polyps. Medical video endoscopy generates several images, which must be reviewed by specialists. The difficulty of manual diagnosis has sparked research towards computer-aided techniques that can quickly and reliably diagnose all generated images. The proposed methodology establishes a framework for diagnosing coloscopy diseases. Endoscopists can lower the risk of polyps turning into cancer during colonoscopies by using more accurate computer-assisted polyp detection and segmentation. With the aim of creating a model that can automatically distinguish polyps from images, we presented a modified DeeplabV3+ model in this study to carry out segmentation tasks successfully and efficiently. The framework's encoder uses a pre-trained dilated convolutional residual network for optimal feature map resolution. The robustness of the modified model is tested against state-of-the-art segmentation approaches. In this work, we employed two publicly available datasets, CVC-Clinic DB and Kvasir-SEG, and obtained Dice similarity coefficients of 0.97 and 0.95, respectively. The results show that the improved DeeplabV3+ model improves segmentation efficiency and effectiveness in both software and hardware with only minor changes.

[Automatic modeling of the knee joint based on artificial intelligence].

Objective: To investigate an artificial intelligence (AI) automatic segmentation and modeling method for knee joints, aiming to improve the efficiency of knee joint modeling. Methods: Knee CT images of 3 volunteers were randomly selected. AI automatic segmentation and manual segmentation of images and modeling were performed in Mimics software. The AI-automated modeling time was recorded. The anatomical landmarks of the distal femur and proximal tibia were selected with reference to previous literature, and the indexes related to the surgical design were calculated. Pearson correlation coefficient ( r) was used to judge the correlation of the modeling results of the two methods; the consistency of the modeling results of the two methods were analyzed by DICE coefficient. Results: The three-dimensional model of the knee joint was successfully constructed by both automatic modeling and manual modeling. The time required for AI to reconstruct each knee model was 10.45, 9.50, and 10.20 minutes, respectively, which was shorter than the manual modeling [(64.73±17.07) minutes] in the previous literature. Pearson correlation analysis showed that there was a strong correlation between the models generated by manual and automatic segmentation ( r=0.999, P<0.001). The DICE coefficients of the 3 knee models were 0.990, 0.996, and 0.944 for the femur and 0.943, 0.978, and 0.981 for the tibia, respectively, verifying a high degree of consistency between automatic modeling and manual modeling. Conclusion: The AI segmentation method in Mimics software can be used to quickly reconstruct a valid knee model.

An intraoperative artificial intelligence system identifying anatomical landmarks for laparoscopic cholecystectomy: a prospective clinical feasibility trial (J-SUMMIT-C-01).

BACKGROUND: We have implemented Smart Endoscopic Surgery (SES), a surgical system that uses artificial intelligence (AI) to detect the anatomical landmarks that expert surgeons base on to perform certain surgical maneuvers. No report has verified the use of AI-based support systems for surgery in clinical practice, and no evaluation method has been established. To evaluate the detection performance of SES, we have developed and established a new evaluation method by conducting a clinical feasibility trial. METHODS: A single-center prospective clinical feasibility trial was conducted on 10 cases of LC performed at Oita University hospital. Subsequently, an external evaluation committee (EEC) evaluated the AI detection accuracy for each landmark using five-grade rubric evaluation and DICE coefficient. We defined LM-CBD as the expert surgeon's "judge" of the cystic bile duct in endoscopic images. RESULTS: The average detection accuracy on the rubric by the EEC was 4.2 ± 0.8 for the LM-CBD. The DICE coefficient between the AI detection area of the LM-CBD and the EEC members' evaluation was similar to the mean value of the DICE coefficient between the EEC members. The DICE coefficient was high score for the case that was highly evaluated by the EEC on a five-grade scale. CONCLUSION: This is the first feasible clinical trial of an AI system designed for intraoperative use and to evaluate the AI system using an EEC. In the future, this concept of evaluation for the AI system would contribute to the development of new AI navigation systems for surgery.

[Comparison of resting state and task-based functional MRI in preoperative mapping in patients with brain gliomas]. / Sravnenie funktsional'noi magnitno-rezonansnoi tomografii v sostoyanii pokoya i stimul-zavisimoi funktsional'noi magnitno-rezonansnoi tomografii v predoperatsionnom kartirovanii u patsientov s gliomami golovnogo mozga.

