Human Annotators Research Articles

Applying Deep Learning to Establish a Total Hip Arthroplasty Radiography Registry: A Stepwise Approach.

Establishing imaging registries for large patient cohorts is challenging because manual labeling is tedious and relying solely on DICOM (digital imaging and communications in medicine) metadata can result in errors. We endeavored to establish an automated hip and pelvic radiography registry of total hip arthroplasty (THA) patients by utilizing deep-learning pipelines. The aims of the study were (1) to utilize these automated pipelines to identify all pelvic and hip radiographs with appropriate annotation of laterality and presence or absence of implants, and (2) to automatically measure acetabular component inclination and version for THA images. We retrospectively retrieved 846,988 hip and pelvic radiography DICOM files from 20,378 patients who underwent primary or revision THA performed at our institution from 2000 to 2020. Metadata for the files were screened followed by extraction of imaging data. Two deep-learning algorithms (an EfficientNetB3 classifier and a YOLOv5 object detector) were developed to automatically determine the radiographic appearance of all files. Additional deep-learning algorithms were utilized to automatically measure the acetabular angles on anteroposterior pelvic and lateral hip radiographs. Algorithm performance was compared with that of human annotators on a random test sample of 5,000 radiographs. Deep-learning algorithms enabled appropriate exclusion of 209,332 DICOM files (24.7%) as misclassified non-hip/pelvic radiographs or having corrupted pixel data. The final registry was automatically curated and annotated in <8 hours and included 168,551 anteroposterior pelvic, 176,890 anteroposterior hip, 174,637 lateral hip, and 117,578 oblique hip radiographs. The algorithms achieved 99.9% accuracy, 99.6% precision, 99.5% recall, and a 99.6% F1 score in determining the radiograph appearance. We developed a highly accurate series of deep-learning algorithms to rapidly curate and annotate THA patient radiographs. This efficient pipeline can be utilized by other institutions or registries to construct radiography databases for patient care, longitudinal surveillance, and large-scale research. The stepwise approach for establishing a radiography registry can further be utilized as a workflow guide for other anatomic areas. Diagnostic Level IV . See Instructions for Authors for a complete description of levels of evidence.

805 Segmentation of cutaneous chronic graft-versus-host disease by a deep learning neural network

Surface area involvement is a key measure for staging and tracking cutaneous chronic graft-versus-host disease (cGVHD) but is limited by availability of expert dermatologist visual evaluation. We developed a deep learning algorithm to segment (mark out) skin affected by cGVHD and tested its performance on unseen patients. We took 179 affected and 181 unaffected skin 3D photos from 36 cGVHD patients. As ground truth, a human annotator marked each region affected by cGVHD. A U-Net algorithm was trained and tested in a leave-one-patient-out validation experiment.

Region interaction and attribute embedding for zero-shot learning

Zero-shot learning (ZSL) aims to recognize unseen class images only by using seen class images for training. In ZSL tasks, unseen classes share local attribute semantics with seen classes. However, most ZSL methods directly learn an embedding from global feature space to semantic space, which may fail to discover local attribute semantics and cause the strong bias problem. In this paper, a novel method called region interaction and attribute embedding (RIAE) is proposed. RIAE consists of two modules: the region graph network (RGN) and the attribute feature embedding (AFE). RGN constructs a region graph where each region (image patch) is regarded as a graph node for region interaction. That is, through the graph-convolution operations of RGN, the node features can aggregate the information from their neighboring node features and update themselves. In order to learn the compatibility between the updated node features and local attribute semantics, AFE is designed to assign high attention weights to the node features which have large compatibilities with attribute semantic vectors. Relying on attribute semantic vectors rather than human annotations, RIAE can adaptively extract and update local attribute features, and further learn an embedding from attribute feature space to semantic space. Extensive experiments on the benchmark datasets show that RIAE gets state-of-the-art or competitive performances.

On the use of multi-objective evolutionary classifiers for breast cancer detection.

