Convolutional Neural Networks For Segmentation Research Articles

Data augmentation via warping transforms for modeling natural variability in the corneal endothelium enhances semi-supervised segmentation.

Image segmentation of the corneal endothelium with deep convolutional neural networks (CNN) is challenging due to the scarcity of expert-annotated data. This work proposes a data augmentation technique via warping to enhance the performance of semi-supervised training of CNNs for accurate segmentation. We use a unique augmentation process for images and masks involving keypoint extraction, Delaunay triangulation, local affine transformations, and mask refinement. This approach accurately captures the natural variability of the corneal endothelium, enriching the dataset with realistic and diverse images. The proposed method achieved an increase in the mean intersection over union (mIoU) and Dice coefficient (DC) metrics of 17.2% and 4.8% respectively, for the segmentation task in corneal endothelial images on multiple CNN architectures. Our data augmentation strategy successfully models the natural variability in corneal endothelial images, thereby enhancing the performance and generalization capabilities of semi-supervised CNNs in medical image cell segmentation tasks.

CIFTC-Net: Cross information fusion network with transformer and CNN for polyp segmentation

Polyp segmentation plays a crucial role in the early diagnosis and treatment of colorectal cancer, which is the third most common cancer worldwide. Despite remarkable successes achieved by recent deep learning-related works, accurate segmentation of polyps remains challenging due to the diversity in their shapes, sizes, appearances, and other factors. To address these problems, a novel cross information fusion network with Transformer and convolutional neural network (CNN) for polyp segmentation, named CIFTC-Net, is proposed to improve the segmentation performance of colon polyps. In particular, a dual-branch encoder with Pyramid Vision Transformer (PVT) and ResNet50 is employed to take full advantage of both the global semantic information and local spatial features to enhance the feature representation ability. To effectively fuse the two types of features, a new global–local feature fusion (GLFF) module is designed. Additionally, in the PVT branch, a multi-scale feature integration (MSFI) module is introduced to fuse multi-scale features adaptively. At the bottom of the model, a multi-scale atrous pyramid bridging (MSAPB) module is proposed to achieve rich and robust multi-level features and improve the segmentation accuracy. Experimental results on four public polyp segmentation datasets demonstrate that CIFTC-Net surpasses current state-of-the-art methods across various metrics, showcasing its superiority in segmentation accuracy, generalization ability, and handling of complex images.

An effective multi-scale interactive fusion network with hybrid Transformer and CNN for smoke image segmentation

Smoke has visually elusive appearances, especially in low-light conditions, so it is quite difficult to quickly and accurately detect smoke from images. To address these challenges, we design a dual-encoder structure of Transformer and Convolutional Neural Network (CNN) to propose an effective Multi-scale Interactive Fusion Network (MIFNet) for smoke image segmentation. To enhance the presentation of features, we propose a Local Feature Enhancement Propagation (LFEP) module to enhance spatial details. To optimize global and local features for efficient fusion, we integrate LFEP into the original Transformer to replace the traditional multi-head self-attention mechanism. Then, we propose a Multi-level Attention Coupled Module (MACM) to fuse Transformer and CNN features of the dual-encoder. MACM can flexibly focus on information interaction between different levels of two encoding paths. Finally, we design a Prior-guided Multi-scale Fusion Decoder (PMFD), which combines prior knowledge with a multi-scale feature fusion strategy to improve the performance of segmentation. Experimental results demonstrate that MIFNet substantially outperforms the state-of-the-art methods. MIFNet achieves a mean Intersection over Union (mIoU) of 81.6% on the synthetic smoke (SYN70K) dataset, and a remarkable accuracy of 98.3% on the forest smoke dataset.

