Size Of Training Dataset Research Articles

Ontology-Based decision tree model for prediction of fatty liver diseases

Non-Alcohol Fatty liver disease is a common clinical complication. The paper aimed to develop a knowledge-based fatty liver detection system based on an ontology and detection rules extracted from a decision tree algorithm. Ontology is created to represent knowledge related to patients and fatty liver disease. By utilizing 43 SWRL rules and the Drool inference engine in ontology, we detected fatty liver patients. The training dataset size is 70% of clean data, including 580 electronic medical records of patients who suffer from liver diseases. After inferencing the rules, the number of patients who suffer from fatty liver disease in ontology is the same as the decision tree model. The paper validated the result generated by the ontology model through the results of the decision tree model.

Preliminary study of artificial intelligence-based fuel-rod pattern analysis of low-quality tomographic image of fuel assembly

Single-photon emission computed tomography is one of the reliable pin-by-pin verification techniques for spent-fuel assemblies. One of the challenges with this technique is to increase the total fuel assembly verification speed while maintaining high verification accuracy. The aim of the present study, therefore, was to develop an artificial intelligence (AI) algorithm-based tomographic image analysis technique for partial-defect verification of fuel assemblies. With the Monte Carlo (MC) simulation technique, a tomographic image dataset consisting of 511 fuel-rod patterns of a 3 × 3 fuel assembly was generated, and with these images, the VGG16, GoogLeNet, and ResNet models were trained. According to an evaluation of these models for different training dataset sizes, the ResNet model showed 100% pattern estimation accuracy. And, based on the different tomographic image qualities, all of the models showed almost 100% pattern estimation accuracy, even for low-quality images with unrecognizable fuel patterns. This study verified that an AI model can be effectively employed for accurate and fast partial-defect verification of fuel assemblies.

SinGAN-Seg: Synthetic training data generation for medical image segmentation.

Analyzing medical data to find abnormalities is a time-consuming and costly task, particularly for rare abnormalities, requiring tremendous efforts from medical experts. Therefore, artificial intelligence has become a popular tool for the automatic processing of medical data, acting as a supportive tool for doctors. However, the machine learning models used to build these tools are highly dependent on the data used to train them. Large amounts of data can be difficult to obtain in medicine due to privacy reasons, expensive and time-consuming annotations, and a general lack of data samples for infrequent lesions. In this study, we present a novel synthetic data generation pipeline, called SinGAN-Seg, to produce synthetic medical images with corresponding masks using a single training image. Our method is different from the traditional generative adversarial networks (GANs) because our model needs only a single image and the corresponding ground truth to train. We also show that the synthetic data generation pipeline can be used to produce alternative artificial segmentation datasets with corresponding ground truth masks when real datasets are not allowed to share. The pipeline is evaluated using qualitative and quantitative comparisons between real data and synthetic data to show that the style transfer technique used in our pipeline significantly improves the quality of the generated data and our method is better than other state-of-the-art GANs to prepare synthetic images when the size of training datasets are limited. By training UNet++ using both real data and the synthetic data generated from the SinGAN-Seg pipeline, we show that the models trained on synthetic data have very close performances to those trained on real data when both datasets have a considerable amount of training data. In contrast, we show that synthetic data generated from the SinGAN-Seg pipeline improves the performance of segmentation models when training datasets do not have a considerable amount of data. All experiments were performed using an open dataset and the code is publicly available on GitHub.

New Explainable Deep CNN Design for Classifying Breast Tumor Response Over Neoadjuvant Chemotherapy.