PurposeBreast cancer is one of the most common tumours in women, nevertheless, it is also one of the cancers that is most usually treated. As a result, early detection is critical, which can be accomplished by routine mammograms. This paper aims to describe, analyze, compare and evaluate three image descriptors involved in classifying breast cancer images from four databases.ApproachMulti–Objective Evolutionary Algorithms (MOEAs) prove themselves as being efficient methods for selection and classification problems. This paper aims to study combinations of well–known classification objectives in order to compare the results of their application in solving very specific learning problems. The experimental results undergo empirical analysis which is supported by a statistical approach. The results are illustrated on a collection of medical image databases, but with a focus on the MOEAs’ performance in terms of several well–known measures. The databases were chosen specifically to feature reliable human annotations, so as to measure the correlation between the gold standard classifications and the various MOEA classifications.ResultsWe have seen how different statistical tests rank one algorithm over the others in our set as being better. These findings are unsurprising, revealing that there is no single gold standard for comparing diverse techniques or evolutionary algorithms. Furthermore, building meta-classifiers and evaluating them using a single, favorable metric is both extremely unwise and unsatisfactory, as the impact is to skew the results.ConclusionsThe best method to address these flaws is to select the right set of objectives and criteria. Using accuracy-related objectives, for example, is directly linked to maximizing the number of true positives. If, on the other hand, accuracy is chosen as the generic metric, the primary classification goal is shifted to increasing the positively categorized data points.

Self-supervised learning methods and applications in medical imaging analysis: a survey.

The scarcity of high-quality annotated medical imaging datasets is a major problem that collides with machine learning applications in the field of medical imaging analysis and impedes its advancement. Self-supervised learning is a recent training paradigm that enables learning robust representations without the need for human annotation which can be considered an effective solution for the scarcity of annotated medical data. This article reviews the state-of-the-art research directions in self-supervised learning approaches for image data with a concentration on their applications in the field of medical imaging analysis. The article covers a set of the most recent self-supervised learning methods from the computer vision field as they are applicable to the medical imaging analysis and categorize them as predictive, generative, and contrastive approaches. Moreover, the article covers 40 of the most recent research papers in the field of self-supervised learning in medical imaging analysis aiming at shedding the light on the recent innovation in the field. Finally, the article concludes with possible future research directions in the field.

Mitigating adversarial evasion attacks by deep active learning for medical image classification

In the Internet of Medical Things (IoMT), collaboration among institutes can help complex medical and clinical analysis of disease. Deep neural networks (DNN) require training models on large, diverse patients to achieve expert clinician-level performance. Clinical studies do not contain diverse patient populations for analysis due to limited availability and scale. DNN models trained on limited datasets are thereby constraining their clinical performance upon deployment at a new hospital. Therefore, there is significant value in increasing the availability of diverse training data. This research proposes institutional data collaboration alongside an adversarial evasion method to keep the data secure. The model uses a federated learning approach to share model weights and gradients. The local model first studies the unlabeled samples classifying them as adversarial or normal. The method then uses a centroid-based clustering technique to cluster the sample images. After that, the model predicts the output of the selected images, and active learning methods are implemented to choose the sub-sample of the human annotation task. The expert within the domain takes the input and confidence score and validates the samples for the model’s training. The model re-trains on the new samples and sends the updated weights across the network for collaboration purposes. We use the InceptionV3 and VGG16 model under fabricated inputs for simulating Fast Gradient Signed Method (FGSM) attacks. The model was able to evade attacks and achieve a high accuracy rating of 95%.

Using ensembles and distillation to optimize the deployment of deep learning models for the classification of electronic cancer pathology reports.

ObjectiveWe aim to reduce overfitting and model overconfidence by distilling the knowledge of an ensemble of deep learning models into a single model for the classification of cancer pathology reports.Materials and MethodsWe consider the text classification problem that involves 5 individual tasks. The baseline model consists of a multitask convolutional neural network (MtCNN), and the implemented ensemble (teacher) consists of 1000 MtCNNs. We performed knowledge transfer by training a single model (student) with soft labels derived through the aggregation of ensemble predictions. We evaluate performance based on accuracy and abstention rates by using softmax thresholding.ResultsThe student model outperforms the baseline MtCNN in terms of abstention rates and accuracy, thereby allowing the model to be used with a larger volume of documents when deployed. The highest boost was observed for subsite and histology, for which the student model classified an additional 1.81% reports for subsite and 3.33% reports for histology.DiscussionEnsemble predictions provide a useful strategy for quantifying the uncertainty inherent in labeled data and thereby enable the construction of soft labels with estimated probabilities for multiple classes for a given document. Training models with the derived soft labels reduce model confidence in difficult-to-classify documents, thereby leading to a reduction in the number of highly confident wrong predictions.ConclusionsEnsemble model distillation is a simple tool to reduce model overconfidence in problems with extreme class imbalance and noisy datasets. These methods can facilitate the deployment of deep learning models in high-risk domains with low computational resources where minimizing inference time is required.