Automated dentition segmentation: 3D UNet-based approach with MIScnn framework

Advancements in technology have led to the adoption of digital workflows in dentistry, which require the segmentation of regions of interest from cone-beam computed tomography (CBCT) scans. These segmentations assist in diagnosis, treatment planning, and research. However, manual segmentation is an expensive and labor-intensive process. Therefore, automated methods, such as convolutional neural networks (CNNs), provide a more efficient way to generate segmentations from CBCT scans. A three-dimensional UNet-based CNN model, utilizing the Medical Image Segmentation CNN framework, was used for training and generating predictions from CBCT scans. A dataset of 351 CBCT scans, with ground-truth labels created through manual segmentation using AI-assisted segmentation software, was prepared. Data preprocessing, augmentation, and model training were performed, and the performance of the proposed CNN model was analyzed. The CNN model achieved high accuracy in segmenting maxillary and mandibular teeth from CBCT scans, with average Dice Similarity Coefficient values of 91.83% and 91.35% for maxillary and mandibular teeth, respectively. Performance metrics, including Intersection over Union, precision, and recall, further confirmed the model's effectiveness. The study demonstrates the efficacy of the three-dimensional UNet-based CNN model within the Medical Image Segmentation CNN framework for automated segmentation of maxillary and mandibular dentition from CBCT scans. Automated segmentation using CNNs has the potential to deliver accurate and efficient results, offering a significant advantage over traditional segmentation methods.

Muscle Fat and Volume Differences in People With Hip-Related Pain Compared With Controls: A Machine Learning Approach.

Hip-related pain (HRP) affects young to middle-aged active adults and impacts physical activity, finances and quality of life. HRP includes conditions like femoroacetabular impingement syndrome and labral tears. Lateral hip muscle dysfunction and atrophy in HRP are more pronounced in advanced hip pathology, with limited evidence in younger populations. While MRI use for assessing hip muscle morphology is increasing, with automated deep-learning techniques showing promise, studies assessing their accuracy are limited. Therefore, we aimed to compare hip intramuscular fat infiltrate (MFI) and muscle volume, in individuals with and without HRP as well as assess the reliability and accuracy of automated machine-learning segmentations compared with human-generated segmentation. This cross-sectional study included sub-elite/amateur football players (Australian football and soccer) with a greater than 6-month history of HRP [n = 180, average age 28.32, (standard deviation 5.88) years, 19% female] and a control group of sub-elite/amateur football players without pain [n = 48, 28.89 (6.22) years, 29% female]. Muscle volume and MFI of gluteus maximus, medius, minimis and tensor fascia latae were assessed using MRI. Associations between muscle volume and group were explored using linear regression models, controlling for body mass index, age, sport and sex. A convolutional neural network (CNN) machine-learning approach was compared with human-performed muscle segmentations in a subset of participants (n = 52) using intraclass correlation coefficients and Sorensen-Dice index. When considering adjusted estimates of muscle volume, there were significant differences observed between groups for gluteus medius (adjusted mean difference 23 858 mm3 [95% confidence interval 7563, 40 137]; p = 0.004) and tensor fascia latae (6660 mm3 [2440, 13 075]; p = 0.042). No differences were observed between groups for gluteus maximus (18 265 mm3 [-21 209, 50 782]; p = 0.419) or minimus (3893 mm3 [-2209, 9996]; p = 0.21). The CNN was trained for 30 000 iterations and assessed its accuracy and reliability on an independent testing dataset, achieving high segmentation accuracy (mean Sorenson-Dice index >0.900) and excellent muscle volume and MFI reliability (ICC2,1 > 0.900). The CNN outperformed manual raters, who had slightly lower interrater accuracy (Sorensen-Dice index >0.800) and reliability (ICC2,1 > 0.800). The increased muscle volumes in the symptomatic group compared with controls could be associated with increased myofibrillar size, sarcoplasmic hypertrophy or both. These changes may facilitate greater muscular efficiency for a given load, enabling the athlete to maintain their normal level of function. In addition, the CNNs for muscle segmentation was more efficient and demonstrated excellent reliability in comparison to manual segmentations.

Convolutional Neural Networks for Segmentation of Pleural Mesothelioma: Analysis of Probability Map Thresholds (CALGB 30901, Alliance).