To reduce breast tumor size before surgery, Neoadjuvant Chemotherapy (NAC) is applied systematically to patients with local breast cancer. However, with the existing clinical protocols, it is not yet possible to have an early prediction of the effect of chemotherapy on a breast tumor. Predicting the response to chemotherapy could reduce toxicity and delay effective treatment. Computational analysis of Dynamic Contrast-Enhanced Magnetic Resonance Images (DCE-MRI) through Deep Convolution Neural Network (CNN) has proved a significant performance in classifying responders and no responder's patients. This study intends to present a new explainable Deep Learning (DL) model predicting the breast cancer response to chemotherapy based on multiple MRI inputs. In this study, a cohort of 42 breast cancer patients who underwent chemotherapy was used to train and validate the proposed DL model. This dataset was provided by the Jules Bordet institute of radiology in Brussels, Belgium. 14 external subjects were used to validate the DL model to classify responding or non-responding patients on temporal DCE-MRI volumes. The model performance was assessed by the Area Under the receiver operating characteristic Curve (AUC), accuracy, and features map visualization according to pathological complete response (Ground truth). The developed deep learning architecture was able to predict the responding breast tumors to chemotherapy treatment in the external validation dataset with an AUC of 0.93 using parallel learning MRI images acquired at different moments. The visual results showed that the most important extracted features from non-responding tumors are in the peripheral and external tumor regions. The model proposed in this study is more efficient compared to those proposed in the literature. Even with a limited training dataset size, the developed multi-input CNN model using DCE-MR images acquired before and following the first chemotherapy was able to predict responding and non-responding tumors with higher accuracy. Thanks to the visualization of the extracted characteristics by the DL model on the responding and non-responding tumors, the latter could be used henceforth in clinical analysis after its evaluation based on more extra data.

Generative Adversarial Network (GAN) for Automatic Reconstruction of the 3D Spine Structure by Using Simulated Bi-Planar X-ray Images.

In this study, we modified the previously proposed X2CT-GAN to build a 2Dto3D-GAN of the spine. This study also incorporated the radiologist’s perspective in the adjustment of input signals to prove the feasibility of the automatic production of three-dimensional (3D) structures of the spine from simulated bi-planar two-dimensional (2D) X-ray images. Data from 1012 computed tomography (CT) studies of 984 patients were retrospectively collected. We tested this model under different dataset sizes (333, 666, and 1012) with different bone signal conditions to observe the training performance. A 10-fold cross-validation and five metrics—Dice similarity coefficient (DSC) value, Jaccard similarity coefficient (JSC), overlap volume (OV), and structural similarity index (SSIM)—were applied for model evaluation. The optimal mean values for DSC, JSC, OV, SSIM_anteroposterior (AP), and SSIM_Lateral (Lat) were 0.8192, 0.6984, 0.8624, 0.9261, and 0.9242, respectively. There was a significant improvement in the training performance under empirically enhanced bone signal conditions and with increasing training dataset sizes. These results demonstrate the potential of the clinical implantation of GAN for automatic production of 3D spine images from 2D images. This prototype model can serve as a foundation in future studies applying transfer learning for the development of advanced medical diagnostic techniques.

Localization and classification of structural damage using deep learning single-channel signal-based measurement

Diligent damage identification is a core thrust of structural health monitoring (SHM). Vibration-based SHM has recently gained paramount importance. Substantial research devoted to Deep Learning (DL) algorithms yielded superior accuracy. This study proposes a novel DL-based damage detection approach to automatically extract features from raw acceleration sensor data. A new One-Dimensional Convolutional Neural Network named BuildingNet was designed to learn features and identify damage locations in real-time under different damage assessment scenarios. Parametric studies were conducted on different layer numbers, size of training datasets, and noise levels. Systematic studies for optimizing the network architecture and training data were performed. Mid-rise building case studies demonstrated the accuracy and efficiency of the proposed model. Time-domain monitoring data, both from multiple and single-channel measurements, were used for training and testing different architectures of BuildingNet. The proposed model achieved excellent damage localization and classification performance on both noise-free and noisy data sets.

Detection of corrosion on steel structures using an artificial neural network

Image based-corrosion detection has become a widespread practice for steel structures, but fine-tuning their model parameters is time-consuming. Alternatively, convolutional neural networks (CNNs) can also be trained fast and automatically, but they demand a huge training dataset. In this paper, a corrosion detection approach based on an artificial neural network (ANN) whose training dataset size is less than 0.1% of that of typical CNNs is introduced. The input layer of the proposed ANN consists of textural and color properties. In the present work, different color spaces and textural properties are examined for their impact on the robustness of the ANN. Results reveal that the best color channels can be achieved by combining CIE L*u*v* and YUV color spaces. Moreover, energy is selected as the best texture feature with respect to the ANN robustness. The proposed ANN outperforms an available image processing algorithm from the perspective of both speed and accuracy. In conclusion, this ANN can be used for actual applications after a fast and straightforward training step.

Decentralized Distributed Multi-institutional PET Image Segmentation Using a Federated Deep Learning Framework.