Correlation Analysis of Variables From the Atherosclerosis Risk in Communities Study.

The need to test chemicals in a timely and cost-effective manner has driven the development of new alternative methods (NAMs) that utilize in silico and in vitro approaches for toxicity prediction. There is a wealth of existing data from human studies that can aid in understanding the ability of NAMs to support chemical safety assessment. This study aims to streamline the integration of data from existing human cohorts by programmatically identifying related variables within each study. Study variables from the Atherosclerosis Risk in Communities (ARIC) study were clustered based on their correlation within the study. The quality of the clusters was evaluated via a combination of manual review and natural language processing (NLP). We identified 391 clusters including 3,285 variables. Manual review of the clusters containing more than one variable determined that human reviewers considered 95% of the clusters related to some degree. To evaluate potential bias in the human reviewers, clusters were also scored via NLP, which showed a high concordance with the human classification. Clusters were further consolidated into cluster groups using the Louvain community finding algorithm. Manual review of the cluster groups confirmed that clusters within a group were more related than clusters from different groups. Our data-driven approach can facilitate data harmonization and curation efforts by providing human annotators with groups of related variables reflecting the themes present in the data. Reviewing groups of related variables should increase efficiency of the human review, and the number of variables reviewed can be reduced by focusing curator attention on variable groups whose theme is relevant for the topic being studied.

EventPointer 3.0: flexible and accurate splicing analysis that includes studying the differential usage of protein-domains.

Alternative splicing (AS) plays a key role in cancer: all its hallmarks have been associated with different mechanisms of abnormal AS. The improvement of the human transcriptome annotation and the availability of fast and accurate software to estimate isoform concentrations has boosted the analysis of transcriptome profiling from RNA-seq. The statistical analysis of AS is a challenging problem not yet fully solved. We have included in EventPointer (EP), a Bioconductor package, a novel statistical method that can use the bootstrap of the pseudoaligners. We compared it with other state-of-the-art algorithms to analyze AS. Its performance is outstanding for shallow sequencing conditions. The statistical framework is very flexible since it is based on design and contrast matrices. EP now includes a convenient tool to find the primers to validate the discoveries using PCR. We also added a statistical module to study alteration in protein domain related to AS. Applying it to 9514 patients from TCGA and TARGET in 19 different tumor types resulted in two conclusions: i) aberrant alternative splicing alters the relative presence of Protein domains and, ii) the number of enriched domains is strongly correlated with the age of the patients.

Soldering Data Classification with a Deep Clustering Approach: Case Study of an Academic-Industrial Cooperation

Modern industries still commonly use traditional methods to visually inspect products, even though automation has many advantages over the skills of human labour. The automation of redundant tasks is one of the greatest successes of Artificial Intelligence (AI). It employs human annotation and finds possible relationships between features within a particular dataset. However, until recently, this has always been the responsibility of AI specialists with a specific type of knowledge that is not available to the industrial domain experts. We documented the joint research of AI and domain experts as a case study on processing a soldering-related industrial dataset. Our image classification approach relies on the latent space representations of neural networks already trained on other databases. We perform dimensionality reduction of the representations of the new data and cluster the outputs in the lower dimension. This method requires little to no knowledge of the underlying architecture of neural networks by the domain experts, meaning it is easily manageable by them, supporting generalization to other use cases that can be investigated in future work. We also suggest a misclassification detecting method. We were able to achieve near-perfect test accuracy with minimal annotation work.

Linguistic Abnormalities in Isolated Rapid Eye Movement Sleep Behavior Disorder.