The purpose of this study was to evaluate the impact of probability map threshold on pleural mesothelioma (PM) tumor delineations generated using a convolutional neural network (CNN). One hundred eighty-six CT scans from 48 PM patients were segmented by a VGG16/U-Net CNN. A radiologist modified the contours generated at a 0.5 probability threshold. Percent difference of tumor volume and overlap using the Dice Similarity Coefficient (DSC) were compared between the reference standard provided by the radiologist and CNN outputs for thresholds ranging from 0.001 to 0.9. CNN-derived contours consistently yielded smaller tumor volumes than radiologist contours. Reducing the probability threshold from 0.5 to 0.01 decreased the absolute percent volume difference, on average, from 42.93% to 26.60%. Median and mean DSC ranged from 0.57 to 0.59, with a peak at a threshold of 0.2; no distinct threshold was found for percent volume difference. The CNN exhibited deficiencies with specific disease presentations, such as severe pleural effusion or disease in the pleural fissure. No single output threshold in the CNN probability maps was optimal for both tumor volume and DSC. This study emphasized the importance of considering both figures of merit when evaluating deep learning-based tumor segmentations across probability thresholds. This work underscores the need to simultaneously assess tumor volume and spatial overlap when evaluating CNN performance. While automated segmentations may yield comparable tumor volumes to that of the reference standard, the spatial region delineated by the CNN at a specific threshold is equally important.

DATA DRIVEN AND CELL SPECIFIC DETERMINATION OF NUCLEI-ASSOCIATED ACTIN STRUCTURE

In the specialized field of cellular mechanosignaling, the intricate relationship between the cytoskeleton and nucleus is crucial for determining the behavior and fate of mesenchymal stem cells (MSCs). Existing analytical methods for filamentous actin fibers (F-actin) suffer from issues of accuracy and reproducibility. To address this, our research introduces a significant advancement in image analysis by utilizing two deep-learning segmentation Convolutional Neural Networks (CNNs) based on U-Net architecture. This algorithm precisely quantifies F-actin and offers reproducible data from 3D confocal microscopy images. Our CNNs automatically extract the 3D shape of both the nucleus and individual actin fibers, thereby enabling the geometric reconstruction of these critical cellular components. We applied this methodology to MSCs exposed to low-intensity vibration, and hypothesized that mechanical stimuli will induce perinuclear actin remodeling. Confocal Z-stack images of MSCs were captured using a 40x oil lens and a Zeiss LSM 900 microscope. Nucleus and F-actin structures were stained with Hoechst 33342 and Phalloidin, respectively. Nucleus-based images were employed to mask and isolate actin fibers within the nuclear vicinity. A specialized deep-learning segmentation model was engineered within the PyTorch framework. For model training, a selective 2% of nuclei and F-actin images were annotated. Specifically, three layers from each of six representative nuclei, among approximately 500 cross-sectional layers, were meticulously selected; a similar approach was employed for actin. The model underwent 200 epochs of training. Post-training, the deep-learning model was employed to process all collected cross-sectional images of F-actin and nuclei. A subsequent analytical step involved the scrutiny of intersecting segmented fibers between consecutive layers for geometrically accurate actin fiber reconstruction. We analyzed 102 control and 110 vibration-treated mesenchymal stem cells (MSCs) by utilizing our deep-learning-based reconstruction algorithm on confocal Z-stack images. Our data revealed an 8% decrease in nuclear height from 5.73 µm to 5.28 µm (p < 0.001) upon application of low-intensity vibration. In contrast, the two groups observed no significant difference in nuclear volume. Accompanying these changes at the nuclear level, the following changes in the F-actin cytoskeleton were observed upon LIV application. The total number of actin filaments per cell increased from 117 to 150. The average volume of individual actin fibers rose from 0.40 µm³ to 0.43 µm³. Notably, the volume of fibers at the apical nuclear surface increased from 0.35 µm³ to 0.46 µm³. While the average length of actin fibers in the apical nuclear surface remained largely unchanged due to the addition of smaller fibers in LIV groups, the LIV group had a larger number of fibers that were larger than 15µm.

Research progress on colorectal cancer identification based on convolutional neural network

Colorectal cancer (CRC) is a common malignant tumor that seriously threatens human health. CRC presents a formidable challenge in terms of accurate identification due to its indistinct boundaries. With the widespread adoption of convolutional neural networks (CNNs) in image processing, leveraging CNNs for automatic classification and segmentation holds immense potential for enhancing the efficiency of colorectal cancer recognition and reducing treatment costs. This paper explores the imperative necessity for applying CNNs in clinical diagnosis of CRC. It provides an elaborate overview on research advancements pertaining to CNNs and their improved models in CRC classification and segmentation. Furthermore, this work summarizes the ideas and common methods for optimizing network performance and discusses the challenges faced by CNNs as well as future development trends in their application towards CRC classification and segmentation, thereby promoting their utilization within clinical diagnosis.