The generalizability and trustworthiness of deep learning (DL)-based algorithms depend on the size and heterogeneity of training datasets. However, because of patient privacy concerns and ethical and legal issues, sharing medical images between different centers is restricted. Our objective is to build a federated DL-based framework for PET image segmentation utilizing a multicentric dataset and to compare its performance with the centralized DL approach. PET images from 405 head and neck cancer patients from 9 different centers formed the basis of this study. All tumors were segmented manually. PET images converted to SUV maps were resampled to isotropic voxels (3 × 3 × 3 mm3) and then normalized. PET image subvolumes (12 × 12 × 12 cm3) consisting of whole tumors and background were analyzed. Data from each center were divided into train/validation (80% of patients) and test sets (20% of patients). The modified R2U-Net was used as core DL model. A parallel federated DL model was developed and compared with the centralized approach where the data sets are pooled to one server. Segmentation metrics, including Dice similarity and Jaccard coefficients, percent relative errors (RE%) of SUVpeak, SUVmean, SUVmedian, SUVmax, metabolic tumor volume, and total lesion glycolysis were computed and compared with manual delineations. The performance of the centralized versus federated DL methods was nearly identical for segmentation metrics: Dice (0.84 ± 0.06 vs 0.84 ± 0.05) and Jaccard (0.73 ± 0.08 vs 0.73 ± 0.07). For quantitative PET parameters, we obtained comparable RE% for SUVmean (6.43% ± 4.72% vs 6.61% ± 5.42%), metabolic tumor volume (12.2% ± 16.2% vs 12.1% ± 15.89%), and total lesion glycolysis (6.93% ± 9.6% vs 7.07% ± 9.85%) and negligible RE% for SUVmax and SUVpeak. No significant differences in performance (P > 0.05) between the 2 frameworks (centralized vs federated) were observed. The developed federated DL model achieved comparable quantitative performance with respect to the centralized DL model. Federated DL models could provide robust and generalizable segmentation, while addressing patient privacy and legal and ethical issues in clinical data sharing.

Sensitivity examination of YOLOv4 regarding test image distortion and training dataset attribute for apple flower bud classification

ABSTRACT Applications of convolutional neural network (CNN)-based object detectors in agriculture have been a popular research topic in recent years. However, complicated agricultural environments bring many difficulties for ground truth annotation as well as potential uncertainties for image data quality. Using YOLOv4 as a representation of state-of-the-art object detectors, this study quantified YOLOv4’s sensitivity against artificial image distortions including white noise, motion blur, hue shift, saturation change, and intensity change, and examined the importance of various training dataset attributes based on model classification accuracies, including dataset size, label quality, negative sample presence, image sequence, and image distortion levels. The YOLOv4 model trained and validated on the original datasets failed at 31.91% white noise, 22.05-pixel motion blur, 77.38° hue clockwise shift, 64.81° hue counterclockwise shift, 89.98% saturation decrease, 895.35% saturation increase, 79.80% intensity decrease, and 162.71% intensity increase with 30% mean average precisions (mAPs) for four apple flower bud growth stages. The performance of YOLOv4 decreased with both declining training dataset size and training image label quality. Negative samples and training image sequence did not make a substantial difference in model performance. Incorporating distorted images during training improved the classification accuracies of YOLOv4 models on noisy test datasets by 13 to 390%. In the context of apple flower bud growth-stage classification, except for motion blur, YOLOv4 is sufficiently robust for potential image distortions by white noise, hue shift, saturation change, and intensity change in real life. Training image label quality and training instance number are more important factors than training dataset size. Exposing models to test-image-alike training images is crucial for optimal model classification accuracies. The study enhances understanding of implementing object detectors in agricultural research.