Patients with synucleinopathies frequently display language abnormalities. However, whether patients with isolated rapid eye movement sleep behavior disorder (iRBD) have prodromal language impairment remains unknown. We examined whether the linguistic abnormalities in iRBD can serve as potential biomarkers for conversion to synucleinopathy, including the possible effect of mild cognitive impairment (MCI), speaking task, and automation of analysis procedure. We enrolled 139 Czech native participants, including 40 iRBD without MCI and 14 iRBD with MCI, compared with 40 PD without MCI, 15 PD with MCI, and 30 healthy control subjects. Spontaneous discourse and story-tale narrative were transcribed and linguistically annotated. A quantitative analysis was performed computing three linguistic features. Human annotations were compared with fully automated annotations. Compared with control subjects, patients with iRBD showed poorer content density, reflecting the reduction of content words and modifiers. Both PD and iRBD subgroups with MCI manifested less occurrence of unique words and a higher number of n-grams repetitions, indicating poorer lexical richness. The spontaneous discourse task demonstrated language impairment in iRBD without MCI with an area under the curve of 0.72, while the story-tale narrative task better reflected the presence of MCI, discriminating both PD and iRBD subgroups with MCI from control subjects with an area under the curve of up to 0.81. A strong correlation between manually and automatically computed results was achieved. Linguistic features might provide a reliable automated method for detecting cognitive decline caused by prodromal neurodegeneration in subjects with iRBD, providing critical outcomes for future therapeutic trials. © 2022 International Parkinson and Movement Disorder Society.

Automated Quantification of Human Osteoclasts Using Object Detection

A balanced skeletal remodeling process is paramount to staying healthy. The remodeling process can be studied by analyzing osteoclasts differentiated in vitro from mononuclear cells isolated from peripheral blood or from buffy coats. Osteoclasts are highly specialized, multinucleated cells that break down bone tissue. Identifying and correctly quantifying osteoclasts in culture are usually done by trained personnel using light microscopy, which is time-consuming and susceptible to operator biases. Using machine learning with 307 different well images from seven human PBMC donors containing a total of 94,974 marked osteoclasts, we present an efficient and reliable method to quantify human osteoclasts from microscopic images. An open-source, deep learning-based object detection framework called Darknet (YOLOv4) was used to train and test several models to analyze the applicability and generalizability of the proposed method. The trained model achieved a mean average precision of 85.26% with a correlation coefficient of 0.99 with human annotators on an independent test set and counted on average 2.1% more osteoclasts per culture than the humans. Additionally, the trained models agreed more than two independent human annotators, supporting a more reliable and less biased approach to quantifying osteoclasts while saving time and resources. We invite interested researchers to test their datasets on our models to further strengthen and validate the results.

Exploring Automated Classification Approaches to Advance the Assessment of Collaborative Problem Solving Skills.

Competency in skills associated with collaborative problem solving (CPS) is critical for many contexts, including school, the workplace, and the military. Innovative approaches for assessing individuals’ CPS competency are necessary, as traditional assessment types such as multiple -choice items are not well suited for such a process-oriented competency. In a move to computer-based environments to support CPS assessment, innovative computational approaches are also needed to understand individuals’ CPS behaviors. In the current study, we describe the use of a simulation-based task on electronics concepts as an environment for higher education students to display evidence of their CPS competency. We further describe computational linguistic methods for automatically characterizing students’ display of various CPS skills in the task. Comparisons between such an automated approach and an approach based on human annotation to characterize student CPS behaviors revealed above average agreement. These results give credence to the potential for automated approaches to help advance the assessment of CPS and to circumvent the time-intensive human annotation approaches that are typically used in these contexts.

Automated detection of gunshots in tropical forests using convolutional neural networks

Unsustainable hunting is one of the leading drivers of global biodiversity loss, yet very few direct measures exist due to the difficulty in monitoring this cryptic activity. Where guns are commonly used for hunting, such as in the tropical forests of the Americas and Africa, acoustic detection can potentially provide a solution to this monitoring challenge. The emergence of low cost autonomous recording units (ARUs) brings into reach the ability to monitor hunting pressure over wide spatial and temporal scales. However, ARUs produce immense amounts of data, and long term and large-scale monitoring is not possible without efficient automated sound classification techniques. We tested the effectiveness of a sequential two-stage detection pipeline for detecting gunshots from acoustic data collected in the tropical forests of Belize. The pipeline involved an on-board detection algorithm which was developed and tested in a prior study, followed by a spectrogram based convolutional neural network (CNN), which was developed in this manuscript. As gunshots are rare events, we focussed on developing a classification pipeline that maximises recall at the cost of increased false positives, with the aim of using the classifier to assist human annotation of files. We trained the CNN on annotated data collected across two study sites in Belize, comprising 597 gunshots and 28,195 background sounds. Predictions from the annotated validation dataset comprising 150 gunshots and 7044 background sounds collected from the same sites yielded a recall of 0.95 and precision of 0.85. The combined recall of the two-step pipeline was estimated at 0.80. We subsequently applied the CNN to an un-annotated dataset of over 160,000 files collected in a spatially distinct study site to test for generalisability and precision under a more realistic monitoring scenario. Our model was able to generalise to this dataset, and classified gunshots with 0.57 precision and estimated 80% recall, producing a substantially more manageable dataset for human verification. Using a classifier-guided listening approach such as ours can make wide scale monitoring of threats such as hunting a feasible option for conservation management.