COVID-19 lesions image segmentation method based on UniFormer

In view of the fact that the convolution neural network segmentation method lacks to capture the global dependency of infected areas in COVID-19 images, which is not conducive to the complete segmentation of scattered lesion areas, this paper proposes a COVID-19 lesion segmentation method UniUNet based on UniFormer with its strong ability to capture global dependency. Firstly, a U-shaped encoder-decoder structure based on UniFormer is designed, which can enhance the cooperation ability of local and global relations. Secondly, Swin spatial pyramid pooling module is introduced to compensate the influence of spatial resolution reduction in the encoder process and generate multi-scale representation. Multi-scale attention gate is introduced at the skip connection to suppress redundant features and enhance important features. Experiment results show that, compared with the other four methods, the proposed model achieves better results in Dice, loU and Recall on COVID-19-CT-Seg and CC-CCIII dataset, and achieves a more complete segmentation of the lesion area.

Lumbar and Thoracic Vertebrae Segmentation in CT Scans Using a 3D Multi-Object Localization and Segmentation CNN.

Radiation treatment of cancers like prostate or cervix cancer requires considering nearby bone structures like vertebrae. In this work, we present and validate a novel automated method for the 3D segmentation of individual lumbar and thoracic vertebra in computed tomography (CT) scans. It is based on a single, low-complexity convolutional neural network (CNN) architecture which works well even if little application-specific training data are available. It is based on volume patch-based processing, enabling the handling of arbitrary scan sizes. For each patch, it performs segmentation and an estimation of up to three vertebrae center locations in one step, which enables utilizing an advanced post-processing scheme to achieve high segmentation accuracy, as required for clinical use. Overall, 1763 vertebrae were used for the performance assessment. On 26 CT scans acquired for standard radiation treatment planning, a Dice coefficient of 0.921 ± 0.047 (mean ± standard deviation) and a signed distance error of 0.271 ± 0.748 mm was achieved. On the large-sized publicly available VerSe2020 data set with 129 CT scans depicting lumbar and thoracic vertebrae, the overall Dice coefficient was 0.940 ± 0.065 and the signed distance error was 0.109 ± 0.301 mm. A comparison to other methods that have been validated on VerSe data showed that our approach achieved a better overall segmentation performance.

Enhancing surgical instrument segmentation: integrating vision transformer insights with adapter

PurposeIn surgical image segmentation, a major challenge is the extensive time and resources required to gather large-scale annotated datasets. Given the scarcity of annotated data in this field, our work aims to develop a model that achieves competitive performance with training on limited datasets, while also enhancing model robustness in various surgical scenarios.MethodsWe propose a method that harnesses the strengths of pre-trained Vision Transformers (ViTs) and data efficiency of convolutional neural networks (CNNs). Specifically, we demonstrate how a CNN segmentation model can be used as a lightweight adapter for a frozen ViT feature encoder. Our novel feature adapter uses cross-attention modules that merge the multiscale features derived from the CNN encoder with feature embeddings from ViT, ensuring integration of the global insights from ViT along with local information from CNN.ResultsExtensive experiments demonstrate our method outperforms current models in surgical instrument segmentation. Specifically, it achieves superior performance in binary segmentation on the Robust-MIS 2019 dataset, as well as in multiclass segmentation tasks on the EndoVis 2017 and EndoVis 2018 datasets. It also showcases remarkable robustness through cross-dataset validation across these 3 datasets, along with the CholecSeg8k and AutoLaparo datasets. Ablation studies based on the datasets prove the efficacy of our novel adapter module.ConclusionIn this study, we presented a novel approach integrating ViT and CNN. Our unique feature adapter successfully combines the global insights of ViT with the local, multi-scale spatial capabilities of CNN. This integration effectively overcomes data limitations in surgical instrument segmentation. The source code is available at: https://github.com/weimengmeng1999/AdapterSIS.git.