Multi-label annotation of text reports from computed tomography of the chest, abdomen, and pelvis using deep learning

BackgroundThere is progress to be made in building artificially intelligent systems to detect abnormalities that are not only accurate but can handle the true breadth of findings that radiologists encounter in body (chest, abdomen, and pelvis) computed tomography (CT). Currently, the major bottleneck for developing multi-disease classifiers is a lack of manually annotated data. The purpose of this work was to develop high throughput multi-label annotators for body CT reports that can be applied across a variety of abnormalities, organs, and disease states thereby mitigating the need for human annotation.MethodsWe used a dictionary approach to develop rule-based algorithms (RBA) for extraction of disease labels from radiology text reports. We targeted three organ systems (lungs/pleura, liver/gallbladder, kidneys/ureters) with four diseases per system based on their prevalence in our dataset. To expand the algorithms beyond pre-defined keywords, attention-guided recurrent neural networks (RNN) were trained using the RBA-extracted labels to classify reports as being positive for one or more diseases or normal for each organ system. Alternative effects on disease classification performance were evaluated using random initialization or pre-trained embedding as well as different sizes of training datasets. The RBA was tested on a subset of 2158 manually labeled reports and performance was reported as accuracy and F-score. The RNN was tested against a test set of 48,758 reports labeled by RBA and performance was reported as area under the receiver operating characteristic curve (AUC), with 95% CIs calculated using the DeLong method.ResultsManual validation of the RBA confirmed 91–99% accuracy across the 15 different labels. Our models extracted disease labels from 261,229 radiology reports of 112,501 unique subjects. Pre-trained models outperformed random initialization across all diseases. As the training dataset size was reduced, performance was robust except for a few diseases with a relatively small number of cases. Pre-trained classification AUCs reached > 0.95 for all four disease outcomes and normality across all three organ systems.ConclusionsOur label-extracting pipeline was able to encompass a variety of cases and diseases in body CT reports by generalizing beyond strict rules with exceptional accuracy. The method described can be easily adapted to enable automated labeling of hospital-scale medical data sets for training image-based disease classifiers.

The Usage of ANN for Regression Analysis in Visible Light Positioning Systems.

In this paper, we study the design aspects of an indoor visible light positioning (VLP) system that uses an artificial neural network (ANN) for positioning estimation by considering a multipath channel. Previous results usually rely on the simplistic line of sight model with limited validity. The study considers the influence of noise as a performance indicator for the comparison between different design approaches. Three different ANN algorithms are considered, including Levenberg–Marquardt, Bayesian regularization, and scaled conjugate gradient algorithms, to minimize the positioning error () in the VLP system. The ANN design is optimized based on the number of neurons in the hidden layers, the number of training epochs, and the size of the training set. It is shown that, the ANN with Bayesian regularization outperforms the traditional received signal strength (RSS) technique using the non-linear least square estimation for all values of signal to noise ratio (SNR). Furthermore, in the inner region, which includes the area of the receiving plane within the transmitters, the positioning accuracy is improved by 43, 55, and 50% for the SNR of 10, 20, and 30 dB, respectively. In the outer region, which is the remaining area within the room, the positioning accuracy is improved by 57, 32, and 6% for the SNR of 10, 20, and 30 dB, respectively. Moreover, we also analyze the impact of different training dataset sizes in ANN, and we show that it is possible to achieve a minimum of 2 cm for 30 dB of SNR using a random selection scheme. Finally, it is observed that is low even for lower values of SNR, i.e., values are 2, 11, and 44 cm for the SNR of 30, 20, and 10 dB, respectively.

Augmenting data with generative adversarial networks: An overview

Performance of neural networks greatly depends on quality, size and balance of training dataset. In a real environment datasets are rarely balanced and training deep models over such data is one of the main challenges of deep learning. In order to reduce this problem, methods and techniques are borrowed from the traditional machine learning. Conversely, generative adversarial networks (GAN) were created and developed, a relatively new type of generative models that are based on game theory and consist of two neural networks, a generator and a discriminator. The generator’s task is to create a sample from the input noise that is based on training data distribution and the discriminator should detect those samples as fake. This process goes through a finite number of iterations until the generator successfully fools the discriminator. When this occurs, sample becomes a part of new (augmented) dataset. Even though the original GAN creates unlabeled samples, variants that soon appeared removed that limitation. Generating artificial data through these networks appears to be a meaningful solution to the imbalance problem since it turned out that artificial samples created by GAN are difficult to differentiate from the real ones. In this manner, new samples of minority class could be created and dataset imbalance ratio lowered.