OBJECTIVE: To analyze and compare the results of cerebral cortex mapping with task-based (tb-fMRI) and resting-state functional MRI in patients with glioma of eloquent cortical areas. MATERIAL AND METHODS: There were 55 patients (24 men and 31 women aged 24 - 74 years, median 39) with glial tumors. In 26 patients, the tumor was located in motor areas. Twenty-nine patients had lesions of Broca and Wernicke's areas. All patients underwent preoperative tb-fMRI and rs-fMRI. Then, resection of tumor was carried out in all cases. RESULTS: Comparison of fMRI and rs-fMRI activation maps was assessed by calculating the Dice coefficient for inclusive speech and motor cortex masks and exclusive masks without brainstem, cerebellum, subcortical nuclei. Inclusive Dice coefficient for motor cortex ranged from 0.11 to 0.50, for speech cortex - from 0.006 to 0.240 (p<0.05). In case of exclusive masks, this value ranged from 0.15 to 0.55 for motor cortex and from 0.004 to 0.205 for speech cortex (p<0.05). CONCLUSION: When comparing the results of cortical mapping in patients with glial tumors, the use of hemispheric exclusive and inclusive masks did not significantly increase activation maps matching. Probably, low degree of correspondence was associated with different genesis of activations, as well as with high variability of speech cortex.

Semantic Segmentation of Extraocular Muscles on Computed Tomography Images Using Convolutional Neural Networks.

Computed tomography (CT) imaging of the orbit with measurement of extraocular muscle size can be useful for diagnosing and monitoring conditions that affect extraocular muscles. However, the manual measurement of extraocular muscle size can be time-consuming and tedious. The purpose of this study is to evaluate the effectiveness of deep learning algorithms in segmenting extraocular muscles and measuring muscle sizes from CT images. Consecutive CT scans of orbits from 210 patients between 1 January 2010 and 31 December 2019 were used. Extraocular muscles were manually annotated in the studies, which were then used to train the deep learning algorithms. The proposed U-net algorithm can segment extraocular muscles on coronal slices of 32 test samples with an average dice score of 0.92. The thickness and area measurements from predicted segmentations had a mean absolute error (MAE) of 0.35 mm and 3.87 mm2, respectively, with a corresponding mean absolute percentage error (MAPE) of 7 and 9%, respectively. On qualitative analysis of 32 test samples, 30 predicted segmentations from the U-net algorithm were accepted while 2 were rejected. Based on the results from quantitative and qualitative evaluation, this study demonstrates that CNN-based deep learning algorithms are effective at segmenting extraocular muscles and measuring muscles sizes.

A deep learning-based model for plant lesion segmentation, subtype identification, and survival probability estimation.

Plants are the primary source of food for world's population. Diseases in plants can cause yield loss, which can be mitigated by continual monitoring. Monitoring plant diseases manually is difficult and prone to errors. Using computer vision and artificial intelligence (AI) for the early identification of plant illnesses can prevent the negative consequences of diseases at the very beginning and overcome the limitations of continuous manual monitoring. The research focuses on the development of an automatic system capable of performing the segmentation of leaf lesions and the detection of disease without requiring human intervention. To get lesion region segmentation, we propose a context-aware 3D Convolutional Neural Network (CNN) model based on CANet architecture that considers the ambiguity of plant lesion placement in the plant leaf image subregions. A Deep CNN is employed to recognize the subtype of leaf lesion using the segmented lesion area. Finally, the plant's survival is predicted using a hybrid method combining CNN and Linear Regression. To evaluate the efficacy and effectiveness of our proposed plant disease detection scheme and survival prediction, we utilized the Plant Village Benchmark Dataset, which is composed of several photos of plant leaves affected by a certain disease. Using the DICE and IoU matrices, the segmentation model performance for plant leaf lesion segmentation is evaluated. The proposed lesion segmentation model achieved an average accuracy of 92% with an IoU of 90%. In comparison, the lesion subtype recognition model achieves accuracies of 91.11%, 93.01 and 99.04 for pepper, potato and tomato plants. The higher accuracy of the proposed model indicates that it can be utilized for real-time disease detection in unmanned aerial vehicles and offline to offer crop health updates and reduce the risk of low yield.