A convolutional neural network (Seg-CNN) for recognition, classification, and segmentation of apple fruit branches and stems in images

This article reports the results of research studies conducted in 2023–2024 on transfer learning of Segmentation Convolutional Neural Networks (Seg-CNN) models for classification, recognition, and segmentation of branches with apple fruits and stems in images. State-of-the-art convolutional neural network architectures, i.e., YOLOv8(n,s,m,l,x)-seg, were used for a detailed segmentation of biological objects in images of varying complexity and scale at the pixel level. An image dataset collected in the field using a GoPro HERO 11 camera was marked up for transfer model training. Data augmentation was performed, producing a total of 2500 images. Image markup was performed using the polygon annotation tool. As a result, polygonal contours around objects were created, outlines of branches, apple tree fruits, and stems were outlined, and segments of objects in the images were indicated. The objects were assigned the following classes: Apple branch, Apple fruit, and Apple stem. Binary classification metrics, such as Precision and Recall, as well as Mean Average Precision (mAP), were used to evaluate the performance of the trained models in recognizing branches with apple fruits and stems in images. The YOLOv8x-seg (mAP50 0.758) and YOLOv8l-seg (mAP50 0.74) models showed high performance in terms of all metrics in recognizing branches, apple fruit, and fruit stems in images, outperforming the YOLOv8n-seg (mAP50 0.7) model due to their more complex architecture. The YOLOv8n-seg model has a faster frame processing speed (11.39 frames/s), rendering it a preferred choice for computing systems with limited resources. The results obtained confirm the prospects of using machine learning algorithms and convolutional neural networks for segmentation and pixel-by-pixel classification of branches with apple fruits and stems on RGB images for monitoring the condition of plants and determining their geometric characteristics.

Adaptive Fusion of Deep Learning With Statistical Anatomical Knowledge for Robust Patella Segmentation From CT Images.

Kneeosteoarthritis (KOA), as a leading joint disease, can be decided by examining the shapes of patella to spot potential abnormal variations. To assist doctors in the diagnosis of KOA, a robust automatic patella segmentation method is highly demanded in clinical practice. Deep learning methods, especially convolutional neural networks (CNNs) have been widely applied to medical image segmentation in recent years. Nevertheless, poor image quality and limited data still impose challenges to segmentation via CNNs. On the other hand, statistical shape models (SSMs) can generate shape priors which give anatomically reliable segmentation to varying instances. Thus, in this work, we propose an adaptive fusion framework, explicitly combining deep neural networks and anatomical knowledge from SSM for robust patella segmentation. Our adaptive fusion framework will accordingly adjust the weight of segmentation candidates in fusion based on their segmentation performance. We also propose a voxel-wise refinement strategy to make the segmentation of CNNs more anatomically correct. Extensive experiments and thorough assessment have been conducted on various mainstream CNN backbones for patella segmentation in low-data regimes, which demonstrate that our framework can be flexibly attached to a CNN model, significantly improving its performance when labeled training data are limited and input image data are of poor quality.

Preoperative CECT-Based Multitask Model Predicts Peritoneal Recurrence and Disease-Free Survival in Advanced Ovarian Cancer: A Multicenter Study

ObjectivePeritoneal recurrence is the predominant pattern of recurrence in advanced ovarian cancer (AOC) and portends a dismal prognosis. Accurate prediction of peritoneal recurrence and disease-free survival (DFS) is crucial to identify patients who might benefit from intensive treatment. We aimed to develop a predictive model for peritoneal recurrence and prognosis in AOC. MethodsIn this retrospective multi-institution study of 515 patients, an end-to-end multi-task convolutional neural network (MCNN) comprising a segmentation convolutional neural network (CNN) and a classification CNN was developed and tested using preoperative CT images, and MCNN-score was generated to indicate the peritoneal recurrence and DFS status in patients with AOC. We evaluated the accuracy of the model for automatic segmentation and predict prognosis. ResultsThe MCNN achieved promising segmentation performances with a mean Dice coefficient of 84.3% (range: 78.8%-87.0%). The MCNN was able to predict peritoneal recurrence in the training (AUC 0.87; 95% CI 0.82–0.90), internal test (0.88; 0.85–0.92), and external test set (0.82; 0.78–0.86). Similarly, MCNN demonstrated consistently high accuracy in predicting recurrence, with an AUC of 0.85; 95% CI 0.82–0.88, 0.83; 95% CI 0.80–0.86, and 0.85; 95% CI 0.83–0.88. For patients with a high MCNN-score of recurrence, it was associated with poorer DFS with P<0.0001 and hazard ratios of 0.1964 (95% CI: 0.1439-0.2680), 0.3249 (95% CI: 0.1896-0.5565), and 0.3458 (95% CI: 0.2582-0.4632). ConclusionsThe MCNN approach demonstrated high performance in predicting peritoneal recurrence and DFS in patients with AOC.