Physics-inspired architecture for neural network modeling of forces and torques in particle-laden flows

We present a physics-inspired neural network (PINN) model for direct prediction of hydrodynamic forces and torques experienced by individual particles in stationary arrays of randomly distributed spheres. In line with our findings, it has recently been demonstrated that conventional fully connected neural networks (FCNN) are incapable of making accurate predictions of force variations in a stationary random array of spheres. The problem arises due to the large number of input variables (i.e., the locations of individual neighboring particles) leading to an overwhelmingly large number of training parameters in a fully connected architecture. Given the typically limited size of training datasets that can be generated by particle-resolved simulations, the NN becomes prone to missing true patterns in the data, ultimately leading to overfitting. Inspired by our observations in developing the microstructure-informed probability-driven point-particle (MPP) model in a previous work, we incorporate two main features in the architecture of the present PINN model: (1) superposition of pairwise hydrodynamic interactions between particles, and (2) sharing training parameters between NN blocks that model neighbor influences. These strategies helps to substantially reduce the number of free parameters and thereby control the model complexity without compromising accuracy. We demonstrate that direct force and torque prediction using NNs is indeed possible, provided that the model structure corresponds to the underlying physics of the problem. For a Reynolds number range of 2⩽Re⩽150 and solid volume fractions of 0.1⩽ϕ⩽0.4, the PINN’s predictions prove to be as accurate as those of other currently available microstructure-informed models.

Cloud removal from satellite imagery using multispectral edge-filtered conditional generative adversarial networks

ABSTRACT We propose a Generative Adversarial Network (GAN) based architecture for removing clouds from satellite imagery. Data used for training comprises of visible light RGB and near-infrared (NIR) band images. The novelty lies in the structure of the discriminator in the GAN architecture, which compares generated and target cloud-free RGB images concatenated with their edge-filtered versions. Experimental results show that our approach to removing clouds outperforms both visually and according to metrics, a benchmark solution that does not take edge filtering into account, and that improvements are robust when varying both training dataset size and NIR cloud penetrability.

Effect of Training Data Volume on Performance of Convolutional Neural Network Pneumothorax Classifiers.

Large datasets with high-quality labels required to train deep neural networks are challenging to obtain in the radiology domain. This work investigates the effect of training dataset size on the performance of deep learning classifiers, focusing on chest radiograph pneumothorax detection as a proxy visual task in the radiology domain. Two open-source datasets (ChestX-ray14 and CheXpert) comprising 291,454 images were merged and convolutional neural networks trained with stepwise increase in training datasetsizes. Model iterations at each dataset volume were evaluated on an external test set of 525 emergency department chest radiographs. Learning curve analysis was performed to fit the observed AUCs for all models generated. For all three network architectures tested, model AUCs and accuracy increased rapidly from 2 × 103 to 20 × 103 training samples, with more gradual increase until the maximum training dataset size of 291 × 103 images. AUCs for models trained with the maximum tested dataset size of 291 × 103 images were significantly higher than models trained with 20 × 103 images: ResNet-50: AUC20k = 0.86, AUC291k = 0.95, p < 0.001; DenseNet-121 AUC20k = 0.85, AUC291k = 0.93, p < 0.001; EfficientNet AUC20k = 0.92, AUC 291k = 0.98, p < 0.001. Our study established learning curves describing the relationship between dataset training size and model performance of deep learning convolutional neural networks applied to a typical radiology binary classification task. These curves suggest a point of diminishing performance returns for increasing training data volumes, which algorithm developers should consider given the high costs of obtaining and labelling radiology data.

Comparison of classic object-detection techniques for automated sewer defect detection

Abstract Sewer systems play a key role in cities to ensure public assets and safety. Timely detection of defects can effectively alleviate system deterioration. Conventional manual inspection is labor-intensive, error-prone and expensive. Object detection is a powerful deep learning technique that can complement and/or replace conventional inspection, especially in complex environments. This study compares two classic object-detection methods, namely faster region-based convolutional neural network (R-CNN) and You Only Look Once (YOLO), for the detection and localization of five types of sewer defects. Model performances are evaluated based on their detection accuracy and processing speed under parameterization impacts of dataset size and training parameters. Results show that faster R-CNN achieved higher prediction accuracy. Training dataset size and maximum number of epochs (MaxE) had dominant impacts on model performances of faster R-CNN and YOLO, respectively. The processing speed increased along with the increasing training data for faster R-CNN, but did not vary significantly for YOLO. The models' abilities to detect disjoint and residential wall were highest, whereas crack and tree root were more difficult to detect. The results help to better understand the strengths and weaknesses of the classic methods and provide a useful user guidance for practical applications in automated sewer defect detection.