Contours modeling of fascicular groups from micro-computed tomography images of peripheral nerves.

The objective is to explore the appropriate method to establish the mathematical model of fascicular groups' contours from micro-CT images of peripheral nerves during the nonsplitting/merging phase. The original contours of fascicular groups from the micro-CT image were described as the discrete pixel points. All discrete pixel points of shapes were extracted into a data set through image processing. The data set was modeled by Bezier, B-spline method, respectively, in which each discrete point was used as a control point for modeling. In the Bezier method, the contour of a nerve bundle needs more than two different Bezier curves to model, making the junction points between two models discontinuous. The contour model described by B-spline is very close to the original contour of nerve bundles when all discrete points are used as the control points. The models described by B-spline have different terms and parameters, making it difficult to calculate in the following research. When the third-order quasi-uniform B-spline method is employed, all nerve bundles models have the same number of terms. The modeling error of third-order quasi-uniform B-spline is less than 3% when the Dice coefficient is more than 95%, and the appropriate number of sampling times is 21. The modeling accuracy is improved with increased sampling times when it is less than 21. However, the modeling accuracy remains stable while the number of sampling times is more than 21. The third-order quasi-uniform B-spline is more efficient in modeling nerve bundles' contour, which is more accurate and straightforward.

Multistage Segmentation of Prostate Cancer Tissues Using Sample Entropy Texture Analysis.

In this study, a multistage segmentation technique is proposed that identifies cancerous cells in prostate tissue samples. The benign areas of the tissue are distinguished from the cancerous regions using the texture of glands. The texture is modeled based on wavelet packet features along with sample entropy values. In a multistage segmentation process, the mean-shift algorithm is applied on the pre-processed images to perform a coarse segmentation of the tissue. Wavelet packets are employed in the second stage to obtain fine details of the structured shape of glands. Finally, the texture of the gland is modeled by the sample entropy values, which identifies epithelial regions from stroma patches. Although there are three stages of the proposed algorithm, the computation is fast as wavelet packet features and sample entropy values perform robust modeling for the required regions of interest. A comparative analysis with other state-of-the-art texture segmentation techniques is presented and dice ratios are computed for the comparison. It has been observed that our algorithm not only outperforms other techniques, but, by introducing sample entropy features, identification of cancerous regions of tissues is achieved with 90% classification accuracy, which shows the robustness of the proposed algorithm.

Imaging quality of PROPELLER diffusion-weighted MR imaging and its diagnostic performance in distinguishing pleomorphic adenomas from Warthin tumors of the parotid gland.

The aim of this study was to evaluate the imaging quality and diagnostic performance of fast spin echo diffusion-weighted imaging with periodically rotated overlapping parallel lines with enhanced reconstruction (FSE-PROP-DWI) in distinguishing parotid pleomorphic adenoma (PMA) from Warthin tumor (WT). This retrospective study enrolled 44 parotid gland tumors from 34 patients, including 15 PMAs and 29 WTs with waived written informed consent. All participants underwent 1.5 T diffusion-weighted imaging including FSE-PROP-DWI and single-shot echo-planar diffusion-weighted imaging (SS-EP-DWI). After imaging resizing and registration among T2WI, FSE-PROP-DWI and SS-EP-DWI, imaging distortion was quantitatively analyzed by using the Dice coefficient. Signal-to-noise ratio and contrast-to-noise ratio were qualitatively evaluated. The mean apparent diffusion coefficient (ADC) of parotid gland tumors was calculated. Wilcoxon signed-rank test was used for paired comparison between FSE-PROP-DWI versus SS-EP-DWI. Mann-Whitney U test was used for independent group comparison between PMAs versus WTs. Diagnostic performance was evaluated by receiver operating characteristics curve analysis. P < 0.05 was considered statistically significant. The Dice coefficient was statistically significantly higher on FSE-PROP-DWI than SS-EP-DWI for both tumors (P < 0.005). Mean ADC was statistically significantly higher in PMAs than WTs on both FSE-PROP-DWI and SS-EP-DWI (P < 0.005). FSE-PROP-DWI and SS-EP-DWI successfully distinguished PMAs from WTs with an AUC of 0.880 and 0.945, respectively (P < 0.05). Sensitivity, specificity, positive predictive value, negative predictive value and accuracy in diagnosing PMAs were 100%, 69.0%, 62.5%, 100% and 79.5% for FSE-PROP-DWI, and 100%, 82.8%, 75%, 100% and 88.6% for SS-EP-DWI, respectively. FSE-PROP-DWI is useful to distinguish parotid PMAs from WTs with less distortion of tumors but lower AUC than SS-EP-DWI.