Batch Normalization Based Convolutional Neural Network for Segmentation and Classification of Brain Tumor MRI Images

Batch Normalization Based Convolutional Neural Network for Segmentation and Classification of Brain Tumor MRI Images

Incorporating Adaptive Sparse Graph Convolutional Neural Networks for Segmentation of Organs at Risk in Radiotherapy

Precisely segmenting the organs at risk (OARs) in computed tomography (CT) plays an important role in radiotherapy’s treatment planning, aiding in the protection of critical tissues during irradiation. Renowned deep convolutional neural networks (DCNNs) and prevailing transformer-based architectures are widely utilized to accomplish the segmentation task, showcasing advantages in capturing local and contextual characteristics. Graph convolutional networks (GCNs) are another specialized model designed for processing the nongrid dataset, e.g., citation relationship. The DCNNs and GCNs are considered as two distinct models applicable to the grid and nongrid datasets, respectively. Motivated by the recently developed dynamic-channel GCN (DCGCN) that attempts to leverage the graph structure to enhance the feature extracted by the DCNNs, this paper proposes a novel architecture termed adaptive sparse GCN (ASGCN) to mitigate the inherent limitations in DCGCN from the aspect of node’s representation and adjacency matrix’s construction. For the node’s representation, the global average pooling used in the DCGCN is replaced by the learning mechanism to accommodate the segmentation task. For the adjacency matrix, an adaptive regularization strategy is leveraged to penalize the coefficient in the adjacency matrix, resulting in a sparse one that can better exploit the relationships between nodes. Rigorous experiments on multiple OARs’ segmentation tasks of the head and neck demonstrate that the proposed ASGCN can effectively improve the segmentation accuracy. Comparison between the proposed method and other prevalent architectures further confirms the superiority of the ASGCN.

Detection of bladder cancer cells using quantitative interferometric label-free imaging flow cytometry.

Bladder cancer is one of the most common cancers with a high recurrence rate. Patients undergo mandatory yearly scrutinies, including cystoscopies, which makes bladder cancer highly distressing and costly. Here, we aim to develop a non-invasive, label-free method for the detection of bladder cancer cells in urine samples, which is based on interferometric imaging flow cytometry. Eight urothelial carcinoma and one normal urothelial cell lines, along with red and white blood cells, imaged quantitatively without staining by an interferometric phase microscopy module while flowing in a microfluidic chip, and classified by two machine-learning algorithms, based on deep-learning semantic segmentation convolutional neural network and extreme gradient boosting. Furthermore, urine samples obtained from bladder-cancer patients and healthy volunteers were imaged, and classified by the system. We achieved accuracy and area under the curve (AUC) of 99% and 97% for the cell lines on both machine-learning algorithms. For the real urine samples, the accuracy and AUC were 96% and 96% for the deep-learning algorithm and 95% and 93% for the gradient-boosting algorithm, respectively. By combining label-free interferometric imaging flow cytometry with high-end classification algorithms, we achieved high-performance differentiation between healthy and malignant cells. The proposed technique has the potential to supplant cystoscopy in the bladder cancer surveillance and diagnosis space.