Cross-Lingual Transfer Learning for Arabic Task-Oriented Dialogue Systems Using Multilingual Transformer Model mT5

Due to the promising performance of pre-trained language models for task-oriented dialogue systems (DS) in English, some efforts to provide multilingual models for task-oriented DS in low-resource languages have emerged. These efforts still face a long-standing challenge due to the lack of high-quality data for these languages, especially Arabic. To circumvent the cost and time-intensive data collection and annotation, cross-lingual transfer learning can be used when few training data are available in the low-resource target language. Therefore, this study aims to explore the effectiveness of cross-lingual transfer learning in building an end-to-end Arabic task-oriented DS using the mT5 transformer model. We use the Arabic task-oriented dialogue dataset (Arabic-TOD) in the training and testing of the model. We present the cross-lingual transfer learning deployed with three different approaches: mSeq2Seq, Cross-lingual Pre-training (CPT), and Mixed-Language Pre-training (MLT). We obtain good results for our model compared to the literature for Chinese language using the same settings. Furthermore, cross-lingual transfer learning deployed with the MLT approach outperform the other two approaches. Finally, we show that our results can be improved by increasing the training dataset size.

Topic Modeling and Sentiment Analysis of Online Education in the COVID-19 Era Using Social Networks Based Datasets

Sentiment Analysis (SA) is a technique to study people’s attitudes related to textual data generated from sources like Twitter. This study suggested a powerful and effective technique that can tackle the large contents and can specifically examine the attitudes, sentiments, and fake news of “E-learning”, which is considered a big challenge, as online textual data related to the education sector is considered of great importance. On the other hand, fake news and misinformation related to COVID-19 have confused parents, students, and teachers. An efficient detection approach should be used to gather more precise information in order to identify COVID-19 disinformation. Tweet records (people’s opinions) have gained significant attention worldwide for understanding the behaviors of people’s attitudes. SA of the COVID-19 education sector still does not provide a clear picture of the information available in these tweets, especially if this misinformation and fake news affect the field of E-learning. This study has proposed denoising AutoEncoder to eliminate noise in information, the attentional mechanism for a fusion of features as parts where a fusion of multi-level features and ELM-AE with LSTM is applied for the task of SA classification. Experiments show that our suggested approach obtains a higher F1-score value of 0.945, compared with different state-of-the-art approaches, with various sizes of testing and training datasets. Based on our knowledge, the proposed model can learn from unified features set to obtain good performance, better results than one that can be learned from the subset of features.

Dictionary-based and machine learning classification approaches: a comparison for tonality and frame detection on Twitter data

ABSTRACT Automated text analysis methods have made it possible to classify large corpora of text by measures such as frames and tonality, with a growing popularity in social, political and psychological science. These methods often demand a training dataset of sufficient size to generate accurate models that can be applied to unseen texts. In practice, however, there are no clear recommendations about how big the training samples should be. This issue becomes especially acute when dealing with texts skewed toward categories and when researchers cannot afford large samples of annotated texts. Leveraging on the case of support for democracy, we provide a guide to help researchers navigate decisions when producing measures of tonality and frames from a small sample of annotated social media posts. We find that supervised machine learning algorithms outperform dictionaries for tonality classification tasks. However, custom dictionaries are useful complements of these algorithms when identifying latent democracy dimensions in social media messages, especially as the method of elaborating these dictionaries is guided by word embedding techniques and human validation. Therefore, we provide easily implementable recommendations to increase estimation accuracy under non-optimal condition.

Accuracy assessment of semantic segmentation for automatic aesthetic control on composite components

This work focuses on the effect of operational and environmental parameters on the image pre-processing step for automatic aesthetic control of composite components by a vision system. The image pre-processing provides first an evaluation of the their quality and a selection of the best ones, then the semantic segmentation of the parts of interest in the images, for an integrated and effective procedure, to improve the accuracy of the following inspection phase. With the aim of defining image selection criteria, a set of photos of the components under analysis are corrupted according increasing levels of blur, contrast degradation and noise. Experimental correlation between image quality indicators and segmentation performance metrics has been found. The training dataset size and the class balance appeared as the most important parameters for semantic segmentation performances. The analysis, carried out on carbon fibre components of complex geometry and superficial appearance, seems promising for automatic inspection.