SVM-Enabled Intelligent Genetic Algorithmic Model for Realizing Efficient Universal Feature Selection in Breast Cyst Image Acquired via Ultrasound Sensing Systems.

In recent years, there are several cost-effective intelligent sensing systems such as ultrasound imaging systems for visualizing the internal body structures of the body. Further, such intelligent sensing systems such as ultrasound systems have been deployed by medical doctors around the globe for efficient detection of several diseases and disorders in the human body. Even though the ultrasound sensing system is a useful tool for obtaining the imagery of various body parts, there is always a possibility of inconsistencies in these images due to the variation in the settings of the system parameters. Therefore, in order to overcome such issues, this research devises an SVM-enabled intelligent genetic algorithmic model for choosing the universal features with four distinct settings of the parameters. Subsequently, the distinguishing characteristics of these features are assessed utilizing the Sorensen-Dice coefficient, T-test, and Pearson's R measure. It is apparent from the results of the SVM-enabled intelligent genetic algorithmic model that this approach aids in the effectual selection of universal features for the breast cyst images. In addition, this approach also accomplishes superior accuracy in the classification of the ultrasound image for four distinct settings of the parameters.

Segmentation and Boundary Detection of Fetal Kidney Images in Second and Third Trimesters Using Kernel-Based Fuzzy Clustering.

Organ segmentation is an important step in Ultrasound fetal images for early prediction of congenital abnormalities and to estimate delivery date. In many applications of 2D medical imaging, they face problems with speckle noise and object contours. Frequent scanning of fetal leads to clinical disturbances to the fetal growth and the quantitative interpretation of Ultrasonic images also a difficult task compared to other image modalities. In the present work a three-stage hybrid algorithm has been developed to segment the US fetal kidney images for the detection of shape and contour. At the first stage the hybrid Mean Median (Hybrid MM) filter is applied to reduce the speckle noise. Then a kernel based Fuzzy C - means clustering is used to detect the shape and contour. Finally, the texture features are obtained from the segmented images. Based on the obtained texture features, the abnormalities are detected. The Gaussian Radial basis function provides an accuracy of 80% at the second and third trimesters with weighted constant ranging from 4 to 8, compared to other global kernel functions. Similarly the proposed method has an accuracy of 86% with compared to other FCM techniques.

Feature enhancement framework for brain tumor segmentation and classification.

Automatic medical image analysis is one of the key tasks being used by the medical community for disease diagnosis and treatment planning. Statistical methods are the major algorithms used and consist of few steps including preprocessing, feature extraction, segmentation, and classification. Performance of such statistical methods is an important factor for their successful adaptation. The results of these algorithms depend on the quality of images fed to the processing pipeline: better the images, higher the results. Preprocessing is the pipeline phase that attempts to improve the quality of images before applying the chosen statistical method. In this work, popular preprocessing techniques are investigated from different perspectives where these preprocessing techniques are grouped into three main categories: noise removal, contrast enhancement, and edge detection. All possible combinations of these techniques are formed and applied on different image sets which are then passed to a predefined pipeline of feature extraction, segmentation, and classification. Classification results are calculated using three different measures: accuracy, sensitivity, and specificity while segmentation results are calculated using dice similarity score. Statistics of five high scoring combinations are reported for each data set. Experimental results show that application of proper preprocessing techniques could improve the classification and segmentation results to a greater extent. However, the combinations of these techniques depend on the characteristics and type of data set used.

AutoVOI: real-time automatic prescription of volume-of-interest for single voxel spectroscopy.