Development and prospective clinical validation of a convolutional neural network for automated detection and segmentation of focal cortical dysplasias

PurposeFocal cortical dysplasias (FCDs) are a leading cause of drug-resistant epilepsy. Early detection and resection of FCDs have favorable prognostic implications for postoperative seizure freedom. Despite advancements in imaging methods, FCD detection remains challenging. House et al. (2021) introduced a convolutional neural network (CNN) for automated FCD detection and segmentation, achieving a sensitivity of 77.8%. However, its clinical applicability was limited due to a low specificity of 5.5%. The objective of this study was to improve the CNN’s performance through data-driven training and algorithm optimization, followed by a prospective validation on daily-routine MRIs. Material and methodsA dataset of 300 3 T MRIs from daily clinical practice, including 3D T1 and FLAIR sequences, was prospectively compiled. The MRIs were visually evaluated by two neuroradiologists and underwent morphometric assessment by two epileptologists. The dataset included 30 FCD cases (11 female, mean age: 28.1 ± 10.1 years) and a control group of 150 normal cases (97 female, mean age: 32.8 ± 14.9 years), along with 120 non-FCD pathological cases (64 female, mean age: 38.4 ± 18.4 years). The dataset was divided into three subsets, each analyzed by the CNN. Subsequently, the CNN underwent a two-phase-training process, incorporating subset MRIs and expert-labeled FCD maps. This training employed both classical and continual learning techniques. The CNN’s performance was validated by comparing the baseline model with the trained models at two training levels. ResultsIn prospective validation, the best model trained using continual learning achieved a sensitivity of 90.0%, specificity of 70.0%, and accuracy of 72.0%, with an average of 0.41 false positive clusters detected per MRI. For FCD segmentation, an average Dice coefficient of 0.56 was attained. The model’s performance improved in each training phase while maintaining a high level of sensitivity. Continual learning outperformed classical learning in this regard. ConclusionsOur study presents a promising CNN for FCD detection and segmentation, exhibiting both high sensitivity and specificity. Furthermore, the model demonstrates continuous improvement with the inclusion of more clinical MRI data. We consider our CNN a valuable tool for automated, examiner-independent FCD detection in daily clinical practice, potentially addressing the underutilization of epilepsy surgery in drug-resistant focal epilepsy and thereby improving patient outcomes.

Out-of-focus artifact removal for Fresnel incoherent correlation holography by deep learning

Fresnel incoherent correlation holography (FINCH) is a promising three-dimensional (3D) imaging method due to its incoherent illumination and inline optical configuration. However, due to information crosstalk, especially out-of-focus artifacts, the axial resolution of FINCH 3D imaging is limited, and the relationship between the artifacts and the reconstructed image is complex and nonlinear, which is difficult to obtain by traditional optical measurement methods. Here, inspired by the superiority of deep learning in nonlinear mapping over traditional physical methods, we propose and demonstrate a deep learning based out-of-focus artifact removal method for FINCH reconstruction, in which a out-of-focus segmentation convolutional neural network (CNN) is designed for obtaining accurate nonlinear relationship between the artifacts and the reconstructed image by training dataset and optimizing parameters, thereby eliminating information crosstalk in the reconstructed image. Both optical experiment and performance analysis demonstrate the effectiveness and superiority of the designed CNN in realizing out-of-focus artifacts removal. Importantly, this deep learning-based out-of-focus artifacts removal method will provide a useful strategy for realizing high quality 3D imaging of FINCH.

SEMANTIC POINT CLOUD SEGMENTATION IN URBAN ENVIRONMENTS WITH 1D CONVOLUTIONAL NEURAL NETWORKS

Abstract. Convolutional Neural Networks (CNNs) have been widely recognized for their efficacy in image analysis tasks. This paper investigates the application of the 1D-CNN variant CNNs for the semantic segmentation of urban point clouds obtained through Mobile Laser Scanning. Ten well-known local geometric features of point clouds were used as input for the 1D CNN. Through an empirical analysis on the Santiago Urban Dataset, the 1D CNN was optimized in terms of numbers of convolution layers, neurons, pooling layers, dropout layers, dense layers, training epochs, and batch size. The performance of the proposed 1D CNN was compared with Support Vector Machine (SVM), Random Forest (RF), and PointNet++. Despite demonstrating a F1-score weighted at 70.3%, outperforming SVM but slightly lagging RF (71.6%) and significantly trailing PointNet++ (90.3%), the proposed 1D-CNN showcases a cost-effective potential for the segmentation of road and building classes. The relative computational requirements of the models were also discussed, highlighting the practical advantages and limitations of each approach.