Data as Expression

ABSTRACT In machine learning literature, the concept of expression is seldom addressed as a general term in relation to information construction and data. The term more often carries a narrower meaning and refers to human bodily, emotional, or artistic dimensions that are recorded to train a model. However, this paper discusses a view in which expression is understood as any human sensory realm that can bring incidents explicit through the act of “pressing out,” as the early etymology of expression indicates. Also, all the sensory data that are used to train a machine learning model and the data the trained model gives as an output are considered meaningful through their expressive mediality—a form that is actively produced and can therefore be subjected to critical phenomenological analysis. This paper contrasts nonverbal expressions with categorical or linguistic expressions and asks, “What do eye movements express when they are used in training a machine learning model? What kind of expression arises in linguistic models? What could be considered aesthetic data (and what would it express)?” For philosopher John Dewey, instinctive reactions in human behavior that for example exhibit mere discharge of an emotion should be separated from purposeful expressions. However, in machine learning, the two Deweyan positions (instinctive and intentional) collapse, as artificial expressions are purely simulations of learned logic. Therefore, the phenomenological question of this paper is traced back to the original stimulus and its human annotators. In this paper, philosopher Don Ihde’s experimental phenomenology explains how appearances can be attended to without subsuming them under any assumptions, whereas art theory—arising especially from philosopher Dieter Mersch’s thinking—provides an understanding of the mediality of expression. This paper introduces examples from machine learning that are not generally considered expressive but are used for regular tasks, such as object detection, and provides an alternative approach from aesthetics and artistic research that understands these modalities as expressive. If data are understood as expressive, it can be critically assessed how current machine learning models constitute knowledge.

A Probabilistic Approach to Surgical Tasks and Skill Metrics.

Identifying and quantifying the activities that compose surgery is essential for effective interventions, computer-aided analyses and the advancement of surgical data science. For example, recent studies have shown that objective metrics (referred to as objective performance indicators, OPIs) computed during key surgical tasks correlate with surgeon skill and clinical outcomes. Unambiguous identification of these surgical tasks can be particularly challenging for both human annotators and algorithms. Each surgical procedure has multiple approaches, each surgeon has their own level of skill, and the initiation and termination of surgical tasks can be subject to interpretation. As such, human annotators and machine learning models face the same basic problem, accurately identifying the boundaries of surgical tasks despite variable and unstructured information. For use in surgeon feedback, OPIs should also be robust to the variability and diversity in this data. To mitigate this difficulty, we propose a probabilistic approach to surgical task identification and calculation of OPIs. Rather than relying on tasks that are identified by hard temporal boundaries, we demonstrate an approach that relies on distributions of start and stop times, for a probabilistic interpretation of when the task was performed. We first use hypothetical data to outline how this approach is superior to other conventional approaches. Then we present similar analyses on surgical data. We find that when surgical tasks are identified by their individual probabilities, the resulting OPIs are less sensitive to noise in the identification of the start and stop times. These results suggest that this probabilistic approach holds promise for the future of surgical data science.

Garbage In, Flowers Out: Noisy Training Data Help Generative Models at Test Time

Despite important progress, conversational systems often generate dialogues that sound unnatural to humans. We conjecture that the reason lies in the different training and testing conditions: agents are trained in a controlled “lab” setting but tested in the “wild”. During training, they learn to utter a sentence given the ground-truth dialogue history generated by human annotators. On the other hand, during testing, the agents must interact with each other, and hence deal with noisy data. We propose to fill this gap between the training and testing environments by training the model with mixed batches containing both samples of human and machine-generated dialogues. We assess the validity of the proposed method on GuessWhat?!, a visual referential game. We show that our method improves the linguistic quality of the generated dialogues, and it leads to higher accuracy of the guessing task; simple perturbations of the ground-truth dialogue history that mimic machine-generated data do not account for a similar improvement. Finally, we run a human evaluation experiment on a sample of machine-machine dialogues to complement the quantitative analysis. This experiment shows that also human annotators successfully exploit dialogues generated by a model trained with mixed batches to solve the task. Hence, the mixed-batch training does not cause a language drift. Moreover, we find that the new training regime allows human annotators to be significantly more confident when selecting the target object, showing that the generated dialogues are informative.