PURPOSE: To develop a fast and automated volume-of-interest (VOI) prescription pipeline (AutoVOI) for single-voxel MRS that removes the need for manual VOI placement, allows flexible VOI planning in any brain region, and enables high inter- and intra-subject consistency of VOI prescription. METHODS: AutoVOI was designed to transfer pre-defined VOIs from an atlas to the 3D anatomical data of the subject during the scan. The AutoVOI pipeline was optimized for consistency in VOI placement (precision), enhanced coverage of the targeted tissue (accuracy), and fast computation speed. The tool was evaluated against manual VOI placement using existing T1 -weighted data sets and corresponding VOI prescriptions. Finally, it was implemented on 2 scanner platforms to acquire MRS data from clinically relevant VOIs that span the cerebrum, cerebellum, and the brainstem. RESULTS: The AutoVOI pipeline includes skull stripping, non-linear registration of the atlas to the subject's brain, and computation of the VOI coordinates and angulations using a minimum oriented bounding box algorithm. When compared against manual prescription, AutoVOI showed higher intra- and inter-subject spatial consistency, as quantified by generalized Dice coefficients (GDC), lower intra- and inter-subject variability in tissue composition (gray matter, white matter, and cerebrospinal fluid) and higher or equal accuracy, as quantified by GDC of prescribed VOI with targeted tissues. High quality spectra were obtained on Siemens and Philips 3T systems from 6 automatically prescribed VOIs by the tool. CONCLUSION: Robust automatic VOI prescription is feasible and can help facilitate clinical adoption of MRS by avoiding operator dependence of manual selection.

Mean curvature and texture constrained composite weighted random walk algorithm for optic disc segmentation towards glaucoma screening.

Accurate optic disc (OD) segmentation is an important step in obtaining cup-to-disc ratio-based glaucoma screening using fundus imaging. It is a challenging task because of the subtle OD boundary, blood vessel occlusion and intensity inhomogeneity. In this Letter, the authors propose an improved version of the random walk algorithm for OD segmentation to tackle such challenges. The algorithm incorporates the mean curvature and Gabor texture energy features to define the new composite weight function to compute the edge weights. Unlike the deformable model-based OD segmentation techniques, the proposed algorithm remains unaffected by curve initialisation and local energy minima problem. The effectiveness of the proposed method is verified with DRIVE, DIARETDB1, DRISHTI-GS and MESSIDOR database images using the performance measures such as mean absolute distance, overlapping ratio, dice coefficient, sensitivity, specificity and precision. The obtained OD segmentation results and quantitative performance measures show robustness and superiority of the proposed algorithm in handling the complex challenges in OD segmentation.

Electric Field Comparison between Microelectrode Recording and Deep Brain Stimulation Systems-A Simulation Study.

The success of deep brain stimulation (DBS) relies primarily on the localization of the implanted electrode. Its final position can be chosen based on the results of intraoperative microelectrode recording (MER) and stimulation tests. The optimal position often differs from the final one selected for chronic stimulation with the DBS electrode. The aim of the study was to investigate, using finite element method (FEM) modeling and simulations, whether lead design, electrical setup, and operating modes induce differences in electric field (EF) distribution and in consequence, the clinical outcome. Finite element models of a MER system and a chronic DBS lead were developed. Simulations of the EF were performed for homogenous and patient-specific brain models to evaluate the influence of grounding (guide tube vs. stimulator case), parallel MER leads, and non-active DBS contacts. Results showed that the EF is deformed depending on the distance between the guide tube and stimulating contact. Several parallel MER leads and the presence of the non-active DBS contacts influence the EF distribution. The DBS EF volume can cover the intraoperatively produced EF, but can also extend to other anatomical areas. In conclusion, EF deformations between stimulation tests and DBS should be taken into consideration as they can alter the clinical outcome.

A large-scale study on the effects of sex on gray matter asymmetry.

Research on sex-related brain asymmetries has not yielded consistent results. Despite its importance to further understanding of normal brain development and mental disorders, the field remains relatively unexplored. Here we employ a recently developed asymmetry measure, based on the Dice coefficient, to detect sex-related gray matter asymmetries in a sample of 457 healthy participants (266 men and 191 women) obtained from 5 independent databases. Results show that women's brains are more globally symmetric than men's (p < 0.001). Although the new measure accounts for asymmetries distributed all over the brain, several specific structures were identified as systematically more symmetric in women, such as the thalamus and the cerebellum, among other structures, some of which are typically involved in language production. These sex-related asymmetry differences may be defined at the neurodevelopmental stage and could be associated with functional and cognitive sex differences, as well as with proneness to develop a mental disorder.

MRI FLAIR lesion segmentation in multiple sclerosis: Does automated segmentation hold up with manual annotation?

INTRODUCTION: Magnetic resonance imaging (MRI) has become key in the diagnosis and disease monitoring of patients with multiple sclerosis (MS). Both, T2 lesion load and Gadolinium (Gd) enhancing T1 lesions represent important endpoints in MS clinical trials by serving as a surrogate of clinical disease activity. T2- and fluid-attenuated inversion recovery (FLAIR) lesion quantification - largely due to methodological constraints - is still being performed manually or in a semi-automated fashion, although strong efforts have been made to allow automated quantitative lesion segmentation. In 2012, Schmidt and co-workers published an algorithm to be applied on FLAIR sequences. The aim of this study was to apply the Schmidt algorithm on an independent data set and compare automated segmentation to inter-rater variability of three independent, experienced raters. METHODS: MRI data of 50 patients with RRMS were randomly selected from a larger pool of MS patients attending the MS Clinic at the Brain and Mind Centre, University of Sydney, Australia. MRIs were acquired on a 3.0T GE scanner (Discovery MR750, GE Medical Systems, Milwaukee, WI) using an 8 channel head coil. We determined T2-lesion load (total lesion volume and total lesion number) using three versions of an automated segmentation algorithm (Lesion growth algorithm (LGA) based on SPM8 or SPM12 and lesion prediction algorithm (LPA) based on SPM12) as first described by Schmidt et al. (2012). Additionally, manual segmentation was performed by three independent raters. We calculated inter-rater correlation coefficients (ICC) and dice coefficients (DC) for all possible pairwise comparisons. RESULTS: We found a strong correlation between manual and automated lesion segmentation based on LGA SPM8, regarding lesion volume (ICC = 0.958 and DC = 0.60) that was not statistically different from the inter-rater correlation (ICC = 0.97 and DC = 0.66). Correlation between the two other algorithms (LGA SPM12 and LPA SPM12) and manual raters was weaker but still adequate (ICC = 0.927 and DC = 0.53 for LGA SPM12 and ICC = 0.949 and DC = 0.57 for LPA SPM12). Variability of both manual and automated segmentation was significantly higher regarding lesion numbers. CONCLUSION: Automated lesion volume quantification can be applied reliably on FLAIR data sets using the SPM based algorithm of Schmidt et al. and shows good agreement with manual segmentation.

Quantifying Variability of Manual Annotation in Cryo-Electron Tomograms.

Although acknowledged to be variable and subjective, manual annotation of cryo-electron tomography data is commonly used to answer structural questions and to create a "ground truth" for evaluation of automated segmentation algorithms. Validation of such annotation is lacking, but is critical for understanding the reproducibility of manual annotations. Here, we used voxel-based similarity scores for a variety of specimens, ranging in complexity and segmented by several annotators, to quantify the variation among their annotations. In addition, we have identified procedures for merging annotations to reduce variability, thereby increasing the reliability of manual annotation. Based on our analyses, we find that it is necessary to combine multiple manual annotations to increase the confidence level for answering structural questions. We also make recommendations to guide algorithm development for automated annotation of features of interest.

Semi-automated quantification of retinal IS/OS damage in en-face OCT image.

A variety of vision ailments are indicated by structural changes in the retinal substructures of the posterior segment of the eye. In particular, integrity of the inner-segment/outer-segment (IS/OS) junction directly relates to the visual acuity. In the en-face optical coherence tomography (OCT) image, IS/OS damage manifests as a dark spot in the foveal region, and its quantification, usually performed by experts, assumes diagnostic significance. In this context, in view of the general scarcity of experts, it becomes imperative to develop algorithmic methods to reduce expert time and effort. Accordingly, we propose a semi-automated method based on level sets. As the energy function, we adopt mutual information which exploits the difference in statistical properties of the lesion and its surroundings. On a dataset of 27 en-face OCT images, segmentation obtained by the proposed algorithm exhibits close visual agreement with that obtained manually. Importantly, our results also match manual results in various statistical criteria. In particular, we achieve a mean Dice coefficient of 85.69%, desirably close to the corresponding observer repeatability index of 89.45%. Finally, we quantify algorithmic accuracy in terms of two quotient measures, defined relative to observer repeatability, which could be used as bases for comparison with future algorithms, even if the latter are tested on disparate datasets.