Active deep image clustering

Deep clustering has attracted increasingly more attention in recent years. However, due to the absence of labels, deep clustering sometimes still provides unreliable clustering results. Although semi-supervised deep clustering can alleviate this problem to some extent by involving few human annotations, we observe that the performance of semi-supervised clustering highly depends on the selection of data for human labeling, but unfortunately, the supervised information selection is still a tough problem as traditional semi-supervised methods pay no attention to it. To tackle this problem, in this paper, we propose a novel deep active clustering method, which can actively select the key data for human labeling and apply the human annotations to improve the deep clustering. Different from conventional semi-supervised deep clustering methods which use fixed pre-given supervised information, we design a simple yet effective strategy to select the informative and uncertain data for querying annotation, which is beneficial to the clustering task. Furthermore, we integrate deep representation learning, clustering, and data selection into a unified framework, so that each task can be boosted by each other. Finally, we conduct extensive experiments on benchmark data sets by comparing it with some state-of-the-art deep clustering methods and semi-supervised clustering methods. The experimental results show that our active clustering methods can outperform both the unsupervised and semi-supervised clustering methods, demonstrating the effectiveness of the proposed method. The codes of this paper are released in https://doctor-nobody.github.io/codes/ADC_codes.zip.

Performance evaluation of the smartphone-based AI cough monitoring app - Hyfe Cough Tracker against solicited respiratory sounds

Background: Emerging technologies to remotely monitor patients’ cough show promise for various clinical applications. Currently available cough detection systems all represent a trade-off between convenience and performance. The accuracy of such technologies is highly contingent on the clinical settings in which they are intended to be used. Moreover, establishing gold standards to measure this accuracy is challenging. Objectives: We present the first performance evaluation study of the Hyfe Cough Tracker app, a passive cough monitoring smartphone application. We evaluate performance for cough detection using continuous audio recordings and cough counting by trained individuals as the gold standard. We propose standard procedures to use multi-observer cough sound annotation from continuous audio recordings as the gold standard for evaluating automated cough detection devices. Methods: This study was embedded in a larger digital acoustic surveillance study (clinicaltrial.gov NCT04762693). Forty-nine participants were included and instructed to produce a diverse series of solicited sounds in 10-minute sessions. Simultaneously, continuous audio recording was performed using a MP3 recorder and two smartphones running Hyfe Cough Tracker app monitored and identified cough events. All continuous audio recordings were independently labeled by three medically-trained researchers. Results: Hyfe Cough Tracker app showed sensitivity of 91% and specificity of 98% with a very high correlation between the cough rate measured by Hyfe and that of human annotators (Pearson correlation of 0.968). A standardized approach to establish an acoustic gold standard for identifying cough sounds with multiple observers is presented. Conclusion: This is the first performance evaluation of a new smartphone-based cough monitoring system. Hyfe Cough Tracker can detect, record and count coughs from solicited cough-like explosive sounds in controlled acoustic environments with very high accuracy. Additional steps are required to validate the system in clinical and community settings.

Unsupervised-learning-based method for chest MRI–CT transformation using structure constrained unsupervised generative attention networks

The integrated positron emission tomography/magnetic resonance imaging (PET/MRI) scanner simultaneously acquires metabolic information via PET and morphological information using MRI. However, attenuation correction, which is necessary for quantitative PET evaluation, is difficult as it requires the generation of attenuation-correction maps from MRI, which has no direct relationship with the gamma-ray attenuation information. MRI-based bone tissue segmentation is potentially available for attenuation correction in relatively rigid and fixed organs such as the head and pelvis regions. However, this is challenging for the chest region because of respiratory and cardiac motions in the chest, its anatomically complicated structure, and the thin bone cortex. We propose a new method using unsupervised generative attentional networks with adaptive layer-instance normalisation for image-to-image translation (U-GAT-IT), which specialised in unpaired image transformation based on attention maps for image transformation. We added the modality-independent neighbourhood descriptor (MIND) to the loss of U-GAT-IT to guarantee anatomical consistency in the image transformation between different domains. Our proposed method obtained a synthesised computed tomography of the chest. Experimental results showed that our method outperforms current approaches. The study findings suggest the possibility of synthesising clinically acceptable computed tomography images from chest MRI with minimal changes in anatomical structures without human annotation.